Blog

  • 「近」是什么意思:接地-穴位阻抗的经络特异性,与五把互相打架的空间尺子

    Evidence level: B

    「局部还是全身」不是数据自带的属性,它取决于你用什么量「近」。而在这个问题上,经络、体表厘米、探测体积、部位基线、事后归类这五把尺子同时在用,它们给出的答案并不一致。

    正文

    上一篇把「病人反应更大」拆成几层假象,结论是先别急着比两组幅度。这一篇往前走一步:假设放大确实存在,它长在哪儿?只出现在病灶所在经络的穴位上,还是全身穴位一起变?

    这问题听着只换了因变量,其实换掉了整个空间框架。「局部」与「全身」不是测量值,是判断——取决于你用什么尺子量「近」。而这里至少五把尺子同时在用。

    第一把:经络。 说「放大局限于胃经」,意思是效应沿一条传统线路分布。可经络不是等距坐标系:穴位间距在四肢与躯干差好几倍,「相关经」与「不相关经」在体表常彼此贴邻。更要紧的是经络分区没法还原成皮节——皮节图本身就不准,现有图谱多基于有缺陷的早期研究,重叠与个体变异被长期低估(PMID 18470936)。所以「胃经」既不等于 L3-L4 皮节,也不等于「膝盖周围十厘米」。把「相关经」当干净二分变量,等于假设三套解剖学恰好重合。

    第二把:体表直线距离。 用厘米建模衰减看着最客观,但它默认效应按半径扩散。经络假说恰恰不这么预测:它讲「上病下取」「内属于脏腑,外络于肢节」,远端取穴靠连通而非邻近。于是同一批数据在两把尺子下结论相反:按厘米算远端效应弱,支持「局部」;按经络算足三里对胃、合谷对面口,支持「跨段连通」。这不是数据矛盾,是尺子不同——沿路径的连通与随距离的衰减是两个模型,不能用一个去否证另一个。

    第三把:探测体积。 这一把最少被提及。四电极法测的不是一个点,是极板之间那块组织的等效阻抗;针间距、进针深度、皮下胶原带都改变读数(PMID 15882468;PMID 20689594;PMID 12467083 报告穴位与肌间结缔组织隔对应率约八成)。相邻穴位间距常与探测体积相当,「相邻测点」实为同一块组织的重复取样。这在统计上叫伪重复——估计不偏,但标准误被系统性低估,第一类错误率严重偏离名义水平(PMID 21687445:组内相关越高、组数越少,偏离越离谱)。后果具体:「全身广泛变化」可能只是探测体积重叠造成的假扩散,「局限」也可能只是同一个热点被数了三次。

    第四把:部位基线。 十个不同体表区域的基线阻抗本就不同,且随年龄系统变化(PMID 27333585)。远端穴位反应小,可能不是「效应传不到」,而是那里基线低、没有下降余地,或电极贴合差、灵敏度低。跨部位比原始阻抗值,等于把部位效应当成距离效应。做法是每个部位都用同部位非穴位点自身对照,只在部位内算差值再比。

    第五把:谁算「相关」。 一个膝痛患者,取胃经(梁丘、犊鼻)、胆经(阳陵泉)、脾经(阴陵泉)、膀胱经(委中)都说得通。若「相关经」是看完数据才定,任何结果都能找到一条自洽的经络解释。这是研究者自由度最经典的入口(PMID 27879038;核查标准见 PMID 33296358)。要拦它只有一条路:「相关经」的定义必须在看数据前写死,并写明依据(教科书条文、定位规则还是解剖标志)——允许事后改,就是允许事后编。

    五把尺子摆在一起,会露出一个更麻烦的结构问题:部位效应与距离效应高度共线。 若衰减曲线在膝、肩、腰三处形状一致,可以说是「按部位可测性调制」;若曲线随病灶整体平移、始终以病灶为峰,可以说是「局部」。两种解释都能拟合同一批数据,除非在多个病灶部位重复同一套几何——真空间结构会跟着病灶走,部位效应不会。

    还有一条不能回避的反向证据:临床上「局部取穴」与「远端取穴」的镇痛效果,在直接比较的研究中未见显著差异(PMID 25690699);催产引产的系统综述里,局部加远端联合与单用远端也无差异(PMID 40801473)。这不能证明穴位没有空间结构,但说明在中观层面「近的比远的强」未被临床数据支持,举证责任在「空间特异性」一侧。

    所以该做的不是「相关经 vs 不相关经」的交互项,而是三件更笨的事。一是把距离当连续变量,同时拟合三种距离(体表厘米、经络路径、皮节段差)比边际解释力——若三者接近,说明数据分辨不出空间结构,任何「局限于某经」的结论都不成立。二是显式写入空间结构,用带空间随机效应的多层模型或距离衰减核,不把相邻测点当独立样本;样本量按独立空间单元数算,不按电极数算。三是多部位重复同一几何,让「跟着病灶走」成为可检验的预测,而不是事后的说法。

    易学落点。《素问·皮部论》讲「皮者脉之部也」,把体表按经脉分成十二皮部——这是按归属分,不是按远近分。经络从设计上就是一张关系网:上下、内外、表里、原络,讲的是谁连着谁、谁管着谁,不是「离病灶多远」。所以拿厘米检验经络,用的是经络自己没用过的尺子;只用经络标签分组,又丢掉了最核心的信息——方向与连通。真正的检验该问的不是「半径多大」,而是「图上哪条边被点亮了」。

    证据与局限

    已验证: 穴位与肌间/肌内结缔组织隔存在约八成解剖对应(PMID 12467083),沿经络的皮下胶原带与更低阻抗相关(PMID 20689594;PMID 15882468,但仅心包经显著、脾经不显著);四电极法测得组织阻抗随针间距与深度变化(PMID 15882468;PMID 20689594);体表不同部位基线阻抗差异显著并随年龄系统变化(PMID 27333585);穴位皮肤电阻在健康人中约四成测次与周边不同、方向可高可低,且短期可重复性高、一周后很低(PMID 19422323);同一设备在同一批穴位上的单次测量信度约 0.72-0.76(PMID 15353016);温湿度、取穴标准、仪器操作是公认混杂(PMID 29354942);现有皮节图不准确、重叠与变异被低估(PMID 18470936);忽略层次结构会使第一类错误率严重偏离(PMID 21687445);小样本与研究者自由度共同推高假阳性(PMID 23571845;PMID 27879038;PMID 33296358)。

    部分验证: 穴位敏化的动物证据支持「病灶→特定皮节段神经源性炎症」的空间局域性:胃黏膜损伤大鼠的伊文思蓝渗出点集中于 T9-11 皮节,并伴 SP/CGRP 纤维与肥大细胞聚集(PMID 28270193;综述见 PMID 35620665);内脏痛的牵涉与继发痛觉过敏由躯体内脏汇聚神经元敏化解释,可扩散至多个皮节(PMID 24903037;PMID 12817663)。但这些是针刺样敏化点而非阻抗读数,两者是否共用同一空间尺度,尚无直接数据。

    待验证: 三种距离模型在同一批接地-阻抗数据上的相对解释力;探测体积重叠造成的空间自相关强度(是否足以单独伪造出「全身扩散」);「多部位重复同一几何」能否把部位效应与距离效应分开;局部与远端取穴在临床镇痛上等效(PMID 25690699;PMID 40801473)与电学层面可能存在空间特异性之间如何调和。

    相关阅读

    证据等级: B
    参考文献: PMID 18470936, PMID 15882468, PMID 20689594, PMID 12467083, PMID 21687445, PMID 27333585, PMID 27879038, PMID 33296358, PMID 25690699, PMID 40801473, PMID 19422323, PMID 15353016, PMID 29354942, PMID 23571845, PMID 28270193, PMID 35620665, PMID 24903037, PMID 12817663


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    The Meaning of \”Near\”: Meridian Specificity of Grounding-Acupuncture Point Impedance and Five Conflicting Spatial Scales

    Evidence level: B

    “Local versus systemic” is not an intrinsic attribute of the data; it depends on which scale you use to measure “nearness.” On this issue, five scales operate simultaneously: the meridian, surface linear distance, probe volume, anatomical baseline, and post-hoc classification. They yield inconsistent answers.

    Main Text

    The previous article deconstructed “greater patient response” into layers of illusion, concluding that one should not rush to compare magnitudes between groups. This article advances further: assuming amplification exists, where does it grow? Does it appear only on acupoints along the affected meridian, or do all acupoints change systemically?

    This question seems to merely swap the dependent variable, yet it replaces the entire spatial framework. “Local” and “systemic” are not measurements; they are judgments dependent on which ruler you use to measure “nearness.” Here, at least five rulers operate simultaneously.

    First Ruler: The Meridian. Stating that “amplification is limited to the Stomach Meridian” implies the effect distributes along a traditional pathway. However, meridians are not equidistant coordinate systems: inter-acupoint spacing differs by several times between limbs and the trunk. Furthermore, “related meridians” and “unrelated meridians” often abut each other on the surface. More critically, meridian zoning cannot be reduced to dermatomes—the dermatome maps themselves are inaccurate, and existing charts rely on flawed early studies where overlaps and individual variations have been systematically underestimated (PMID 18470936). Therefore, the “Stomach Meridian” is neither equivalent to the L3-L4 dermatome nor to “ten centimeters around the knee.” Treating the “related meridian” as a clean binary variable assumes three distinct anatomical systems coincide perfectly.

    Second Ruler: Surface Linear Distance. Modeling attenuation with centimeters appears most objective, yet it defaults to radial diffusion. The meridian hypothesis predicts otherwise: it posits “treat upper diseases with lower points” and “internal organs belong internally, limbs/joints externally,” relying on connectivity rather than proximity for distal point selection. Consequently, the same dataset yields opposite conclusions under two different rulers: calculating by centimeters shows weak distal effects, supporting “locality”; calculating by meridian shows Zusanli affecting the stomach and Hegu affecting the mouth, supporting “cross-segment connectivity.” This is not a data contradiction but a difference in rulers—pathway connectivity and distance-dependent attenuation are two distinct models; neither can refute the other using a single metric.

    Third Ruler: Probe Volume. This ruler is least frequently mentioned. The four-electrode method does not measure a single point but the equivalent impedance of the tissue between the plates; needle spacing, insertion depth, and subcutaneous collagen bands all alter readings (PMID 15882468; PMID 20689594; PMID 12467083 reports an ~80% correspondence rate between acupoints and intermuscular connective tissue). Adjacent acupoint spacing often matches the probe volume; thus, “adjacent measurement points” constitute repeated sampling of the same tissue block. Statistically, this is pseudoreplication—estimates remain unbiased, but standard errors are systematically underestimated, causing severe deviation from nominal Type I error rates (PMID 21687445: higher intra-class correlation and fewer groups lead to greater deviation). The consequence is specific: “systemic widespread changes” may merely be false diffusion caused by overlapping probe volumes, while “locality” may simply be the same hotspot counted three times.

    Fourth Ruler: Anatomical Baseline. Baseline impedance across ten different surface regions naturally differs and changes systematically with age (PMID 27333585). Small distal responses may not indicate “failure of effect transmission” but rather low baselines with no room for decrease, poor electrode contact, or low sensitivity. Comparing across regions treats regional effects as distance effects. The correct approach is to use same-region non-acupoint points for internal control within each region, calculating differences only within regions before comparison.

    Fifth Ruler: Who Defines “Relevance”. For a patient with knee pain, the Stomach Meridian (Liangqiu, Dubini), Gallbladder Meridian (Yanglingquan), Spleen Meridian (Yinlingquan), and Bladder Meridian (Weizhong) all provide plausible explanations. If the “related meridian” is determined after viewing the data, any result can find a self-consistent meridian explanation. This is the classic entry point for researcher degrees of freedom (PMID 27879038; verification criteria see PMID 33296358). To constrain this, there is only one path: the definition of the “related meridian” must be fixed prior to data viewing, with justification stated (textbook clauses, localization rules, or anatomical landmarks)—allowing post-hoc changes is equivalent to allowing post-hoc fabrication.

    Placing all five rulers together reveals a more complex structural issue: high collinearity between regional effects and distance effects. If attenuation curves have consistent shapes at the knee, shoulder, and waist, it can be termed “region-specific measurability modulation”; if curves shift globally with the lesion peak remaining fixed at the lesion site, it can be termed “local.” Both explanations can fit the same dataset unless the same geometric setup is repeated across multiple lesion sites—true spatial structure follows the lesion, whereas regional effects do not.

    There is also unavoidable counter-evidence: clinical studies directly comparing analgesic efficacy between “local point selection” and “distal point selection” show no significant difference (PMID 25690699); systematic reviews of labor induction show no difference between combined local and distal use versus distal use alone (PMID 40801473). This does not prove acupoints lack spatial structure, but it indicates that “near is stronger than far” at the meso-level is not supported by clinical data, placing the burden of proof on the “spatial specificity” side.

    Therefore, the task is not an interaction term of “related vs. unrelated meridians,” but three more laborious actions. First, treat distance as a continuous variable, simultaneously fitting three distances (surface centimeters, meridian path length, dermatome segment difference) to compare marginal explanatory power—if all three are similar, data cannot resolve spatial structure, and any conclusion of “limitation to a specific meridian” is invalid. Second, explicitly encode spatial structure using multi-level models with spatial random effects or distance-decay kernels, treating adjacent measurement points as non-independent samples; calculate sample size based on independent spatial units, not electrode counts. Third, repeat the same geometry across multiple regions, making “following the lesion” a testable prediction rather than a post-hoc explanation.

    Yijing Studies perspective. The Su Wen · Pi Bu Lun states “The skin is the department of the vessels,” dividing the surface into twelve dermatomes according to meridians—this is classification by affiliation, not classification by proximity. Meridians are designed as a relational network: upper/lower, interior/exterior, exterior/internal, original/branching, describing who connects to whom and who governs whom, not “how far from the lesion.” Thus, testing meridians with centimeters uses a ruler the meridian system itself does not employ; grouping solely by meridian labels discards the core information—direction and connectivity. The true test should not ask “what radius,” but “which edges on the map are illuminated.”

    Evidence and Limitations

    Verified: There is approximately 80% anatomical correspondence between acupoints and intermuscular/intramuscular connective tissue septa (PMID 12467083); subcutaneous collagen bands along meridians correlate with lower impedance (PMID 20689594; PMID 15882468, though significant only for the Pericardium Meridian and not the Spleen Meridian); four-electrode methods show tissue impedance varies with needle spacing and depth (PMID 15882468; PMID 20689594); significant baseline impedance differences exist across surface regions and change systematically with age (PMID 27333585); acupoint skin resistance in healthy individuals varies across ~40% of measurements relative to surroundings with variable direction and high short-term repeatability but very low repeatability after one week (PMID 19422323); inter-device reliability for single measurements on the same batch of acupoints is approximately 0.72-0.76 (PMID 15353016); humidity, temperature, point selection standards, and instrument operation are recognized confounders (PMID 29354942); existing dermatome maps are inaccurate with underestimated overlaps and variations (PMID 18470936); ignoring hierarchical structure causes severe deviation in Type I error rates (PMID 21687445); small sample sizes and researcher degrees of freedom jointly inflate false positives (PMID 23571845; PMID 27879038; PMID 33296358).

    Partially Verified: Animal evidence supporting acupoint sensitization supports “lesion-specific dermatome segment neurogenic inflammation” spatial locality: Evans blue extravasation points in rats with gastric mucosal injury concentrate at T9-11 dermatomes accompanied by SP/CGRP fibers and mast cell aggregation (PMID 28270193; review see PMID 35620665); referred pain and secondary hyperalgesia are explained by sensitization of trunk visceral convergence neurons, which can diffuse to multiple dermatomes (PMID 24903037; PMID 12817663). However, these are needle-like sensitization points, not impedance readings; whether they share the same spatial scale lacks direct data.

    To Be Verified: Relative explanatory power of the three distance models on the same batch of grounding-impedance data; spatial autocorrelation intensity caused by probe volume overlap (whether sufficient to independently fabricate “systemic diffusion”); whether “repeating the same geometry across multiple regions” can separate regional effects from distance effects; how to reconcile clinical equivalence of local vs. distal point selection for analgesia (PMID 25690699; PMID 40801473) with potential spatial specificity at the electrical level.

    Related Reading

    Evidence Level: B
    References: PMID 18470936, PMID 15882468, PMID 20689594, PMID 12467083, PMID 21687445, PMID 27333585, PMID 27879038, PMID 33296358, PMID 25690699, PMID 40801473, PMID 19422323, PMID 15353016, PMID 29354942, PMID 23571845, PMID 28270193, PMID 35620665, PMID 24903037, PMID 12817663


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 「病人反应更大」这句话,先要问是哪把尺子量的:接地-穴位阻抗的病理放大效应与四层假象

    Evidence level: B

    交互项显著只说明两组的变化量不相等,不说明病理状态放大了效应——在两组基线不同、入组规则不同、测量系统不同的情况下,「放大」可以完全由除法和招募方式造出来。

    正文

    上一篇确立了一个前提:接地会系统性地扰动穴位阻抗,而这种扰动是否具穴位特异性,要看「接地×测点类别」的交互项。本主题把交互项里的第二个因素从「穴位 vs 旁开点」换成「病人 vs 健康人」:如果慢性疼痛或失眠患者的接地-阻抗响应幅度大于健康对照,是否说明病理状态放大了这个界面效应?

    这看似只换了分组变量,其实换了一整套统计风险。因为「穴位 vs 旁开点」是在同一个人身上比,两组基线天然共享;而「病人 vs 健康人」是两个人群比,基线、入组规则、用药、测量条件全都不同。交互项在数学上仍然只是一个乘积项,但它此刻承载的因果含义要脆弱得多。至少四层假象,可各自独立造出同一个「放大」。

    第一层:放大需要一个共同的「1」。 假设健康组基线 100 kΩ、患者组 70 kΩ,接地后两组都下降 10 kΩ。绝对变化完全相等,但换成相对变化,患者组是 14.3%、健康组是 10%——凭空多出四成「放大」。所以「放大」在绝对 Ω、相对 %、标准化 d 三种尺度下不是同一句话。只报一种尺度的研究,读者分不清它报的是生理还是除法。

    第二层:患者组可能根本没有放大的空间。 穴位阻抗对组织水合高度敏感,电容法与阻抗法测的其实是同一件事的不同侧面(PMID 22092664);慢性疼痛与失眠人群常伴局部水肿、出汗改变、皮温异常——复杂区域疼痛综合征的皮温不对称本身就是诊断特征(PMID 35005383)。这些因素普遍把基线阻抗压低。基线已低,再往下的余地就小,真实机制更可能是「压缩」;而压缩与放大在交互项里长得一样,只是符号相反。

    第三层,也是最容易被忽略的:差异性回归到均值。 回归到均值不是误差,是选择的必然结果(PMID 15333621;一般表达式见 PMID 40126924)。当两组入组规则不同——健康组随便招、患者组按症状严重度招——两组的极端程度天然不等,重复测量时向均值回落的幅度也就不等;这种差异性回归足以单独造出一个显著的组间变化差(控制思路见 PMID 37284516)。须说清一点:入组按症状选、非按阻抗选,故回归严格讲作用于症状;但本假说的全部内容恰恰是「症状与阻抗相关」,于是症状的回归会顺着这条相关传导到阻抗上。换言之——这个假说越成立,这个假象就越强。

    第四层:患者组不是「健康人加一个病」,而是另一套测量系统。 纤维肌痛患者的自主神经功能本身偏离常态(PMID 27152281;昼夜节律分析见 PMID 9811051),失眠的高唤醒模型意味着交感张力基线持续抬高(PMID 19481481),而精神状态与用药都会实质改变皮肤电活动(PMID 2567035)。于是「病/健康」与基线交感张力、皮温、水合、用药史高度共线。交互项显著,只能说明两组在某个未拆分的复合变量上不同,不能说明是疾病机制。

    第五层:即使一切干净,检验本身也不给力。 交互项检验的功效通常只有同等主效应检验的四分之一量级(PMID 15066682);亚组分析的假阳性与假阴性风险早有量化(PMID 11701102),「总体无效而某亚组有效」在统计上完全可能(PMID 23939947)。更麻烦的是把严重度做成「病/不病」的二分——二分连续变量会系统性低估效应并损失功效(PMID 22457338)。十几人对十几人的研究报出「放大四成」,多半是几层假象的合力。

    所以真正该做的不是把两类人放一起比谁变得多,而是三件更朴素的事。一是用配对内差值,不用组间事后比:交叉设计里每个人自己的「接地减非接地」差值已经带走了他自己的基线,回归到均值在配对差值里被大幅抵消——这是交叉设计最被低估的价值,也最该被本主题利用。二是三种尺度同时报,并事先声明「放大」只在绝对、相对、标准化三向同向时成立;三向不一致时正确的表述是「两组变化在尺度上不可比」,不是「病人更敏感」。三是把严重度当连续变量做调节,用连续×处理的交互项代替两个人群的硬分组,并预注册方向与最小可检交互量;估不出交互量就先做剂量探索,别先做人群对比。

    还有一条反向的、不能回避的证据:在安慰剂镇痛这个最经典的「病人是否更敏感」问题上,慢性疼痛患者(骨关节炎与纤维肌痛)与健康人的反应幅度、可重复性并无差异(PMID 32639369)。这不能证明接地也如此,但它把举证责任放回了「放大」这一侧。

    易学落点。《素问·标本病传论》讲「知标本者,万举万当;不知标本,是谓妄行」——标本在这里就是「基线从哪来」。中医从不假设病人是健康人的放大版,它假设病人是另一种格局,所以才有同病异治、异病同治。风水里「福地福人居」其实也在说同一件事:同一个位置对不同的人,不是用同一把尺子量出来的。翻成统计语言就是:别拿两个基线不同的人群的原始变化量直接相减,然后把那个差叫作敏感性。

    证据与局限

    已验证: 接地即时、全身性、可逆地改变体表电位(PMID 22420736),并伴随血液流变学改变(PMID 22757749),效应综述见 PMID 22291721;接地缓解延迟性肌肉酸痛的初步证据(PMID 20192911);对基线未校正的疼痛试验做 ANCOVA 重分析会实质性改变结论与功效(PMID 15322438);回归到均值是选择效应而非误差,且在任意双变量分布下可量化(PMID 15333621;PMID 40126924);差异性回归可独立造成组间变化差,倾向评分可部分校正(PMID 37284516);交互检验功效远低于主效应检验(PMID 15066682);慢性疼痛患者的安慰剂反应幅度与健康人无差异(PMID 32639369)。

    部分验证: 纤维肌痛与失眠人群存在自主神经基线偏移(PMID 27152281;PMID 9811051;PMID 19481481),但这是否改变对外部电学干预的响应幅度,尚无直接数据;穴位阻抗与组织水合、皮温的耦合关系明确(PMID 22092664;PMID 35005383),但耦合系数在患病人群中的标定缺失。

    待验证: 接地×人群(健康/患病)交互项在三种尺度下是否同向;严重度作为连续调节变量的剂量-反应形状(该课题见后续 #467 方向);配对内差值能否把差异性回归压到可忽略的量级。

    相关阅读

    证据等级: B
    参考文献: PMID 22420736, PMID 22757749, PMID 20192911, PMID 15333621, PMID 40126924, PMID 37284516, PMID 15066682, PMID 11701102, PMID 23939947, PMID 22457338, PMID 22092664, PMID 35005383, PMID 27152281, PMID 9811051, PMID 19481481, PMID 2567035, PMID 32639369, PMID 22291721, PMID 15322438


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    \”Greater Patient Response\”: Measuring the Pathological Amplification Effect of Ground-Point Acupoint Impedance and the Four-Layer Illusion

    Evidence level: B

    A significant interaction term only indicates that the magnitude of change differs between two groups; it does not confirm that a pathological state amplifies the effect. Under conditions of differing baselines, varying inclusion criteria, and distinct measurement systems, “amplification” can be entirely generated by division ratios and recruitment methods.

    Main Text

    The previous article established a premise: grounding systematically perturbs acupoint impedance, and whether this perturbation exhibits acupoint specificity depends on the interaction term of “grounding × measurement site category.” This topic shifts the second factor in that interaction from “acupoint vs. adjacent point” to “patient vs. healthy control”: if the grounding-impedance response magnitude in chronic pain or insomnia patients exceeds that of healthy controls, does this indicate that the pathological state amplifies this interface effect?

    This appears to be merely a change in grouping variables, yet it introduces an entirely different statistical risk profile. Comparing “acupoints vs. adjacent points” involves individuals within the same person, sharing natural baselines; conversely, comparing “patients vs. healthy controls” involves distinct populations with differing baselines, inclusion criteria, medication histories, and measurement conditions. Mathematically, the interaction term remains merely a product term, yet its causal interpretation is now far more fragile. At least four illusions can each independently generate the same appearance of “amplification.”

    First Illusion: Amplification Requires a Common “1”. Assume the healthy group baseline is 100 kΩ and the patient group is 70 kΩ; after grounding, both groups decrease by 10 kΩ. The absolute change is identical, but when converted to relative change, the patient group shows a 14.3% reduction versus 10% for the healthy group—凭空 creating an extra 40% “amplification.” Thus, “amplification” is not a single statement across absolute Ω, relative %, and standardized d scales. Studies reporting only one scale leave readers unable to distinguish whether they are observing a physiological effect or a mathematical artifact of division.

    Second Illusion: The Patient Group May Have No Room for Amplification. Acupoint impedance is highly sensitive to tissue hydration levels; capacitive and impedance methods measure different facets of the same phenomenon (PMID 22092664). Chronic pain and insomnia populations frequently exhibit localized edema, altered sweating, and abnormal skin temperature—skin temperature asymmetry itself is a diagnostic feature for complex regional pain syndrome (PMID 35005383). These factors universally depress the baseline impedance. With a low baseline already established, the room for further reduction is small; the true mechanism is more likely “compression.” Compression and amplification appear identical in interaction terms, differing only in sign.

    Third Illusion (Most Easily Overlooked): Regression to the Mean. Regression to the mean is not error but an inevitable result of selection (PMID 15333621; general expression see PMID 40126924). When inclusion criteria differ between groups—random recruitment for healthy controls versus symptom severity-based recruitment for patients—the extremity of the two groups naturally differs. Consequently, the magnitude of return toward the mean during repeated measurements also differs; such differential regression alone can generate a significant between-group difference in change (control logic see PMID 37284516). One clarification is necessary: since selection is based on symptoms rather than impedance, regression strictly acts upon symptoms. However, the entirety of this hypothesis posits a correlation between “symptoms and impedance”; thus, symptom regression transmits along this correlation to impedance. In other words—the more valid this hypothesis is, the stronger this illusion becomes.

    Fourth Illusion: The Patient Group Is Not “Healthy Plus a Disease,” But a Different Measurement System. Autonomic function in fibromyalgia patients inherently deviates from norms (PMID 27152281; circadian rhythm analysis see PMID 9811051); the high-arousal model for insomnia implies a persistently elevated baseline sympathetic tone (PMID 194811481); and mental state alongside medication substantially alter skin conductance (PMID 2567035). Therefore, “patient/healthy” status is highly collinear with baseline sympathetic tone, skin temperature, hydration, and medication history. A significant interaction term merely indicates that the two groups differ on some undecomposed composite variable; it cannot confirm a disease mechanism.

    Fifth Illusion: Even If Everything Is Clean, the Test Itself Is Weak. The power of interaction tests is typically only one-quarter the magnitude of equivalent main effect tests (PMID 15066682); false positive and false negative risks in subgroup analysis have long been quantified (PMID 11701102), and “overall null with a specific subgroup effect” is statistically entirely possible (PMID 23939947). More problematic is converting severity into a binary “patient/non-patient” classification—binarizing continuous variables systematically underestimates effects and reduces power (PMID 22457338). Studies reporting “40% amplification” from dozens of patients versus dozens of controls are likely the合力 (synergy) of several layers of illusions.

    Therefore, the true task is not to compare two groups to see who changed more, but to pursue three simpler actions. First, use within-pair differences rather than between-group post-hoc comparisons: in crossover designs, each individual’s “grounding minus non-grounding” difference already removes their own baseline; regression to the mean is largely canceled out within paired differences—this is the most undervalued value of crossover design, and precisely what this topic should leverage. Second, report all three scales simultaneously, explicitly stating that “amplification” only holds when absolute, relative, and standardized changes align directionally; when they do not, the correct phrasing is “changes between groups are incomparable across scales,” not “patients are more sensitive.” Third, treat severity as a continuous moderator, using a continuous × treatment interaction term instead of hard grouping into two populations, and preregister the expected direction and minimum detectable interaction effect; if an interaction effect cannot be estimated, conduct dose exploration first rather than population comparison.

    There is also a reverse piece of evidence that cannot be avoided: in placebo analgesia, the most classic question of “whether patients are more sensitive,” chronic pain patients (osteoarthritis and fibromyalgia) show no difference in response magnitude or reproducibility compared to healthy controls (PMID 32639369). This does not prove the same holds for grounding, but it returns the burden of proof to the side of “amplification.”

    Yijing Studies Implication. The Su Wen · Ben Bian Lun states, “He who knows the root and branch acts correctly in all endeavors; he who does not is acting blindly.” Here, “root and branch” refers to “where the baseline originates.” Traditional Chinese medicine never assumes patients are an amplified version of healthy people; it assumes they represent a different pattern, hence the principles of treating the same disease differently or different diseases similarly. In Feng Shui, “a blessed land and a blessed resident” say the same thing: the same location is not measured with the same ruler for different people. Translated into statistical language: do not simply subtract raw change amounts from two populations with different baselines and label that difference as “sensitivity.”

    Evidence and Limitations

    Verified: Grounding immediately, systemically, and reversibly alters surface potential (PMID 22420736) and is accompanied by hemorheological changes (PMID 22757749); systematic review see PMID 22291721; preliminary evidence for grounding alleviating delayed-onset muscle soreness (PMID 20192911); reanalysis of pain trials without baseline correction via ANCOVA substantially alters conclusions and power (PMID 15322438); regression to the mean is a selection effect not error and is quantifiable under any bivariate distribution (PMID 15333621; PMID 40126924); differential regression can independently generate between-group change differences, propensity scores can partially correct (PMID 37284516); interaction test power is far lower than main effect tests (PMID 15066682); placebo response magnitude in chronic pain patients shows no difference from healthy controls (PMID 32639369).

    Partially Verified: Baseline autonomic shifts exist in fibromyalgia and insomnia populations (PMID 27152281; PMID 9811051; PMID 19481481), but whether this alters response magnitude to external electrical interventions lacks direct data; coupling between acupoint impedance and tissue hydration/skin temperature is clear (PMID 22092664; PMID 35005383), but calibration of coupling coefficients in diseased populations is missing.

    To Be Verified: Whether the grounding × population (healthy/diseased) interaction is consistent across all three scales; dose-response shape of severity as a continuous moderator (see subsequent #467 direction); whether within-pair differences can suppress differential regression to negligible levels.

    Related Reading

    Evidence Level: B
    References: PMID 22420736, PMID 22757749, PMID 20192911, PMID 15333621, PMID 40126924, PMID 37284516, PMID 15066682, PMID 11701102, PMID 23939947, PMID 22457338, PMID 22092664, PMID 35005383, PMID 27152281, PMID 9811051, PMID 19481481, PMID 2567035, PMID 32639369, PMID 22291721, PMID 15322438


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 池③ 引文错配全量定位:从”假号”到”真号伪书目”的判定升级

    Evidence level: C

    池③ · 课题 #190(本轮新登记)· 承接第 1/2/6 轮引文审计
    证据等级 C|状态 draft|本轮对既有正文写入 = 0


    0. 一句话结论

    跑了 721 篇、3162 个唯一 PMID,真正”不存在”的号只有 4 个——前几轮”假号很多”的印象是错的。真正的问题在另一头:PMID 真实,但配在它旁边的作者名、标题、期刊、年份全部是编的。我把它命名为「真号伪书目」(real-ID / fabricated-bibliography)。这一类,PubMed 存在性检查永远查不出。


    1. 方法与工具(本轮新建)

    新工具:08-ops/_tools/citation-pool-scan.py(纯标准库,py_compile 通过)

    设计原则来自第 37 轮的教训——只报不需要语义判断就能证伪的错,宁可漏报,绝不臆断:

    判据 是否确定性 是否进入结论
    E1 PMID 不存在 确定性(esummary + efetch 双确认) ✅
    E2 PMC 号当 PMID 确定性(elink 反查) ✅(本轮 0 命中)
    E3 年份矛盾 不确定(文本年份常是队列年/事件年) ⛔ 降级为 INFO,不进结论
    主题是否相关 不确定 ⛔ 不判,留人工
    作者是否对 不确定 ⛔ 不判,留人工

    1.1 工具自身踩的三个坑(全部实测修掉,写在这里防复发)

    # 坑 后果 修法 验证
    1 PMID \*\*40892475\*\* 这种 markdown 加粗写法匹配不到 静默漏提,一个文件 5 个号被读成 0 个 标记与数字间允许 ≤6 个非数字字符 修正后同文件读到 5 个号
    2 pmid: ["36106418", ...] 带引号、单数、frontmatter 内 漏提 frontmatter 层 按 [, \s"'] 切分 + 加单值/短横线写法 未提取行 74 → 13
    3 PMID 正则未加 re.I,正文小写 pmid: 全漏 大规模静默漏提 全部加 re.I 唯一 PMID 3153(此前 3152,差 1 但覆盖率从 74 缺到 13 缺)

    1.2 E3 为什么必须降级(实测数据)

    首版用”邻域 ±1 行所有年份”判年份矛盾,报出 866 条。人工抽 12 条,≥11 条是假阳性:

    声明年 真实年 实际含义
    2015 2000 文本写的是「720BC–AD2015 的日食记录」——数据跨度,不是发表年
    1960 2007 文本写的是「1960 年代电气化后」——事件年代
    1996 2013 文本写的是「1940、1977、1996 三个出生队列」——队列年份

    收紧到”同一行 PMID 前 45 字符内、剔除区间写法”后降到 125 条,抽样 12 条仍 ≥11 条假阳性(同一批问题换了个位置)。→ 结论:这类判据在本语料上不可用,降级 INFO。

    这条教训值得单独记:池③正文大量使用「某年某研究」的叙事句式,年份与 PMID 的邻近关系不构成发表年声明。任何基于邻近性的年份校验,在这个语料上都会大面积误报。


    2. 全池扫描结果

    指标 数值
    扫描文件 721 篇(03-sites/cn/research/science-yixue/)
    声明 PMID 唯一数 3162
    覆盖率缺口(含 PMID 字样但未提取) 13 行(均为审计文内的举例文本,非引用)
    E1 不存在 15 处 / 4 个唯一号 / 6 个文件
    E2 PMC 当 PMID 0
    E3 年份矛盾 0(已降级,见 §1.2)

    2.1 E1 全部命中(4 个唯一号,共 15 处)

    PMID 出现文件 声明的用途
    15199541 standing-wave-antinode-psychophysiology.md(3 处) 「HPA 2009《Infrasound: a brief overview of sources and effects》」
    34479992 dosha-qi-rsa-falsification-framework-methodology.md(2)、dosha-qi-rsa-falsification-preregistration-specification.md(2)、pool5-standby-validation-entry-exit-gate.md(1) 「Chung 2021 中医证型 HRV」/「Putnick & Bornstein 2016 测量等值性系统综述」/「HRV 跨族裔参考值等值性」——同一个号在三处被赋予三种不同身份
    28726973 pulse-signature-detraining-regression-dynamics.md(2) 参考文献列表内
    16030677 homeopathy-high-dilution-frequency-memory.md(1) 疑为 Rey 热释光研究

    (另有 rtribe-omop-citation-integrity-audit.md 4 处,属审计文自述”这些号不存在”,不是新错。)

    34479992 这个号的三种身份最值得注意:同一个不存在的号,被三篇文章分别说成”Chung 2021 HRV””Putnick & Bornstein 2016 测量等值性””HRV 跨族裔等值性”。这与第 2 轮判定的「跨子线挪用」失效模式完全一致——只不过这次被挪用的号本身就是假的。


    3. 真正的主问题:真号伪书目(本轮新命名)

    3.1 什么是”真号伪书目”

    pulse-pressure-normalization-hri-algorithm.md 第 105–106 行:

    3. Zhang Y, et al. "Real-time pressure feedback in pulse acquisition: impact on
       signal stability." Biomed Signal Process. 2021; PMID: 31730401
    4. Chen X, et al. "Pressure-harmonic gain curves for six primary pulse types."
       Chin J Integr Med. 2022; PMID: 33901427
    

    两个 PMID 都真实存在。但:

    PMID 真实文献 文档声称
    31730401 Schnyer RN 等,J Altern Complement Med 2019,”Can Reliability of the Chinese Medicine Diagnostic Process Be Improved? Results of a Prospective RCT” 「Zhang Y, Biomed Signal Process 2021,实时压力反馈」
    33901427 Weaver NA 等,Lancet Neurol 2021,”Strategic infarct locations for post-stroke cognitive impairment”(卒中) 「Chen X, Chin J Integr Med 2022,六种脉象压力-谐波增益曲线」

    → 作者、标题、期刊、年份、主题,五项全错;只有那串数字是真的。

    3.2 为什么这条最危险

    层面 后果
    存在性检查 查不出——PMID 是真的,E1 全绿
    引用数统计 不受影响——看起来引了 5 篇,实际可用 0 篇
    读者抽查 点进 PubMed 会看到一篇真实论文,反而增强可信感
    后果 「弦脉 G(P) 斜率是浮脉 1.7 倍」这种具体数值论断,挂在一篇完全无关的卒中影像论文上

    第 1 轮把这类判为「张冠李戴(B 类)」,当时判断是”不需重新找文献,只需归位“。本轮修正:这个判断只在”号本身有正确出处”时成立。 本例中,被挪用的号在池内没有任何一处正确引用(33901427 例外,见 §3.3),所以”归位”无门——必须重新找文献。

    3.3 同一个号,一处对一处错(第 2 轮判据仍成立)

    33901427 在 tcm-pulse-hri-inter-operator-reliability-training-standardization.md 里被正确使用:

    「虽然 Weaver 等人(2021)关于卒中认知损伤的研究并非脉诊研究,其核心方法论价值在于提供了多中心、大样本条件下评估操作者间信度的标准化范式」

    → 这是完全合规的引用:作者对、年份对、主题说清了、并明确声明了”不是脉诊研究”的边界。

    同一个号在 pulse-pressure-normalization-hri-algorithm.md 里被伪装成”陈 2022 六脉增益曲线”。

    → 结论:判断错配必须看”这个号在这篇里被用来支撑什么”,全池扫描 + 上下文比对才能定性。第 2 轮的方法论正确,本轮只是把它的适用面从 OMOP 扩到全池。

    3.4 伪书目的三个结构性特征(供后续自动检测用)

    特征 表现 可否自动化
    作者姓名为常见中文拼音(Zhang/Chen/Li/Wang + 单字母) 池③伪条目高发 弱信号,单用误报高
    期刊名与主题匹配但非该领域主刊 「Biomed Signal Process」「Chin J Integr Med」配脉诊 需领域词表
    同一条目在不同文件里被赋予不同标题 34479992 三身份 ✅ 可自动(同一 PMID 跨文件标题不一致 → 强信号)

    → 唯一可低成本自动化的是第三条。已建议加入扫描器(见 §6)。


    4. 本轮额外确认:谐波主线”域外文献”41 个,但多数是合法的

    对 53 篇谐波/HRI 文件做”标题不含脉搏/心血管/中医词”的粗筛,得 41 个域外文献。逐一核对后,绝大多数是合规引用:

    类型 例 判定
    方法学借用(明确说明) 26653405(中介分析指南)、27330520(ICC 指南)、24240655(NRI 评价) ✅ 合规
    运动生理学(去训练课题) 10999420、2692122、7219129、8440543、33629972 ✅ 合规(主题就是去训练)
    联邦学习(隐私课题) 37044052、38340728、39766014、39081567 ✅ 合规
    完全无关 34166332(HIV 检测趋势)、33081257(铸铁柱 CFRP 加固)、34637011(LGBTQ 青少年干预)、31787578(口香糖黏附性)、35262535(牦牛奶乳酸菌)、35343188(肺癌 EGFR 液体活检)、39157688(癫痫手术)、32451380(风速预测)、35127882(马牙修复) ⛔ 错配

    → 错配集中在 pulse-harmonic-gut-microbiome-organ-axis.md 与 pulse-harmonic-spectrum-individualized-zangfu-functional-scoring.md 两个文件,其余为合规引用。这说明错配不是全池弥漫,而是文件级爆发。


    5. 处置建议(分级,全部待 JOE 决策,本轮不动手)

    级 对象 动作 成本 是否需 JOE
    P0 2 个”肠-心轴”文件(gut-microbiome-organ-axis、spectrum-individualized-zangfu-scoring) 降级 frontmatter evidence_level 至 C,并在正文加”引文待重建”横幅(不改论断,只加免责) 低 ✅ 需批准(改既有正文)
    P1 4 个不存在号(E1) 删除或替换为真实文献;34479992 三处必须分别处理(三身份) 中 ✅ 需批准
    P2 pulse-pressure-normalization-hri-algorithm.md 第 105–106 条伪书目 重新检索真实文献(实时压力反馈 / 六脉压力-谐波增益,确有其题但号错) 中 ✅ 需批准
    P3 OMOP 9 篇(第 1/2 轮遗留) 原 A/B 方案,范围须含健康主线(第 6 轮已提) 高 ✅ 已挂 6 轮未决
    P4 全池 3162 个 PMID 的语义比对 工具化 + 分批人工 极高 建议只做”跨文件同号异题”自动检测

    5.1 本轮最想请 JOE 拍的一句话

    是否接受”先用横幅标注 + 降级 evidence_level,暂不逐条重检文献”这个过渡方案?

    理由:全量重检 3162 个 PMID 的语义,成本远超本池季度产能;而”横幅 + 降级”是可逆、低成本、且能立刻止住”读者被误导”这个最大风险。文献重建可以按 P1→P2→P3 分批做。


    6. 工具侧待办(下轮可自主)

    1. 加”跨文件同号异题”检测(§3.4 第三条,唯一低成本可自动化的伪书目信号)——输入两个文件、同一 PMID、两处声明的标题首词不同 → 报警。
    2. citation-verify.sh 升级:现有版本只回查”号是否真实 + 真实文献是什么”,不比对文中的声明。建议增加 --claim 参数,让写作者粘贴自己的书目行,脚本做词元重叠度打分(阈值 <0.2 即报”疑伪书目”)。
    3. 覆盖率闸门:任何扫描跑完,”未提取行”必须为 0 才允许出结论(本轮首版漏提 74 行,差点得出错误结论)。

    7. 证据边界(必读)

    • 本轮只对谐波主线 53 篇做了全量域外文献筛查 + 17 篇人工逐条比对;其余 668 篇只过了 E1/E2 确定性判据。
    • 因此「错配总数」是下界,不是全量。真实的真号伪书目数量可能远高于本文列出的 9 个文件。
    • 本文所有”真实文献”信息来自 NCBI E-utilities esummary/efetch,未做全文精读。
    • 本文未修改任何既有正文(含错误的 9 个文件一字未动)。
    • 本文不构成任何健康/医学结论。

    7.1 方法学锚点(3 个 PMID,E-utilities 逐条回查 title/首作者/期刊一致)

    本文所用审计方法并非自创,医学文献学界对”引用不准确”已有系统性量化研究:

    # PMID 首作者 载体 年 在本文中的角色
    M1 40696441 Baethge C Res Integr Peer Rev 2025 引用不准确性的系统综述与 meta 分析——本文的”错配是系统性而非偶发”这一判断的上游方法学依据
    M2 28910404 Mogull SA PLoS One 2017 医学论文中”被引事实”准确性的方法学与报告规范综述——支撑本文”必须比对声明而非只查存在性”的方法论选择
    M3 32977312 Montenegro TS J Neurosurg Spine 2021 单个专科的参考文献准确性实测——说明这类问题在各领域普遍存在,非本池特有

    三点共同的支撑作用:把本池的引文问题从「agent 出错的个案」重新定位为「文献学上已知的系统性现象」,从而支持 §3.2 的核心判断——存在性检查对这类失效无效,必须比对声明内容。


    免责声明

    本文为引文完整性审计,不涉及任何医学结论,也不构成对既有文章观点的支持或否定。所列错配仅为书目层面的定位,不代表原文的学术论断必然错误——但在文献重建完成前,相关文章的论断不得被引用为证据。


    池③ 课题 #190|勘探员:WB-FIVEPOOLS|2026-10-18|证据等级:C|状态:draft(未标 accepted,未写 779/782)


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    证据与局限

    本文证据等级为 C,属 measurable,结论限于研究层面的推断,不构成对个体情形的判定。不是假号(只有 4 个),而是”PMID 真实 + 书目信息整体虚构”这一类此前没被命名的失效模式。

    局限:本文未覆盖全部相关文献,效应量与结论随样本与方法变化;传统体系与现代研究的对照属结构类比而非等价证明,请结合专业判断参考。

    Pool 3: Full Localization of Citation Mismatch – Upgrade in Determination from \”False ID\” to \”True ID with Fake Bibliography\

    Evidence level: C

    Pool ③ · Project #190 (Newly Registered in This Round) · Building upon Citation Audits from Rounds 1/2/6
    Evidence Level C | Status draft | Writings to existing text in this round = 0


    0. One-Sentence Conclusion

    Across 721 articles and 3,162 unique PMIDs, only four IDs are genuinely non-existent; the impression from earlier rounds that “many fake IDs exist” is incorrect. The real issue lies elsewhere: the PMIDs are authentic, but the author names, titles, journals, and years listed alongside them are fabricated. I term this phenomenon “Real-ID / Fabricated-Bibliography.” For this category, PubMed’s existence verification will never flag a discrepancy.

    1. Methods and Tools (New in this Round)

    New tool: 08-ops/_tools/citation-pool-scan.py (pure standard library, passes py_compile).

    Design principles are derived from the lessons of Round 37: only report errors that can be falsified without semantic judgment, preferring false negatives over false positives:

    Criterion Certainty Enters Conclusion
    E1 PMID missing Certain (confirmed via esummary + efetch) ✅
    E2 PMC ID treated as PMID Certain (verified via elink reverse lookup) ✅ (0 hits in this round)
    E3 Year contradiction Uncertain (text years often refer to cohort/event years) ⛔ Downgrade to INFO; exclude from conclusion
    Topic relevance Uncertain ⛔ No judgment; leave for human review
    Author validity Uncertain ⛔ No judgment; leave for human review

    1.1 Three Pitfalls Encountered by the Tool Itself (All Fixed and Logged Here to Prevent Recurrence)

    # Pitfall Consequence Fix Verification
    1 Markdown bolding syntax like PMID \*\*40892475\*\* fails to match Silent omission: reads 5 IDs in a file as 0 Allow ≤6 non-digit characters between marker and number File now reads all 5 IDs after fix
    2 pmid: ["36106418", ...] with quotes, singular form, or in frontmatter Misses frontmatter layer Split by [,\s"'] + support single-value/dash notation Line 74 now extracted (was missing)
    3 PMID regex lacks re.I, causing all lowercase pmid: in text to be missed Large-scale silent omission Added re.I globally Unique PMID 3153 now captured (previously 3152; coverage improved from 74 missing to 13 missing)

    1.2 Why E3 Must Be Downgraded (Empirical Data)

    The initial version used “neighborhood ±1 lines” to judge year contradictions, reporting 866 cases. Manual sampling of 12 cases showed ≥11 were false positives:

    Stated Year Actual Year Actual Meaning
    2015 2000 Text refers to “solar eclipse records from 720 BC–AD 2015” — data span, not publication year
    1960 2007 Text refers to “after electrification in the 1960s” — event era
    1996 2013 Text refers to “three birth cohorts: 1940, 1977, 1996” — cohort year

    Restricting to “within the first 45 characters preceding each PMID on the same line, excluding interval notation” reduced cases to 125. Sampling 12 again showed ≥11 false positives (same issues, just repositioned). → Conclusion: This criterion is unreliable for this corpus; downgrade to INFO.

    This lesson warrants separate documentation: The pool corpus heavily employs narrative phrasing like “Year X Study.” The proximity between year and PMID does not constitute a declaration of publication year. Any year validation based on proximity will yield widespread false positives in this corpus.

    2. Full Pool Scan Results

    Indicator Value
    Scanned Files 721 (03-sites/cn/research/science-yixue/)
    Unique Declared PMID Count 3,162
    Coverage Gap (PMID present but not extracted) 13 lines (all illustrative text within audit documents, not citations)
    E1 Non-Existent 15 instances / 4 unique IDs / 6 files
    E2 PMC Match for PMID 0
    E3 Year Contradiction 0 (downgraded, see §1.2)

    2.1 E1 Full Hits (4 unique IDs, 15 instances total)

    PMID Affected Files Stated Purpose
    15199541 standing-wave-antinode-psychophysiology.md (3 instances) «HPA 2009 Infrasound: a brief overview of sources and effects»
    34479992 dosha-qi-rsa-falsification-framework-methodology.md (2), dosha-qi-rsa-falsification-preregistration-specification.md (2), pool5-standby-validation-entry-exit-gate.md (1) «Chung 2021 TCM Syndrome HRV» / «Putnick & Bornstein 2016 Measurement Equivalence Systematic Review» / «HRV Cross-Ethnic Reference Value Equivalence» — the same ID assigned three distinct identities across three locations
    28726973 pulse-signature-detraining-regression-dynamics.md (2) Listed in reference section
    16030677 homeopathy-high-dilution-frequency-memory.md (1) Suspected Rey luminescence study

    (Additionally, rtribe-omop-citation-integrity-audit.md contains 4 instances; these belong to the audit document’s own statement that “these IDs do not exist,” not new errors.)

    The three identities assigned to 34479992 are particularly notable: A non-existent ID is described by three articles as “Chung 2021 HRV,” “Putnick & Bornstein 2016 Measurement Equivalence,” and “HRV Cross-Ethnic Equivalence.” This aligns perfectly with the failure mode identified in Round 2 as «cross-subline appropriation» — except here, the appropriated ID itself is fabricated.


    3. The Core Question: Real IDs, Fake Bibliographies (New Naming for This Round)

    3.1 What is a “Real ID, Fake Bibliography”

    Lines 105–106 in pulse-pressure-normalization-hri-algorithm.md:

    3. Zhang Y, et al. "Real-time pressure feedback in pulse acquisition: impact on
       signal stability." Biomed Signal Process. 2021; PMID: 31730401
    4. Chen X, et al. "Pressure-harmonic gain curves for six primary pulse types."
       Chin J Integr Med. 2022; PMID: 33901427
    

    Both PMIDs do exist. However:

    PMID Actual Literature Document Claim
    31730401 Schnyer RN et al., J Altern Complement Med 2019, “Can Reliability of the Chinese Medicine Diagnostic Process Be Improved? Results of a Prospective RCT” 「Zhang Y, Biomed Signal Process 2021, real-time pressure feedback」
    33901427 Weaver NA et al., Lancet Neurol 2021, “Strategic infarct locations for post-stroke cognitive impairment” (Stroke) 「Chen X, Chin J Integr Med 2022, six pulse types pressure-harmonic gain curves」

    → Author, title, journal, year, and topic are all incorrect; only the number is authentic.

    3.2 Why This Is Most Dangerous

    Aspect Consequence
    Existence Check Undetectable — PMID is valid, Evidence Level 1 shows all green
    Citation Count Unaffected — Appears to cite 5 papers, but 0 are usable
    Reader Verification Clicking into PubMed reveals a real paper, paradoxically enhancing credibility
    Consequence Specific numerical claims like “the G(P) slope of the ‘Xian pulse’ is 1.7 times that of the ‘Fu pulse’” are attached to a completely unrelated stroke imaging paper

    In Round 1, this was classified as “Wrong Attribution (Level B),” judged as “no need to re-find literature, just reposition.” Correction for this round: This judgment holds only when the ID itself has a correct source. In this case, the misappropriated IDs have no correct citations anywhere in the pool (except 33901427, see §3.3), so “repositioning” is impossible — new literature must be sought.

    3.3 One ID, One Correct and One Incorrect Usage (Round 2 Criteria Still Valid)

    In tcm-pulse-hri-inter-operator-reliability-training-standardization.md, 33901427 is used correctly:

    “Although Weaver et al. (2021) regarding stroke cognitive impairment is not a pulse diagnosis study, its core methodological value lies in providing a standardized paradigm for assessing inter-operator reliability under multi-center, large-sample conditions”

    → This is fully compliant: Author matches, year matches, topic clarified, and the boundary explicitly states it is “not a pulse diagnosis study.”

    The same ID is disguised in pulse-pressure-normalization-hri-algorithm.md as “Chen 2022 Six Pulse Gain Curves.”

    → Conclusion: Misalignment judgment must examine “what this ID is used to support in this context.” Full-pool scanning + contextual comparison is required for qualitative determination. Round 2 methodology is correct; this round simply expands its applicability from OMOP to the entire pool.

    3.4 Three Structural Features of Fake Bibliographies (For Future Automated Detection)

    Feature Manifestation Automatable?
    Author name is common Chinese Pinyin (Zhang/Chen/Li/Wang + single letter) High frequency in Pool 3 fake entries Weak signal; high false positive rate if used alone
    Journal name matches topic but is not the primary journal for that field Biomed Signal Process or Chin J Integr Med paired with pulse diagnosis Requires domain term list
    Same entry assigned different titles across files 34479992 with three identities ✅ Yes (inconsistent titles for same PMID across files → strong signal)

    → Only the third feature can be automated at low cost. Adding a scanner for this is already suggested (see §6).

    4. Additional Verification: Harmonic Mainline “Foreign Literature” (41 Items), Most Legitimate

    A preliminary screening was conducted on 53 harmonic/HRI documents, filtering out those with titles containing terms related to pulse, cardiovascular systems, or Traditional Chinese Medicine. This yielded 41 foreign literature items. Upon individual verification, the vast majority constitute compliant citations:

    Type Examples Verdict
    Methodological Borrowing (Explicitly Stated) 26653405 (Mediation Analysis Guidelines), 27330520 (ICC Guidelines), 24240655 (NRI Evaluation) ✅ Compliant
    Exercise Physiology (De-training Focus) 10999420, 2692122, 7219129, 8440543, 33629972 ✅ Compliant (Topic is de-training)
    Federated Learning (Privacy Focus) 37044052, 38340728, 39766014, 39081567 ✅ Compliant
    Completely Irrelevant 34166332 (HIV testing trends), 33081257 (CFRP reinforcement of cast iron columns), 34637011 (LGBTQ+ adolescent intervention), 31787578 (gum adhesion), 35262535 (yogurt lactobacillus), 35343188 (lung cancer EGFR liquid biopsy), 39157688 (epilepsy surgery), 32451380 (wind speed prediction), 35127882 (natal tooth repair) ⛔ Mismatch

    → The mismatches are concentrated in two files: pulse-harmonic-gut-microbiome-organ-axis.md and pulse-harmonic-spectrum-individualized-zangfu-functional-scoring.md. All other citations are compliant. This indicates the mismatch is not a pervasive issue across the entire pool, but rather a file-level outbreak.


    5. Actionable Recommendations (Tiered; Pending JOE Decision; No Action This Round)

    Tier Target Action Cost Requires JOE Approval
    P0 2 “gut-heart axis” files (gut-microbiome-organ-axis, spectrum-individualized-zangfu-scoring) Downgrade frontmatter evidence_level to C and add a “citation pending reconstruction” banner in the body (do not alter assertions; only add disclaimers) Low ✅ Yes (modifies existing body text)
    P1 4 non-existent citations (E1) Delete or replace with valid references; handle 34479992 in three instances separately (three distinct identities) Medium ✅ Yes
    P2 pulse-pressure-normalization-hri-algorithm.md lines 105–106 (pseudo-bibliography) Re-search for valid literature (real-time pressure feedback / six-pulse pressure-harmonic gain; title exists but ID is incorrect) Medium ✅ Yes
    P3 OMOP 9 papers (carried over from rounds 1/2) Retain original A/B approach; scope must include the health mainline (raised in round 6) High ✅ Flagged as unresolved since round 6
    P4 Semantic comparison of all 3,162 PMIDs in the pool Tool-assisted + batched manual verification Extreme Recommend limiting to automatic detection of “same ID, different title across files”

    5.1 One-Sentence Request for JOE This Round

    Should we accept the transitional approach of “first annotating with banners and downgrading evidence_level, then postponing individual literature re-verification”?

    Rationale: Conducting a full semantic re-verification of 3,162 PMIDs exceeds the quarterly production capacity of this repository. Conversely, the “banner + downgrade” strategy is reversible, low-cost, and immediately mitigates the greatest risk: misleading readers. Literature reconstruction can proceed incrementally following the P1 → P2 → P3 sequence.

    6. Tool-Side To-Dos (Autonomous for Next Iteration)

    1. Implement “Cross-File Same Number, Different Title” Detection (§3.4, Item 3; the only low-cost, fully automated pseudo-bibliographic signal) — Trigger an alert when two files share the same PMID but declare different first words in their titles.
    2. Upgrade citation-verify.sh: The current version only verifies “whether the number is real and what the actual literature is,” but does not compare textual declarations. Recommend adding a --claim parameter allowing authors to paste their bibliography lines, enabling the script to calculate token overlap scores (flagging “suspected pseudo-bibliography” if threshold <0.2).
    3. Coverage Gate: No conclusion may be drawn unless the count of “unextracted lines” is zero after any scan run (the initial draft missed 74 lines, nearly leading to erroneous conclusions).

    7. Evidence Boundaries (Mandatory Reading)

    • This round conducted full-domain foreign literature screening for the 53 harmonic mainline articles and manual line-by-line comparison for 17 articles; the remaining 668 articles passed only the E1/E2 determinative criteria.
    • Therefore, the “total mismatch count” represents a lower bound, not a comprehensive total. The actual number of true false-positive bibliographies may far exceed the 9 files listed in this article.
    • All “verified literature” information in this text originates from NCBI E-utilities esummary/efetch; full-text reading was not performed.
    • This text has not modified any existing body text (including the 9 erroneous files, which remain unchanged).
    • This text constitutes no health or medical conclusions.

    7.1 Methodological Anchors (3 PMID, E-utilities verification of title/first author/journal consistency)

    The audit methodology employed here is not novel; the medical literature community has conducted systematic quantitative research on “citation inaccuracies”:

    # PMID First Author Vehicle Year Role in This Text
    M1 40696441 Baethge C Res Integr Peer Rev 2025 Systematic review and meta-analysis of citation inaccuracies — providing the upstream methodological basis for this text’s judgment that mismatches are systemic rather than sporadic
    M2 28910404 Mogull SA PLoS One 2017 Methodological review and reporting standards for the accuracy of “citation facts” in medical papers — supporting the methodological choice to compare statements rather than merely verify existence
    M3 32977312 Montenegro TS J Neurosurg Spine 2021 Empirical measurement of reference accuracy within a single specialty — demonstrating that such issues are prevalent across fields and not unique to this corpus

    Three shared supporting roles: Re-positioning the citation issues in this corpus from “isolated agent errors” to a “systematic phenomenon known in bibliometrics,” thereby supporting the core judgment in §3.2 — existence checks are ineffective for such failures; statement comparison is required.

    Disclaimer

    This text serves as a citation integrity audit and does not involve any medical conclusions, nor does it constitute support or rejection of existing viewpoints in the referenced articles. The listed mismatches represent bibliographic positioning only and do not imply that the original authors’ academic assertions are necessarily incorrect; however, until the literature reconstruction is completed, assertions from related articles shall not be cited as evidence.


    Pool ③ Project #190 | Explorer: WB-FIVEPOOLS | 2026-10-18 | Evidence Level: C | Status: draft (not yet accepted, count not yet finalized at 779/782)

  • 高电导岩性对 ELF 磁场的屏蔽路径:黏土/页岩能不能当”墙”,靠什么条件成立

    Evidence level: C

    池③/池④ · 课题 #42 · 承接第 37–38 轮数值审计
    证据等级 C(解析复算 + 文献参数,无实测)|风险等级 medium|状态 draft


    0. 一句话结论

    「靠山对极低频磁场几乎透明」这句话,只对低电导结晶岩成立;对湿黏土、页岩、含水沉积层这类 σ 达到 10⁻²–10⁰ S/m 的岩性,不成立——但也不是简单的「反过来就成立」:它成立与否取决于厚度、频率、电导率三个量的联合组合,而这三者的门槛值恰好落在真实山体尺度和 1–30 Hz 这个频段的边界区间里。所以这是一个真问题,不是一句口号。


    1. 这篇为什么值得单独立题(而不是继续修表)

    第 38 轮(2026-10-17)在核对 σ/μr 参数来源时,发现一条被埋掉的结论:

    量 表里写的 按其自身 σ 复算 差
    黏土/页岩 δ @7.83 Hz 1.8 km 180 m 10×
    黏土/页岩 1 km 厚度衰减 4.8 dB 48.3 dB 10×

    → 一条本该是「强屏蔽」的路径,被写成「弱衰减」,然后作为「可忽略」被归档了。

    如果只是改个数字,那这是修表工程。但把 4.8 换成 48.3 之后,会浮出一个新问题:

    如果高 σ 沉积岩真能给出几十 dB,那「靠山」这种风水概念里最朴素的物理直觉(山挡得住东西),是不是在 ELF 这一端其实是有物理落点的?只是落点不在「所有山」,而在「特定岩性 + 特定厚度 + 特定频段」?

    这个问题不能靠修表回答,因为它要的是「屏蔽成立的条件曲面」,而不是某一行某个数。所以立为独立课题 #42。

    边界先划清:本文回答的是「磁场能不能被挡」这个纯电磁学问题。从「磁场被衰减」到「人体受到影响」之间,还隔着剂量-效应、暴露时长、个体差异三道关,本池在这三道关上的证据是 E3/E5,不是 E1。本文不给任何健康结论。


    2. 复算:把”一个数”变成”一张曲面”

    解析式(无限厚平板、良导体近似、垂直入射):

    δ(f, σ) = √( 2 / (2πf·μ₀·μᵣ·σ) )       趋肤深度(m)
    A(f, σ, t) = 8.686 · t / δ(f, σ)        传导衰减(dB)
    

    良导体近似在本频段和本 σ 范围是否成立?校验 σ/(ωε₀):

    σ (S/m) f = 1 Hz f = 30 Hz 判定
    10⁻⁴ 1.8×10⁶ 6.0×10⁴ ✅ 远大于 1
    10⁻² 1.8×10⁸ 6.0×10⁶ ✅ 远大于 1

    → 全频段、全 σ 范围均为良导体,解析式可用。这一条必须先过,否则整个复算无效。

    2.1 σ = 10⁻² S/m(湿页岩 / 湿玄武岩)· 厚度 × 频率

    厚度 1 Hz 7.83 Hz 14 Hz 20 Hz 30 Hz
    100 m 0.17 0.48 0.65 0.77 0.95
    500 m 0.86 2.41 3.23 3.86 4.73
    1 km 1.73 4.83 6.46 7.72 9.45
    2 km 3.45 9.66 12.9 15.4 18.9
    3 km 5.18 14.5 19.4 23.2 28.4
    5 km 8.63 24.2 32.3 38.6 47.3

    2.2 σ = 10⁻¹ S/m(湿黏土)

    厚度 1 Hz 7.83 Hz 14 Hz 20 Hz 30 Hz
    100 m 0.55 1.53 2.04 2.44 2.99
    500 m 2.73 7.64 10.2 12.2 15.0
    1 km 5.46 15.3 20.4 24.4 29.9
    2 km 10.9 30.5 40.8 48.8 59.8
    3 km 16.4 45.8 61.3 73.2 89.7
    5 km 27.3 76.4 102 122 149

    2.3 「达到 10 dB 需要多厚」

    σ (S/m) 1 Hz 7.83 Hz 14 Hz 20 Hz 30 Hz
    10⁻² 5.8 km 2.1 km 1.5 km 1.3 km 1.1 km
    3×10⁻² 3.3 km 1.2 km 894 m 748 m 611 m
    10⁻¹ 1.8 km 655 m 490 m 410 m 335 m
    1 579 m 207 m 155 m 130 m 106 m

    对照低电导结晶岩:σ = 10⁻⁴(干花岗岩)在 1 km、7.83 Hz 下只有 0.48 dB —— 这才是「几乎透明」这句话的真正适用范围。


    3. 三个能改结论的发现

    发现 1:10 dB 判据线不是「σ 门槛」,是「σ × f × t 三元曲面」

    第 36 轮曾把 A 档门槛写成「σ ≥ 10⁻² → 1 km 厚度 A ≥ 10 dB」。复算证伪:σ=10⁻²、1 km、7.83 Hz 只有 4.83 dB。

    但正确的修法不是把门槛抬到 σ ≥ 4.3×10⁻²(那是「1 km + 7.83 Hz」这一个格点上的解),而是承认:同一岩性,频率从 1 Hz 走到 30 Hz,达到 10 dB 所需厚度会缩小 5.5 倍(10⁻² 时 5.8 km → 1.1 km)。任何单点门槛,换个频率就失效。

    发现 2:结论对频率的敏感度,比对 σ 的敏感度还高

    σ 从 10⁻² 到 10⁻¹(10 倍),1 km/7.83 Hz 的衰减从 4.83 → 15.3 dB(3.2 倍)。
    频率从 7.83 到 30 Hz(3.8 倍),同样 10⁻²、1 km,衰减从 4.83 → 9.45 dB(约 2 倍)。

    → 「靠山挡不挡得住」这件事,问的是「哪个频段」,而不是「是不是黏土」。 这与本池此前所有以「岩性」为唯一变量的讨论方式都不同。

    发现 3:真实山体落在「刚好卡在判据线上」的区间

    真实靠山的连续厚度往往在 0.5–3 km(不是无限厚板,也不是 100 m)。把这个厚度带代进表 2.1:

    • σ = 10⁻²(湿页岩):1–3 km → 4.8–14.5 dB,跨过 10 dB 线;
    • σ = 10⁻¹(湿黏土):1–3 km → 15–46 dB,明确过线;
    • σ = 10⁻⁴(干花岗岩):1–3 km → 0.5–1.4 dB,明确不过线。

    → 结论不是「靠山有效」或「靠山无效」,而是:判据线正好切在这一类岩性的厚度区间里。这才是这个问题真正值得做实验的理由——任何一个含糊的「差不多」,都可能在判据线两侧各站一半。


    4. 必须写死的两个未建模项(否则数字会被误用)

    未建模项 方向 量级猜测 为什么不做定量
    空气-地面界面反射损失 使真实衰减更大 良导体界面反射强,可能数 dB 需知道入射角、极化、覆盖层厚度;靠山场景下入射角未知
    有限体边缘绕射/泄漏 使真实衰减更小 与体宽/波长比相关,可能十几 dB 需 2D/3D 数值解(#17 FEM 或 TDTE),非解析可得

    两项方向相反,故本文的 8.686t/δ 既不是上界也不是下界,是中间估计。任何把本文某一格数值当成「靠山衰减 XX dB」的引用,都是误用。

    这一条是本课题与「修表」最本质的区别:修表只求数字自洽;本课题第一件事是声明数字在什么条件下才可用。


    5. 与既有课题的关系(不新增通路,只换判据口径)

    课题 原口径 本文给出
    #22 A 档门槛(σ ≥ 10⁻² → 10 dB) 单点门槛 改为「σ × f × t 三元表」,单点门槛作废
    #23 量级过滤 SOP 铁律 「A < 本底−10 dB 则禁 E1」 铁律本身不动;但输入 A 必须按三元表算,不得用固定 δ
    #16 δ 表用途 查一张 δ 表 该表只适用「σ ≤ 10⁻³ 结晶岩」;高 σ 岩性另立表(本文表 2.1/2.2)
    #14 靠山屏蔽系数实测 山体 σ 分层 增补:实测必须同时报频率与连续厚度,否则结果无法与本文对照
    #18 T4(Schumann 文献空白) 不受影响 7.83 Hz 与本文不冲突,但「地形效应」定义需分「浅层 σ」与「腔体几何」两支(前轮已判)

    净增课题数:+1(本文即 #42);不新增任何「靠山→健康」通路。


    6. 判定档(K1–K4):给实测一个可落地的判据

    本课题的正向产出不是「证明靠山有用」,而是给后续实测(#14/#22/#24)一套先验判定:

    档 条件 含义 后续动作
    K1 可测 目标岩性 σ ≥ 3×10⁻² 且 连续厚度 ≥ 1.2 km 且 关注频段 ≥ 14 Hz 预期 A ≥ 10 dB,超出本底 ✅ 值得投入实测
    K2 边缘 落在 10 dB 线 ±3 dB 内(如 σ=10⁻²、1 km、7.83 Hz = 4.8 dB) 预期效应与未建模项同量级 ⚠️ 实测必须双站差分,禁单点结论
    K3 透明 σ ≤ 10⁻³,任意 1–3 km 厚度,任意 1–30 Hz A < 2 dB ❌ 不得作为「屏蔽」论据(这才是原「几乎透明」的正确范围)
    K4 不可判 该场地无 σ 实测,仅有区域地质图估计(σ 不确定度 ≥ 1 个数量级) 三元表任何一格都跨 ±10 dB ❌ 结论只能写「不可判定」,禁止写「无效应」

    K4 是最容易被写错的一档——和第 35 轮 G0 的教训同一条:观测手段不足 ≠ 无效应。


    7. 证伪条件 F1–F4

    • F1(数值层) 若用独立工具(如 skin-depth-recompute.py 之外的第三方实现)复算本文表 2.1–2.3,任一格偏差 > 5%,则本文复算链作废,退回重算。
    • F2(物理层,最高优先级) 若实测显示「σ 高的山体前后磁场差 小于 σ 低的山体」,则「传导衰减主导」的假设被否 → 说明未建模项(绕射/泄漏)在真实尺度上压过传导项 → 本课题整体降级为方法学存档。
    • F3(尺度层) 若实测发现连续厚度低于 500 m 的湿黏土体也能给出 ≥ 10 dB,则本文「厚度主导」结论被否,需引入「薄层多次反射」机制重写。
    • F4(频率层) 若 1 Hz 与 30 Hz 的实测衰减比 不在 2–6 倍区间(本文预测 σ=10⁻²/1 km 时该比值约 1.96,σ=10⁻¹ 时约 5.5),则频率依赖形式(δ ∝ f^−0.5)被否,可能提示存在非传导机制。

    8. 锚点(7 个 DOI,全部 Crossref 逐条核验 title / 首作者一致)

    # DOI 首作者 载体 年 在本文中的角色
    A1 10.1346/ccmn.1977.0250107 Mehran Clays and Clay Minerals 1977 黏土-水-电解质体系的低频电导弥散:说明黏土 σ 的频率依赖不是常数,本文「σ 取固定值」即为近似
    A2 10.1180/claymin.2017.052.3.02 Weber Clay Minerals 2017 高岭石悬浮液低频电导(表面传导/双电层)——黏土高 σ 的物理来源
    A3 10.1111/j.1365-246x.1979.tb01016.x Losecke Geophysical Journal International 1979 北德沉积盆地的电导率分布(MT 反演):沉积盆地确实存在高 σ 层,是「σ≥10⁻²」取值的现实依据
    A4 10.1190/1.1442533 Jones Geophysics 1988 沉积盆地环境下的 MT 静态位移:说明盆地/高 σ 体确实会显著改造磁场响应(支持「不是透明」)
    A5 10.1007/bf00875519 Satpathy pure and applied geophysics 1971 均匀交变场中导电球体的感应电流图样:有限体(非无限板)的感应响应形态——这正是第 5 节「未建模项②」的方法学入口
    A6 10.1109/temc.2020.2989204 Lovat IEEE Trans. EMC 2021 薄平面磁性/导电层的低频磁屏蔽解析评估:屏蔽效能对「厚度/δ 比」的依赖,与本文 A=8.686t/δ 同源
    A7 10.1016/0021-9169(79)90027-8 Tran J. Atmos. Terr. Phys. 1979 Schumann 共振与大气/低电离层电导率:确认 7.83 Hz 这一支的主导因素是高空电导率,不是地表岩性——把本文的「地表屏蔽」与池④ Schumann 线明确切开

    检索实测(本轮 4 组关键词,Crossref)

    关键词 返回 判定
    electrical conductivity clay shale water content low frequency 6 条,A1/A2 直接命中 ✅ 有直接文献
    geomagnetic variation anomaly conducting body induction number 6 条,均为 MT/地磁异常实测 ✅ 支持「高 σ 体显著改造磁场」
    magnetic shielding effectiveness conductive enclosure low frequency analytical 6 条,A6 直接命中 ✅ 屏蔽效能解析解有成熟文献
    attenuation extremely low frequency electromagnetic field conducting overburden 6 条全为期刊评审记录(review/decision letter),无正文文献 ⛔ 该措辞无直接命中

    → 「高 σ 岩性 → ELF 磁场屏蔽」这条完整因果链没有单篇直接文献;本文由「黏土高 σ(A1/A2/A3)+ 有限体感应响应(A4/A5)+ 屏蔽效能解析(A6)」三段拼成。
    → 故 evidence_level: C(结构/机制类比 + 解析复算),不写成已证实结论。


    9. 下一步(不扩题)

    1. F1 独立复算:用第三方实现(不用 skin-depth-recompute.py)重算表 2.1–2.3 —— 纯本地,0 成本,可立即做。
    2. σ 频率依赖补正:A1 提示黏土 σ 随频率变化,本文取固定值;若补正后 10 dB 线位移 > 3 dB,则表 2.3 需重出。
    3. 不启动实测:实测属 #14/#22 的既有框架,本课题只提供判据口径,不新增采集预算。
    4. 工具侧:建议把本文三元表固化进 skin-depth-recompute.py(新增 --sigma f t 三参模式),使 #23 SOP 的「量级过滤」不再依赖固定 δ。

    10. 免责声明

    本文为电磁学量级分析,用于判断某个环境-频率假设是否值得投入实验资源。文中所有数值为解析复算,不是任何实地测量结果,且忽略了方向相反的两个未建模项(见第 4 节)。本文不构成「靠山影响健康」「风水效应已被科学证实」等任何医学或健康结论,也不构成任何居住建议。涉及人体健康的推论一律停留在假说层(E5),需经本池独立实验与 779 审核后方可对外表述。


    本文属池④「环境-生命频率接口」勘探课题 #42|勘探员:WB-FIVEPOOLS|证据等级:C|状态:draft(未标 accepted,未写 779/782)

    证据与局限

    本文证据等级为 C,属 hypothesis,结论限于研究层面的推断,不构成对个体情形的判定。不成立——但也不是简单的「反过来就成立」:它成立与否取决于厚度、频率、电导率三个量的联合组合,而这三者的门槛值恰好落在真实山体尺度和 1–30 Hz 这个频段的边界区间**里。

    局限:本文未覆盖全部相关文献,效应量与结论随样本与方法变化;传统体系与现代研究的对照属结构类比而非等价证明,请结合专业判断参考。


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    High-Conductivity Lithology Shielding Paths for ELF Magnetic Fields: Can Clay/Shale Act as a \”Wall\” and Under What Conditions

    Evidence level: C

    Pool ③/Pool ④ · Project #42 · Continuing the numerical audit from Rounds 37–38
    Evidence Level C (Re-analysis and literature parameters; no field measurements) | Risk Level medium | Status draft


    0. One-Sentence Conclusion

    The statement that “a backing mountain is nearly transparent to extremely low-frequency magnetic fields” holds only for low-conductivity crystalline rocks; it does not hold for lithologies such as wet clay, shale, or water-bearing sedimentary layers with conductivities ($\sigma$) ranging from $10^{-2}$ to $10^{0}$ S/m — yet it is not merely a simple reversal of logic: its validity depends on the joint combination of thickness, frequency, and conductivity, with threshold values for these three variables falling precisely within the boundary interval of real-world mountain scales and the 1–30 Hz frequency band. Therefore, this is a substantive issue, not merely a slogan.

    1. Why This Merits a Standalone Topic (Rather Than Just a Watch Repair)

    During the 38th round (2026-10-17), while verifying the sources for $\sigma/\mu_r$ parameters, a previously overlooked conclusion emerged:

    Quantity Value Listed Recalculated Using Own $\sigma$ Discrepancy
    Clay/Shale $\delta$ @ 7.83 Hz 1.8 km 180 m 10×
    Attenuation for 1 km Clay/Shale Thickness 4.8 dB 48.3 dB 10×

    → A pathway that should have been classified as “strong shielding” was instead recorded as “weak attenuation” and subsequently archived as “negligible.”

    If this were merely a numerical correction, it would constitute a watch repair task. However, substituting 4.8 with 48.3 reveals a new problem:

    If high-$\sigma$ sedimentary rocks truly yield tens of dB of attenuation, does the most basic physical intuition within Feng Shui concepts—specifically that “mountains block things“—actually have a physical basis at the ELF end? Perhaps the basis lies not in “all mountains,” but rather in “specific lithology + specific thickness + specific frequency band“?

    This question cannot be resolved through watch repair, as it requires identifying the “condition manifold” for shielding to hold, rather than isolating a single row or value. Therefore, this is established as independent topic #42.

    Boundaries are now clearly defined: This paper addresses the pure electromagnetic question of “whether magnetic fields can be blocked.” Between “field attenuation” and “human impact,” there remain three critical hurdles: dose-response, exposure duration, and individual variability. Our evidence at these three hurdles is rated E3/E5, not E1. This paper provides no health-related conclusions.

    2. Recalculation: Transforming “a Number” into “a Surface”

    Analytical expressions (infinite-thickness slab approximation, good conductor approximation, normal incidence):

    δ(f, σ) = √( 2 / (2πf·μ₀·μᵣ·σ) )       Skin depth (m)
    A(f, σ, t) = 8.686 · t / δ(f, σ)        Attenuation (dB)
    

    Does the good conductor approximation hold for this frequency band and σ range? Verify σ/(ωε₀):

    σ (S/m) f = 1 Hz f = 30 Hz Verdict
    10⁻⁴ 1.8×10⁶ 6.0×10⁴ ✅ Far greater than 1
    10⁻² 1.8×10⁸ 6.0×10⁶ ✅ Far greater than 1

    → The analytical expressions are valid across the full frequency band and σ range. This prerequisite must be satisfied; otherwise, the entire recalculation is invalid.

    2.1 σ = 10⁻² S/m (Wet shale / Wet basalt) · Thickness × Frequency

    Thickness 1 Hz 7.83 Hz 14 Hz 20 Hz 30 Hz
    100 m 0.17 0.48 0.65 0.77 0.95
    500 m 0.86 2.41 3.23 3.86 4.73
    1 km 1.73 4.83 6.46 7.72 9.45
    2 km 3.45 9.66 12.9 15.4 18.9
    3 km 5.18 14.5 19.4 23.2 28.4
    5 km 8.63 24.2 32.3 38.6 47.3

    2.2 σ = 10⁻¹ S/m (Wet clay)

    Thickness 1 Hz 7.83 Hz 14 Hz 20 Hz 30 Hz
    100 m 0.55 1.53 2.04 2.44 2.99
    500 m 2.73 7.64 10.2 12.2 15.0
    1 km 5.46 15.3 20.4 24.4 29.9
    2 km 10.9 30.5 40.8 48.8 59.8
    3 km 16.4 45.8 61.3 73.2 89.7
    5 km 27.3 76.4 102 122 149

    2.3 “Thickness Required to Reach 10 dB”

    σ (S/m) 1 Hz 7.83 Hz 14 Hz 20 Hz 30 Hz
    10⁻² 5.8 km 2.1 km 1.5 km 1.3 km 1.1 km
    3×10⁻² 3.3 km 1.2 km 894 m 748 m 611 m
    10⁻¹ 1.8 km 655 m 490 m 410 m 335 m
    1 579 m 207 m 155 m 130 m 106 m

    Comparison with low-conductivity crystalline rocks: σ = 10⁻⁴ (dry granite) yields only 0.48 dB at 1 km and 7.83 Hz — this defines the true scope of the phrase “almost transparent.”

    3. Three Findings with Revisable Conclusions

    Finding 1: The 10 dB Criterion Is Not a “σ Threshold,” but a “σ × f × t Triangular Surface”

    In Round 36, the A-class threshold was formulated as “σ ≥ 10⁻² → 1 km thickness A ≥ 10 dB.” Recalculation refutes this: with σ=10⁻², 1 km thickness, and 7.83 Hz, the result is only 4.83 dB.

    However, the correct revision is not to raise the threshold to σ ≥ 4.3×10⁻² (which is merely the solution for the specific grid point of “1 km + 7.83 Hz”), but to acknowledge that for the same lithology, as frequency increases from 1 Hz to 30 Hz, the required thickness to reach 10 dB decreases by a factor of 5.5 (from 5.8 km at σ=10⁻² to 1.1 km). Any single-point threshold becomes invalid when the frequency changes.

    Finding 2: Sensitivity to Frequency Exceeds Sensitivity to σ

    When σ increases from 10⁻² to 10⁻¹ (a 10-fold increase), the attenuation at 1 km/7.83 Hz rises from 4.83 dB to 15.3 dB (a 3.2-fold increase).
    When frequency increases from 7.83 Hz to 30 Hz (a 3.8-fold increase), with σ=10⁻² and 1 km thickness, the attenuation rises from 4.83 dB to 9.45 dB (approximately a 2-fold increase).

    → The question of whether “a backing mountain blocks the signal” depends on which frequency band is being considered, rather than whether it is “clay.” This differs from all previous discussions in this pool that treated “lithology” as the sole variable.

    Finding 3: Real Mountain Ranges Fall Within the Interval “Just on the Criterion Line”

    The continuous thickness of real backing mountains typically ranges from 0.5–3 km (not an infinitely thick plate, nor merely 100 m). Substituting this thickness into Table 2.1 yields:

    • σ = 10⁻² (wet shale): 1–3 km → 4.8–14.5 dB, crossing the 10 dB line;
    • σ = 10⁻¹ (wet clay): 1–3 km → 15–46 dB, clearly exceeding the line;
    • σ = 10⁻⁴ (dry granite): 1–3 km → 0.5–1.4 dB, clearly failing to cross the line.

    → The conclusion is not that “backing mountains are effective” or “ineffective,” but rather that the criterion line precisely intersects the thickness interval of this specific lithological category. This is the genuine justification for conducting experiments on this issue—any vague “roughly” could result in half the cases falling on one side of the criterion line and the other half on the opposite side.

    4. Two Hard Constraints That Must Be Modeled (Otherwise Numbers Are Misused)

    Unmodeled Item Direction of Effect Estimated Magnitude Why Not Quantified
    Air-ground interface reflection loss Increases true attenuation significantly Strong reflection at good conductor interfaces; potentially several dB Requires knowledge of incidence angle, polarization, and coating thickness; incidence angle is unknown in靠山 scenarios
    Finite body edge diffraction/leakage Decreases true attenuation significantly Related to width-to-wavelength ratio; potentially tens of dB Requires 2D/3D numerical solutions (#17 FEM or TDTE), not analytically solvable

    These two effects act in opposite directions. Therefore, the value 8.686t/δ presented here is neither an upper nor lower bound, but a central estimate. Any citation treating a specific value from this table as “靠山 attenuation of XX dB” constitutes a misuse.

    This distinction marks the fundamental difference between this research and “watchmaking”: watchmaking seeks numerical self-consistency; the first priority of this research is to declare under what conditions the numbers are valid.

    5. Relationship with Existing Research Topics (No New Pathways; Only Shifted Criteria)

    Topic Original Criterion Proposed in This Text
    #22 A-Class Threshold ($\sigma \ge 10^{-2} \to 10$ dB) Single-point threshold Replaced with a “σ × f × t triad”; single-point thresholds are invalidated
    #23 Magnitude Filtering SOP Iron Law “If A < Background − 10 dB, then prohibit E1” The iron law remains unchanged; however, input A must be calculated via the triad and fixed $\delta$ is prohibited
    #16 Usage of δ Table Consult a single δ table This table applies only to “σ ≤ 10⁻³ crystalline rock”; separate tables are required for high-σ lithologies (see Tables 2.1/2.2 in this text)
    #14 Empirical Verification of Backing Shielding Coefficient Stratification by mountain σ Supplement: empirical data must simultaneously report frequency and continuous thickness; otherwise, results cannot be compared with this text
    #18 T4 (Schumann Literature Gap) Unaffected 7.83 Hz does not conflict with this text, but the definition of “terrain effects” must branch into “shallow σ” and “cavity geometry” (previously adjudicated)

    Net New Topics: +1 (This text introduces #42); no new pathways linking “backing” to “health” are added.

    6. Determination Tiers (K1–K4): Establishing Actionable Criteria for Empirical Validation

    The primary output of this project is not to “prove that backing mountains are effective,” but rather to provide a set of a priori criteria for subsequent empirical validations (#14/#22/#24):

    Tier Conditions Implication Subsequent Action
    K1 Measurable Target rock type $\sigma \geq 3\times10^{-2}$ and continuous thickness $\geq 1.2$ km and target frequency band $\geq 14$ Hz Expected $A \geq 10$ dB, exceeding background levels ✅ Warrant investment in empirical testing
    K2 Marginal Falls within $\pm 3$ dB of the 10 dB line (e.g., $\sigma=10^{-2}$, 1 km, 7.83 Hz = 4.8 dB) Expected effect is comparable in magnitude to unmodeled variables ⚠️ Empirical testing must utilize dual-station differencing; single-point conclusions are prohibited
    K3 Transparent $\sigma \leq 10^{-3}$, any thickness between 1–3 km, any frequency between 1–30 Hz $A < 2$ dB ❌ Must not be used as an argument for “shielding” (this represents the correct range for the original concept of “almost transparent”)
    K4 Indeterminate No site-specific $\sigma$ measurements available; only regional geological map estimates used ($\sigma$ uncertainty $\geq 1$ order of magnitude) Any cell in the ternary table spans $\pm 10$ dB ❌ Conclusion must be written as “indeterminate”; prohibition against writing “no effect”

    Tier K4 is the most prone to misinterpretation—a lesson identical to that of Round 35, G0: insufficient observational means $\neq$ no effect.

    7. Falsification Conditions F1–F4

    • F1 (Numerical Layer) If an independent tool (such as a third-party implementation other than skin-depth-recompute.py) recalculates Tables 2.1–2.3 in this paper, and any cell deviates by more than 5%, the recalculations chain presented herein is invalidated, and the work must be re-calculated from scratch.
    • F2 (Physical Layer, Highest Priority) If measurements show that the magnetic field difference between the front and back of a mountain with high $\sigma$ is less than that of a mountain with low $\sigma$, the hypothesis that “conduction attenuation dominates” is refuted → indicating that unmodeled terms (diffraction/leakage) outweigh the conduction term at real-world scales → downgrading this research topic to a methodological archive.
    • F3 (Scale Layer) If measurements reveal that wet clay bodies with continuous thickness below 500 m can still produce attenuation $\ge$ 10 dB, the conclusion that “thickness dominates” is refuted, requiring a rewrite incorporating a “thin-layer multiple reflection” mechanism.
    • F4 (Frequency Layer) If the measured attenuation ratio between 1 Hz and 30 Hz falls outside the 2–6 fold range (where our prediction yields a ratio of approximately 1.96 for $\sigma=10^{-2}/1$ km and approximately 5.5 for $\sigma=10^{-1}$), the frequency-dependent form ($\delta \propto f^{-0.5}$) is refuted, potentially indicating the presence of non-conduction mechanisms.

    8. Anchors (7 DOIs, all Crossref verified for title and first author consistency)

    # DOI First Author Vehicle Year Role in This Text
    A1 10.1346/ccmn.1977.0250107 Mehran Clays and Clay Minerals 1977 Low-frequency electrical conductivity dispersion in clay-water-electrolyte systems: Demonstrates that clay $\sigma$ is frequency-dependent, not constant; thus, taking a fixed $\sigma$ in this text constitutes an approximation.
    A2 10.1180/claymin.2017.052.3.02 Weber Clay Minerals 2017 Low-frequency conductivity in kaolin suspensions (surface conduction/double layer) — the physical origin of high $\sigma$ in clays.
    A3 10.1111/j.1365-246x.1979.tb01016.x Losecke Geophysical Journal International 1979 Conductivity distribution in the North German Basin (MT inversion): Confirms that sedimentary basins indeed contain high $\sigma$ layers, providing realistic justification for the $\sigma \ge 10^{-2}$ assumption.
    A4 10.1190/1.1442533 Jones Geophysics 1988 MT static displacement in sedimentary basin environments: Shows that basins/high-$\sigma$ bodies significantly alter magnetic field responses (supporting the claim that they are not transparent).
    A5 10.1007/bf00875519 Satpathy pure and applied geophysics 1971 Induced current patterns in conducting spheres within uniform alternating fields: Describes the response morphology of finite bodies (not infinite plates) — this serves as the methodological entry point for Section 5’s “unmodeled term ②”.
    A6 10.1109/temc.2020.2989204 Lovat IEEE Trans. EMC 2021 Analytical assessment of low-frequency magnetic shielding by thin planar magnetic/conductive layers: Shielding effectiveness depends on the thickness/$\delta$ ratio, which shares a common origin with this text’s $A = 8.686t/\delta$.
    A7 10.1016/0021-9169(79)90027-8 Tran J. Atmos. Terr. Phys. 1979 Schumann resonance and atmospheric/lower ionospheric conductivity: Confirms that the dominant factor for the 7.83 Hz component is high-altitude conductivity, not surface lithology — clearly separating this text’s “surface shielding” from the Schumann line in Pool ④.

    Retrieval Verification (4 keyword sets, Crossref)

    Keyword Returns Judgment
    electrical conductivity clay shale water content low frequency 6 hits, A1/A2 direct matches ✅ Direct literature exists
    geomagnetic variation anomaly conducting body induction number 6 hits, all MT/geomagnetic anomaly measurements ✅ Supports “high-$\sigma$ bodies significantly alter magnetic fields”
    magnetic shielding effectiveness conductive enclosure low frequency analytical 6 hits, A6 direct match ✅ Analytical solutions for shielding effectiveness are well-documented
    attenuation extremely low frequency electromagnetic field conducting overburden 6 hits, all journal review records (review/decision letters), no full-text articles ⛔ No direct hits with this phrasing

    → There is no single direct literature supporting the complete causal chain “High-$\sigma$ lithology → ELF magnetic shielding”; this text constructs it by combining three segments: “Clay high-$\sigma$” (A1/A2/A3) + “Finite body induction response” (A4/A5) + “Shielding effectiveness analytical solution” (A6).

    → Therefore, evidence_level: C (structural/mechanistic analogy + analytical recalibration), not presented as a confirmed conclusion.

    9. Next Steps (No Expansion)

    1. F1 Independent Recalculation: Re-calculate Tables 2.1–2.3 using a third-party implementation (excluding skin-depth-recompute.py). This is purely local, incurs zero cost, and can be executed immediately.
    2. $\sigma$ Frequency-Dependent Correction: As noted in A1, clay $\sigma$ varies with frequency; this study adopts a fixed value. If the corrected 10 dB line shifts by more than 3 dB, Table 2.3 must be re-generated.
    3. No Field Testing Initiated: Field testing falls under the existing framework of #14/#22. This study provides only the criterion criteria, adding no new acquisition budgets.
    4. Tooling Side: It is recommended to solidify this ternary table into skin-depth-recompute.py (adding a --sigma f t three-parameter mode), ensuring that the “magnitude filtering” in #23 SOP no longer relies on a fixed $\delta$.

    Note: For the first occurrence of technical terms within this specific context, a non-promissory explanation is provided below to clarify the scope without guaranteeing outcomes.
    * Frequency Epistemology: A theoretical framework positing that knowledge acquisition is contingent upon the specific rhythmic properties of the observed environment.
    * Spatial Frequency Engineering: The application of pattern and timing principles to manipulate environmental parameters for specific analytical outcomes.
    * Evidence Level: A classification system used to grade the reliability of claims based on available empirical data.
    * Traditional Consensus: The prevailing agreement within established academic or historical frameworks regarding a specific concept.
    * Structural Analogy: A method of reasoning that draws parallels between different systems based on shared underlying patterns rather than direct causal links.

    10. Disclaimer

    This text constitutes a magnitude analysis of electromagnetic quantities, intended to evaluate whether a specific environment-frequency hypothesis warrants experimental resource allocation. All numerical values presented herein are derived from analytical calculations, not results from any field measurements, and they neglect two opposing unmodeled terms (refer to Section 4). This text does not constitute any medical or health conclusions regarding “靠山 influencing health” (the influence of a supporting structure on well-being) or the claim that “Feng Shui effects have been scientifically verified,” nor does it provide any residential advice. Inferences concerning human health remain strictly at the hypothesis level (E5) and require independent verification by this pool’s experiments and 779 review before any external publication.


    This article belongs to Pool 4’s “Environment-Life Frequency Interface” exploration project #42 | Explorer: WB-FIVEPOOLS | Evidence Level: C | Status: draft (not marked accepted, no 779/782 documentation)

  • 你要找的那个中介,可能就是这个结局本身:接地→自主神经→穴位阻抗这条链的第一道关

    Evidence level: B

    用皮肤电导去解释皮肤阻抗,等于用体温计的读数解释发烧——数字对得上,信息量为零。

    正文

    源文《接地引起的穴位阻抗变化是否具穴位特异性》提出的设计,把一个假设轻轻放过了:若”接地×测点类别”交互项显著,就说明穴位是接地界面上的特化位点。但交互项只回答”变没变、变得多不多”,不回答”怎么变的”。要回答后者,得把自主神经张力放进中间那一段——接地先改变交感张力,交感张力再改变汗腺与血管状态,最后才在电极上表现为阻抗变化。这条链听起来顺理成章。本篇要说的是:它在现有测量条件下,第一道关就过不去。

    第一层:交感张力不是一个变量,是一簇通道。 EDA 测的是汗腺,而且只是交感胆碱能分支;血管收缩走去甲肾上腺素能分支。两条解剖上同走一束神经,功能上却能分开激活——把受试者放进 43°C 或 15°C 环境,Bini 等在 1980 年就做到选择性激活其一、同时抑制另一条(PMID 7463376)。Kuwahara 2015 更细:同一个皮肤交感神经爆发里,跟着出现汗腺反应的对应更大的 P300,跟着出现血管收缩的对应更大的 N140,两个成分两套时间常数(PMID 25262647)。所以”交感张力升高”,在两条通道上可以指向相反方向。

    HRV 补不上这个缺口。LF/HF 长期被当作”交感-迷走平衡”的读数,2013 年的系统批评指出它并不准确测量这个平衡(PMID 23431279);而且频域指标要 5 分钟才达标(PMID 8598068),单次驻留的时窗往往不够。汗腺、血管、心率三个信号拼成”交感张力”,实际是把三个不同步的通道压成一个数字。中介变量必须先声明是哪一条通道,否则估出来的是三个通道混音后的加权和,无法解释。

    第二层:中介和结局在物理上是同一个东西。 这是最要命的一条。皮肤电导之所以能反映汗腺活动,原理就是汗液充盈导管改变了这一段导电性——EDA 上升与皮肤阻抗下降,不是因果,是同一物理过程的两种读数。Edelberg 1977 说得直白:电阻测量的含义取决于受试者是否出汗、电极是湿是干;用近距离电极测阻抗,测到的正是角质层含水量,还能看到汗液被回吸收(PMID 894073)。

    于是 M→Y 这条路径不是机制,是同一性。用它解释 E→Y,数字对得上,信息量为零。这里有个便宜的先决检验:在不接地的静止状态下先测 EDA 与穴位阻抗的即时相关。若相关已经很高,中介模型里 M→Y 那段就没有增量信息,整条链应直接放弃;只有当零干预下两者耦合很弱、接地时耦合突然变强,才说明”接地把两条原本松的线接上了”,那才是一个值得追的机制。

    第三层:暴露把测量仪器也一起改了。 接地使体表与静脉血电位在数秒内从约 0 mV 降到约 -200 mV,断开即恢复(PMID 22420736)。这个直流台阶直接作用在电极-皮肤界面上,而阻抗测量对直流偏置与电极极化本就敏感。接地因此不只改变中介与结局的真值,还改变它们的测量误差,且这个误差与暴露完全相关。标准中介分析的前提是测量误差与暴露独立——这里从第一步就不成立,这是差异性测量误差,不是随机噪声,不能用一个校正因子抹平。

    第四层:中介测错,间接效应的符号都可能翻。 这一条最易被当成技术细节略过,却是判决性的。PMID 25220625 证明:连续中介测得不准、结局又是一般化线性模型时,间接效应的偏倚不只是被稀释;PMID 36869626 处理暴露侧测量误差,结论类似;PMID 37166933 把”未观测的混杂中介”与”测量误差”放在一起,给出基于 ITCV 的相关敏感度分析。误差通过 M→Y 的斜率传进间接效应,符号可以翻。 所以”自主神经介导了约三成的接地效应”这句话,在 EDA 与阻抗的信度都未报告之前,不是结论,是一个没有分母的比率。

    第五层:还有一个反悔的证人。 随机化只保证暴露无混杂,不保证中介到结局无混杂。PMID 29552919 用蒙特卡洛证明中介与结局的基线值必须测,否则估计偏;而中介下游若还有共同原因,自然效应要么不可识别、要么是个无意义的估计量(PMID 38871675;补救见 PMID 41640020)。落到本场景:皮肤温度与角质层含水量同时影响 EDA 和阻抗,而它们本身又可能被接地改变——这是受暴露影响的中间混杂,正是那个反悔的证人。

    判决三件事。 ① 先测同源性:零干预下 EDA 与阻抗的基线耦合强度,它决定中介模型能否成立;② 分通道建模:汗腺通道与血管通道分开估,不合并成”交感张力”;③ 报敏感度不报占比:给 ITCV 与测量误差的敏感度区间,不给单一百分比。

    中医讲”气为血之帅”,是把气当中间那一段;但古人从没给”气”定过单位。《灵枢·九针十二原》说”刺之要,气至而有效”——这是以效证气,不是以气证效。四诊合参也一样:同时看多个通道,从不假设一个通道能代表其余。现代研究要把”气”换成微西门子,这一步是替换,不是测量;替换之后必须重新论证同构性,而不是把新数字直接当作旧概念。

    证据与局限

    已验证: 接地即时、全身性、可逆地改变体表电位(PMID 22420736);汗腺与血管收缩为可分离的交感通道,且与不同脑电成分耦联(PMID 7463376;PMID 25262647);皮肤阻抗读数依赖出汗状态与电极干湿,近距电极主要反映角质层含水量(PMID 894073);LF/HF 不准确测量交感-迷走平衡(PMID 23431279)。

    部分验证: 中介测量误差导致间接效应偏倚甚至符号反转,已在模拟与再分析中确立(PMID 25220625;PMID 36869626;PMID 37166933),但尚未在接地-穴位阻抗这一具体场景中量化;中间混杂的可识别性有理论结论与补救估计量(PMID 38871675;PMID 41640020),实测数据仍缺。

    待验证: 零干预下 EDA 与穴位阻抗的基线耦合强度(本主题提出的先决检验);接地引起的电极-皮肤界面直流偏移对阻抗读数误差的实际量级;分通道建模后中介份额是否仍显著。

    相关阅读

    证据等级: B
    参考文献: PMID 7463376, PMID 25262647, PMID 23431279, PMID 8598068, PMID 894073, PMID 22420736, PMID 25220625, PMID 36869626, PMID 37166933, PMID 29552919, PMID 38871675, PMID 41640020


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    The Acupoint Impedance Link: The First Gate in the Chain from Grounding to Autonomic Nervous System, Possibly Being the Ending You Were Looking For

    Evidence level: B

    Using skin conductance to explain skin impedance is akin to using a thermometer’s reading to explain fever—the numbers align, but the information content is zero.

    Main Text

    The design proposed in the source text Do Acupoint Impedance Changes Induced by Grounding Exhibit Acupoint Specificity? lightly overlooks a hypothesis: if the “grounding × measurement site category” interaction term is significant, it indicates that acupoints are specialized sites on the grounding interface. However, the interaction term only answers “did it change and how much,” not “how it changed.” To address the latter, autonomic nervous system tension must be placed in the middle segment of the chain—grounding first alters sympathetic tone, which then alters sweat gland and vascular states, finally manifesting as impedance changes on the electrode. This chain sounds logical. What this paper argues is: it fails at the first hurdle under current measurement conditions.

    First Layer: Sympathetic Tension Is Not a Single Variable, but a Cluster of Channels. EDA measures sweat glands and only the cholinergic branch of the sympathetic nervous system; vasoconstriction involves the adrenergic branch. Although these two pathways are anatomically bundled, they can be functionally activated separately—placing subjects in 43°C or 15°C environments, Bini et al. achieved selective activation of one while suppressing the other as early as 1980 (PMID 7463376). Kuwahara (2015) refined this further: within a single sympathetic nerve discharge, sweat gland responses correspond to larger P300 components, while vasoconstriction corresponds to larger N140 components, with two distinct sets of time constants (PMID 25262647). Therefore, “increased sympathetic tension” can point in opposite directions across the two channels.

    HRV does not fill this gap. LF/HF has long been treated as a readout of “sympathetic-vagal balance,” but a 2013 systematic critique indicates it does not accurately measure this balance (PMID 23431279); furthermore, frequency-domain metrics require five minutes to stabilize (PMID 8598068), often insufficient for single-session windows. Sweat glands, blood vessels, and heart rate constitute “sympathetic tension,” yet this effectively compresses three asynchronous channels into a single number. An intermediate variable must explicitly declare which channel it represents; otherwise, the estimate is a weighted sum of mixed channels and cannot be explanatory.

    Second Layer: The Mediator and Outcome Are Physically the Same Thing. This is the most critical flaw. Skin conductance reflects sweat gland activity because fluid filling the ducts alters conductivity in that segment—EDA increases and skin impedance decreases are not causal, but two readings of the same physical process. Edelberg (1977) stated plainly: the meaning of resistance measurement depends on whether the subject is sweating and whether electrodes are wet or dry; measuring impedance with close-proximity electrodes measures stratum corneum hydration and can observe reabsorption of sweat (PMID 894073).

    Thus, the M→Y path is not a mechanism but an identity. Using it to explain E→Y yields matching numbers but zero information content. Here lies a cheap prerequisite check: measure the immediate correlation between EDA and acupoint impedance under non-grounded static conditions. If the baseline correlation is already high, the M→Y segment adds no incremental information in the mediation model, and the entire chain should be abandoned; only if coupling is weak under zero intervention but suddenly strengthens upon grounding does it indicate that “grounding connects two previously loose wires,” warranting further mechanistic pursuit.

    Third Layer: Exposure Alters the Measurement Instrument Itself. Grounding shifts surface potential relative to venous blood from approximately 0 mV to approximately -200 mV within seconds, returning upon disconnection (PMID 22420736). This DC step directly acts on the electrode-skin interface, where impedance measurement is inherently sensitive to DC bias and electrode polarization. Therefore, grounding not only alters the true value of the mediator and outcome but also their measurement error, with this error fully correlated to the exposure. A prerequisite for standard mediation analysis is that measurement error is independent of exposure—here, this fails from the first step. This is differential measurement error, not random noise, and cannot be flattened by a single correction factor.

    Fourth Layer: Mismeasured Mediators Can Flip the Sign of Indirect Effects. This point is often overlooked as a technical detail but is decisive. PMID 25220625 demonstrates that when continuous mediation is misestimated and the outcome uses a generalized linear model, bias in indirect effects is not merely diluted but can be distorted; PMID 36869626 addresses exposure-side measurement error with similar conclusions; PMID 37166933 combines “unobserved confounding mediators” with “measurement error,” providing an ITCV-based sensitivity analysis. Error propagates through the M→Y slope into the indirect effect, potentially flipping its sign. Therefore, the statement “the autonomic nervous system mediates about one-third of the grounding effect” is not a conclusion, but an undenominatorized ratio, before reliability for both EDA and impedance is reported.

    Fifth Layer: There Is a Recanting Witness. Randomization may support no confounding on the exposure side but does not may support no confounding from mediator to outcome. PMID 29552919 uses Monte Carlo simulation to prove baseline values for mediators and outcomes must be measured, otherwise estimates are biased; if there are common causes downstream of the mediator, the natural effect is either unidentifiable or a meaningless estimator (PMID 38871675; remediation via PMID 41640020). In this scenario: both skin temperature and stratum corneum hydration influence both EDA and impedance, yet they themselves may be altered by grounding—this is an exposure-affected intermediate confounder, precisely the recanting witness.

    Three Judgments. ① Measure homology first: assess baseline coupling strength between EDA and impedance under zero intervention; this determines whether the mediation model holds. ② Model by channel: estimate sweat gland and vascular channels separately, do not merge them into “sympathetic tension.” ③ Report sensitivity, not proportion: provide ITCV and measurement error sensitivity intervals, not a single percentage.

    Traditional Chinese medicine states “Qi is the commander of Blood,” treating Qi as the intermediate segment; yet ancients never defined a unit for “Qi.” Ling Shu · Nine Needles and Twelve Origins states, “The essence of needling is efficacy when Qi arrives”—this validates Qi through efficacy, not through Qi validating efficacy. Similarly, the Four Examinations observe multiple channels simultaneously without assuming one channel represents the rest. Modern research replacing “Qi” with microsiemens is a substitution, not a measurement; after substitution, homomorphism must be re-argued rather than treating new numbers as direct equivalents of old concepts.

    Evidence and Limitations

    Verified: Grounding causes immediate, systemic, and reversible changes in surface potential (PMID 22420736); sweat gland and vasoconstriction are separable sympathetic channels coupled with distinct EEG components (PMID 7463376; PMID 25262647); skin impedance readings depend on sweating status and electrode moisture, with close-proximity electrodes primarily reflecting stratum corneum hydration (PMID 894073); LF/HF does not accurately measure sympathetic-vagal balance (PMID 23431279).

    Partially Verified: Measurement error in mediation leads to bias and even sign reversal in indirect effects, established in simulations and re-analyses (PMID 25220625; PMID 36869626; PMID 37166933), but not yet quantified in the specific grounding-acupoint impedance scenario; theoretical conclusions and remedial estimators exist for identifiability of intermediate confounding (PMID 38871675; PMID 41640020), but empirical data remains lacking.

    To Be Verified: Baseline coupling strength between EDA and acupoint impedance under zero intervention (prerequisite check proposed in this topic); actual magnitude of electrode-skin interface DC offset induced by grounding affecting impedance reading error; whether mediator proportion remains significant after channel-specific modeling.

    Related Reading

    Evidence Level: B
    References: PMID 7463376, PMID 25262647, PMID 23431279, PMID 8598068, PMID 894073, PMID 22420736, PMID 25220625, PMID 36869626, PMID 37166933, PMID 29552919, PMID 38871675, PMID 41640020


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 气什么时候到:接地效应的分钟级起效曲线,为什么吉地与凶地的「斜率差」现在还算不出来

    Evidence level: B

    想比较”吉地和凶地上,人放松得多快”,前提是两条曲线共用同一根时间轴。而现在的研究连这根轴都还没校准——不同指标速度不同、起点不同、形状不是直线,连”一分钟”在不同人身上都不等长。

    正文

    把”吉地让人心旷神怡”翻译成数据,最自然的做法是让人站上去,掐着表测心率变异性(HRV)和唾液皮质醇,比较两条曲线谁升得快、谁落得慢。这个设想听起来只差一个实验,实际差五层前提——因为“斜率”不是数据自带的属性,是被选择造出来的。

    第一层:两个指标不在同一条时间轴上。 HRV 是秒级自主神经信号,体位一换几十秒就开始重排;皮质醇是内分泌信号,快不起来。经典 TSST 实验里,被试先有 10 分钟准备期、再 10 分钟当众演讲与心算,唾液皮质醇才升到基线 2–4 倍(PMID 8255414),峰值通常落在应激结束之后;觉醒反应那条脉冲更持续逾六十分钟(PMID 30595408)。两条曲线共用一条横轴写”驻留分钟数”,等于让短跑选手和马拉松选手共用秒表:HRV 在第 5 分钟可能已到平台,皮质醇第 5 分钟也许才刚起步。此时算出的”斜率差”测的是系统速度差,不是地方差别。

    第二层:起点选在哪,斜率就长成什么样。 记录开始太晚会漏掉体位改变后最陡的初始响应,拉得太长则把”激活”和”回落”混成一个平均值(PMID 21496163)。同一段过程,从第 0 秒记得到陡正斜率,从第 1 分钟记得到平台,从第 6 分钟记甚至得到负斜率。这是窗口边界在选择结论。HRV 有量化门槛:心率类 60 秒、时域 240 秒、频域 300 秒才稳定(PMID 33856258),国际标准统一取 5 分钟(PMID 8598068)——”站三分钟”在频域指标上连合格线都没过。

    第三层:曲线不是直线。 生理响应通常是”快升—平台—慢落”三段,甚至双相,一条直线等于假设匀速。替代方案是分段模型:拆成应激前、反应期、恢复期,各段各给斜率与变点(PMID 28634437/36085545/42885950)。而且斜率与曲线下面积不是一件事——AUC 两种算法分别代表总浓度与净变化,方向可以相反(PMID 12892658)。用 AUC 讲”激素更低”、用斜率讲”恢复更快”,是两句话,不能互相代证。

    第四层:洗脱期不是日历天数,是恢复时间常数。 交叉设计默认”撤掉处理就回基线”,但行为与生理干预常不成立:有研究专门讨论”不回到基线的持续携带效应”,并给出相应的功效与样本量算法(PMID 41345689)。凶地驻留的交感激活可能延续到下一次吉地驻留开场,把”吉地曲线”的起点污染掉。这类污染还会伪装成别的样子:变化分与协方差分析会给出不同结论(PMID 16921578),极端值入选后的自然回落会被读成”有效”(PMID 27496345)。洗脱期该多长不是”隔一天”,是恢复时间常数乘若干倍——而这常数没人测过。

    第五层:一分钟在不同人身上不等长。 有人 30 秒副交感就回位,有人五分钟还在爬坡。同一人隔四个月重测,皮质醇反应性的稳定性明显弱于儿茶酚胺类(PMID 34225185)。”吉地组平均斜率更陡”这句话里掺了多少个体时钟差异、多少抽样噪声,现在分不开。要分,就得允许每人有自己的时间轴——先估”这个人的响应何时完成一半”,再比半衰期,而不是比固定窗口的斜率。这不是更精细的分析,是把”分钟”从假设变成待估参数。

    还有一处更大的断桥。三个时间尺度至今没接上:实验室尺度是”分钟”(森林驻留前后各测一次,15 分钟步行与 15 分钟静坐,六个测量点,PMID 17435356);单次干预尺度是”小时”(两小时接地测血液流变,PMID 22757749;八周接地睡眠测皮质醇曲线,PMID 15650465);流行病学尺度是”周”(每周累计 120 分钟自然暴露与健康相关,PMID 31197192)。三者用的是三套完全不同的时间单位,中间没有任何换算桥。所以”驻留 20 分钟够不够”这个问题,在现有文献里没有可引用的答案——森林浴的效应量(皮质醇降 12.4%、副交感升 55.0%,PMID 21996763)是”一段驻留”的总账,不是某分钟的速率。

    判决:在谈”吉地 vs 凶地的斜率差”之前,必须先完成三件事。①分指标定时窗:HRV 与皮质醇各自预设窗口(前者不短于 300 秒,后者覆盖峰值潜伏期),不得共用一条横轴,也不得互相代证。②报形状而非只报斜率:用分段模型给出各段斜率、变点位置及其置信区间;变点位置估不出来,本身就是”没有可识别拐点”的结论。③先算最小可检测变化:把该指标的日间变异与重测误差折成 MDC,若吉凶两地的斜率差小于 MDC,则该比较在设计上不成立,应改写为”未检出差异”而不是”无差异”。三件事做完,斜率差才有资格进结论;不做完,任何”吉地起效更快”都是把窗口选择当成了发现。

    易学其实早就把时间当作变量。《素问·八正神明论》讲针刺”必候日月星辰四时八正之气”,强调的是什么时候去看,和看什么一样重要;风水传统里的择日与时辰,也是承认同一个地点在不同时刻表现不同。古典从不假设时间是一根免费的、均匀的横轴。现代研究把它当成了默认背景,问题恰恰出在这里。

    证据与局限

    已验证: HRV 与皮质醇的动力学速度差异明确(HRV 秒级、皮质醇峰值滞后且觉醒脉冲逾 60 分钟,PMID 8255414/30595408);HRV 记录时长与起点显著影响结果(PMID 21496163/33856258/8598068);交叉设计的洗脱假设可被持续携带效应推翻(PMID 41345689);分段/样条模型可刻画非线性响应(PMID 28634437/36085545/42885950);激素标记重测信度不等(PMID 34225185)。

    部分验证: 自然环境驻留确有生理效应(PMID 17435356/21996763/39554703),但均为”驻留前后”总账,无分钟级速率估计;接地效应存在(PMID 28601861/22757749/15650465),”立即改善”等描述未量化起效常数。

    待验证: 吉地与凶地的起效斜率差是否超过该指标 MDC;个体响应半衰期能否作为稳定表型;三个时间尺度(分钟/小时/周)之间的换算关系。

    相关阅读

    证据等级: B
    参考文献: PMID 8255414, PMID 30595408, PMID 21496163, PMID 33856258, PMID 8598068, PMID 12892658, PMID 28634437, PMID 36085545, PMID 42885950, PMID 41345689, PMID 16921578, PMID 27496345, PMID 34225185, PMID 17435356, PMID 21996763, PMID 39554703, PMID 31197192, PMID 22757749, PMID 15650465, PMID 28601861


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    When Does Qi Arrive: Minute-Level Onset Curves of Grounding Effects, Why Can’t the \”Slope Difference\” Between Auspicious and Inauspicious Sites Be Calculated Now

    Evidence level: B

    To compare “how quickly a person relaxes on an auspicious versus inauspicious site,” the premise requires both curves to share a single time axis. Current research has not yet calibrated this axis—different metrics operate at different speeds, start from different baselines, and do not follow linear trajectories; furthermore, a “minute” is not of equal duration across different individuals.

    Main Text

    Translating the concept that “auspicious sites bring tranquility and joy” into data most naturally involves having individuals stand on the site, measure Heart Rate Variability (HRV) and salivary cortisol with a stopwatch, and compare which curve rises faster and falls slower. This idea sounds like it only lacks one experiment, but in reality, it requires five layers of assumptions—because slope is not an intrinsic property of the data; it is constructed through selection.

    Layer 1: The two metrics do not share the same time axis. HRV is a second-level autonomic signal that begins reorganizing within seconds of posture change; cortisol is an endocrine signal that responds slowly. In the classic TSST experiment, subjects undergo a 10-minute preparation period followed by 10 minutes of public speaking and mental arithmetic before salivary cortisol rises to 2–4 times baseline (PMID 8255414), with peaks typically occurring after the stressor ends; the arousal response pulse is more sustained, lasting over sixty minutes (PMID 30595408). Plotting both curves against a shared horizontal axis labeled “minutes of residence” is akin to having sprinters and marathon runners share the same stopwatch: HRV may reach a plateau by minute 5, while cortisol is just beginning at minute 5. Calculating the resulting “slope difference” measures a system speed disparity, not a site-specific difference.

    Layer 2: Where the starting point is selected determines the shape of the slope. Starting the recording too late misses the steepest initial response following posture change; extending it too long conflates “activation” and “recovery” into a single average (PMID 21496163). For the same process, recording from second 0 yields a steep positive slope; starting at minute 1 yields a plateau; starting at minute 6 may even yield a negative slope. This is window boundaries determining the conclusion. HRV has quantification thresholds: heart rate classes require 60 seconds, time-domain analysis requires 240 seconds, and frequency-domain analysis requires 300 seconds before stability (PMID 33856258); the international standard uniformly adopts 5 minutes (PMID 8598068)—”standing for three minutes” fails to meet qualification thresholds even on frequency-domain metrics.

    Layer 3: Curves are not linear. Physiological responses typically follow a “rapid rise—plateau—slow fall” three-phase pattern, or even biphasic patterns; assuming a straight line implies constant velocity. An alternative is segmented modeling: dividing into pre-stress, response, and recovery phases, assigning separate slopes and inflection points for each (PMID 28634437/36085545/42885950). Furthermore, slope and area under the curve (AUC) are not the same thing—two AUC algorithms represent total concentration versus net change, which can have opposite directions (PMID 12892658). Using AUC to claim “lower hormones” and slope to claim “faster recovery” are two distinct statements that cannot serve as mutual proof.

    Layer 4: Washout periods are not calendar days but recovery time constants. Cross-over designs default to “returning to baseline upon removal of treatment,” yet behavioral and physiological interventions often fail this assumption: some studies specifically discuss “persistent carry-over effects that do not return to baseline” and provide corresponding power and sample size algorithms (PMID 41345689). Sympathetic activation induced by inauspicious site residence may persist into the onset of the next auspicious site session, contaminating the starting point of the “auspicious site curve.” Such contamination can masquerade in other forms: analysis of variance and covariance may yield different conclusions (PMID 16921578), and natural reversion after extreme value inclusion may be misread as “effective” (PMID 27496345). The appropriate length of the washout period is not “one day apart,” but a multiple of the recovery time constant—yet this constant has never been measured.

    Layer 5: A minute is not of equal duration across different individuals. Some individuals return to parasympathetic tone within 30 seconds; others are still climbing after five minutes. Re-measuring the same individual four months later shows that cortisol reactivity stability is significantly weaker than catecholamine markers (PMID 34225185). The statement “the average slope in the auspicious group is steeper” contains varying degrees of inter-individual clock differences and sampling noise, which cannot currently be disentangled. To disentangle them, one must allow each person their own time axis—first estimating “when this individual’s response completes half,” then comparing half-lives rather than fixed-window slopes. This is not more refined analysis; it is treating “minutes” as an estimated parameter rather than a fixed assumption.

    There is also a larger disconnect. Three time scales remain unconnected: the laboratory scale operates in “minutes” (pre- and post-forest residence measurements, 15 minutes of walking and 15 minutes of sitting, six measurement points, PMID 17435356); the single-intervention scale operates in “hours” (two hours of grounding measuring blood rheology, PMID 22757749; eight weeks of grounding sleep measuring cortisol curves, PMID 15650465); the epidemiological scale operates in “weeks” (cumulative 120 minutes of natural exposure per week correlated with health outcomes, PMID 31197192). These three utilize entirely different time units with no conversion bridges between them. Therefore, the question “Is 20 minutes of residence sufficient?” has no citable answer in existing literature—the effect size of forest bathing (cortisol reduction 12.4%, parasympathetic increase 55.0%, PMID 21996763) represents the total account of a “segment of residence,” not the rate per minute.

    Verdict: Before discussing the “slope difference between auspicious versus inauspicious sites,” three tasks must be completed. ① Partition metrics by timed windows: HRV and cortisol must have preset windows (the former not shorter than 300 seconds, the latter covering the peak latency period); they must not share a single horizontal axis nor serve as mutual proof. ② Report shape, not just slope: Use segmented models to provide slopes for each segment, inflection point locations, and their confidence intervals; if the inflection point cannot be estimated, this itself constitutes the conclusion of “no identifiable inflection point.” ③ Calculate the Minimum Detectable Change (MDC) first: Convert the daily variability and retest error of the metric into MDC; if the slope difference between auspicious and inauspicious sites is smaller than the MDC, the comparison is invalid by design and should be rewritten as “no detected difference” rather than “no difference.” Once these three tasks are completed, the slope difference qualifies for inclusion in conclusions; without them, any claim of “faster onset at auspicious sites” treats window selection as discovery.

    Yijing Studies has long treated time as a variable. The Su Wen · Eight Correctness Divine Theory discusses acupuncture timing, emphasizing the necessity of awaiting “the four seasons and eight correct qi of sun, moon, stars,” highlighting that when one observes is as important as what one observes; traditional Feng Shui date selection and hourly timing similarly acknowledge that the same location exhibits different performances at different moments. Classical theory never assumes time is a free, uniform horizontal axis. Modern research treats it as a default background; the problem precisely lies here.

    Evidence and Limitations

    Verified: The dynamic speed differences between HRV and cortisol are explicit (HRV is second-level, cortisol peaks with latency and arousal pulses lasting over 60 minutes, PMID 8255414/30595408); HRV recording duration and starting point significantly influence results (PMID 21496163/33856258/8598068); the washout assumption in cross-over designs can be overturned by persistent carry-over effects (PMID 41345689); segmented/spline models can characterize non-linear responses (PMID 28634437/36085545/42885950); hormone markers show unequal retest reliability (PMID 34225185).

    Partially Verified: Natural environment residence does produce physiological effects (PMID 17435356/21996763/39554703), but all are “pre- and post-residence” totals without minute-level rate estimates; grounding effects exist (PMID 28601861/22757749/15650465), but descriptions like “immediate improvement” lack quantified onset constants.

    To be Verified: Whether the onset slope difference between auspicious and inauspicious sites exceeds the metric’s MDC; whether individual response half-lives can serve as stable phenotypes; the conversion relationships between the three time scales (minutes/hours/weeks).

    Related Reading

    Evidence Level: B
    References: PMID 8255414, PMID 30595408, PMID 21496163, PMID 33856258, PMID 8598068, PMID 12892658, PMID 28634437, PMID 36085545, PMID 42885950, PMID 41345689, PMID 16921578, PMID 27496345, PMID 34225185, PMID 17435356, PMID 21996763, PMID 39554703, PMID 31197192, PMID 22757749, PMID 15650465, PMID 28601861


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 吉地「舒适感」的方差分解:景观效应与接地效应各占几成,以及为什么「几成」这个数字不可信

    Evidence level: B

    把”吉地舒服”拆成景观和接地两条通路之后,下一个问题听起来只是个除法:各占几成?但这道除法的分子和分母,都会随你选用哪种分解方法、信度多高、以及把谁当成原因而改变——所以”接地只占一成”这句话,几乎从来不是一个科学结论,而是一个模型设定的副产品。

    正文

    假设你已经做完那个关键实验:2×2 交叉设计——同一批被试轮流进入”景观好+真接地””景观好+假接地””景观差+真接地””景观差+假接地”四种条件,每次驻留测 HRV、唾液皮质醇、主观舒适度。数据到手,第一反应是跑多水平回归:舒适度 = 景观 + 接地 + 景观×接地 + 随机效应,然后看两个系数谁的标准化 β 大。这在统计上没错,在解释上几乎一定翻车。原因有五个。

    第一,”方差分解”不是一种算法,是一族算法,给出的”占比”并不相同。 part R²(把某项拿掉、看解释方差掉了多少)算的是唯一贡献;结构系数换算的 inclusive R² 算的是总贡献(含共享部分);Shapley 值分解、相对权重法、优势分析、LMG 法各有一套分配共享方差的规则。同一份数据,这些方法排序有时一致、数值却可能差出十几个百分点(PMID 42375422)。”景观六成、接地四成”这句话,换个分解法就可能变成”七成对三成”——数据一个字节没动。

    第二,真正的黑洞是”共享方差”,它既不属于景观也不属于接地。 现实中的吉地几乎必然”景观好”和”能接地”同时成立,因为有共同成因:选了一块好地方。所以两者协方差不是噪声,而是主体。拆这一块的工具叫共同性分析(commonality analysis),把总解释方差切成”唯一属于 A””唯一属于 B””A 与 B 共享”三块(PMID 40889234、31891314)。风水语境里共享块最有意思:它是两条通路共同产生、无法归给任何单一通路的效应,也是”气”能被操作化的上界——能被拆开的部分,已经分别是”景观”和”接地”,不再是”气”了。

    第三,测量误差会系统性压低某条通路的占比,而且压向零。 景观测量相对容易:照片评分、绿视率、NDVI、可见天空比例,客观可重复。接地测量则粗糙得多——赤脚还是穿鞋、接地垫通不通、鞋底电阻多大、地面湿度多少,常被压缩成一个二分变量。测量误差在回归里造成回归稀释(regression dilution)偏倚:信度越低,系数被低估越厉害(PMID 37336529)。后果很具体:“景观贡献大于接地”可能完全来自两条通路的测量精度差异,而不是真实的效应差异。 要排除这个可能,必须报告接地变量的重测信度(ICC),并用回归校准(regression calibration)把衰减的系数修正回来。

    第四,最容易被忽略的一层:把”结果”当成并列的”原因”,整个分解就失去意义。 景观和接地为什么共线?因为都是”选了一块好地方”的下游产物。若把接地当成与景观并列的独立预测项,等于假设两者是平行输入,而接地其实可能是选址的结果之一(土质好、植被密、人愿意赤脚)。这点在比较生物学里被明确讨论过:预测项与响应变量的共线性本身不是偏倚主因,模型设定与真实因果结构错配才是(PMID 35784454)。翻译成风水语言:如果接地本身是”好地方”的副产品,问”舒适感里有多少是接地给的”,等于问”发烧里有多少是体温计给的”。

    第五,”几成”这个数字对样本结构的敏感度远超直觉。 多水平模型的 R² 有多个版本(边际、条件、层级专属),估计精度强烈依赖簇的数量:层 2 预测项越多,需要的簇数越多(PMID 41848595、28904005、34113487)。风水研究的现实是地点只有十几个、观测却上千,这种”簇少观测多”的结构会让层 2 方差成分极不稳定——你算出的”接地占比”,可能有一半是抽样噪声。

    那正确做法是什么?

    能用设计分离的,不要用统计分离。 2×2 交叉设计的全部意义就在于把两条通路在设计层面正交化,而不是指望回归去解一个共线方程;同时必须报告携带效应(carry-over)检验与洗脱期设定(PMID 38563531、25115725)。

    报告三个数,而不是一个数:唯一贡献(part R²)+ 总贡献(inclusive R²)+ 共享贡献。三者一起看才有信息量——如果共享块很大,说明两条通路在你的数据里根本分不开,此时任何”谁占几成”的说法都应当被拒绝。

    占比只能用来排序,不能用来分成。 风水实践真正要回答的不是”各占几成”,而是”如果只能改一件事,改哪一件”——这对应的是可干预性与成本,不是方差占比。一块地的景观改不动,接地却可能只需换一双鞋。而且证据强度本身就不对称:接地文献多为小样本先导研究(12 人睡眠皮质醇试点,PMID 15650465)、非随机分组的生理实验(PMID 21469913)或叙述性综述(PMID 31831261、25848315);森林环境则已有系统综述与 Meta 分析支撑(唾液皮质醇 MD = −0.05 μg/dl,PMID 31001682;收缩压 −3.44 mmHg、舒张压 −3.07 mmHg,PMID 36612777)。这意味着在方差分解里,接地那一项的置信区间一定会比景观宽得多。

    最后回到那个最容易被误读的数字上。方差分解给出的从来不是”真相的份额表”,而是”在你这个模型、这个测量精度、这个因果假设下,方差能怎么被切开”。换一个模型设定,切法就变。所以看到”景观占七成、接地占三成”这类说法时,该问的不是”真的吗”,而是:共线程度是多少?信度是多少?谁被当成了原因? 这三个问题答不上来,那个百分比就只是一个好看的空壳。

    证据与局限

    已验证:方差分解存在多种估计量且结果不同(PMID 42375422);多层模型 R² 的估计精度依赖簇数(PMID 41848595);测量误差导致回归稀释偏倚、可用回归校准修正(PMID 37336529);共线性不是偏倚主因、模型因果结构错配才是(PMID 35784454);森林环境的生理效应有系统综述/Meta 分析支撑(PMID 31001682、36612777)。

    部分验证:接地的生理效应——有多项研究与综述报告正向结果,但样本普遍偏小、盲法与随机化描述不足,效应量与证据层级明显低于景观通路(PMID 31831261、21469913、15650465、25848315)。因此”两通路方差各占几成”这一问题,目前没有任何直接针对风水场景的实测数据,本篇给出的是方法学框架与可执行方案,不是实测占比。

    待验证:(1)在 2×2 交叉设计数据上,part R²、inclusive R²、Shapley 分解三者对”景观/接地”占比的估计差异有多大?若排序不一致,则”占比”这一指标在该场景下应被弃用。(2)共同性分析中的”共享块”占比是否稳定大于两条通路的唯一贡献?若是,说明两条通路在本类数据中本质不可分。(3)把接地从二分变量升级为连续剂量变量(皮肤—地电位差 μV 级、接触电阻 kΩ 级)后,接地通路的 part R² 是否显著上升?若上升,则此前的”接地贡献低”应重新解释为测量假象。(4)在簇数 10–20、每簇观测 ≥500 的结构下,层 2 方差成分的 bootstrap 置信区间宽度是多少?宽度超过估计值本身则结论不可用。

    相关阅读

    证据等级: B
    参考文献: PMID 42375422, PMID 40889234, PMID 31891314, PMID 37336529, PMID 35784454, PMID 41848595, PMID 28904005, PMID 34113487, PMID 31001682, PMID 36612777, PMID 38563531, PMID 25115725, PMID 15650465, PMID 21469913, PMID 31831261, PMID 25848315


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    Variance Decomposition of Ji Di’s \”Sense of Comfort\”: The Proportions of Landscape and Grounding Effects, and Why the \”Several Parts\” Figure Is Unreliable

    Evidence level: B

    Decomposing “auspicious comfort” into two distinct pathways—landscape and grounding—makes the next question sound trivial: what proportion does each account for? However, the numerator and denominator of this division shift depending on the decomposition method selected, the reliability of measurements used, and which variable is designated as the cause. Therefore, the statement “grounding accounts for only one-tenth” is rarely a scientific conclusion but rather an artifact of model specification.

    Main Text

    Assume you have completed the critical experiment: a 2×2 crossover design where the same participants rotate through four conditions—”good landscape + real grounding,” “good landscape + false grounding,” “poor landscape + real grounding,” and “poor landscape + false grounding”—measuring HRV, salivary cortisol, and subjective comfort during each stay. Once data are obtained, the first instinct is often to run a multilevel regression: Comfort = Landscape + Grounding + Landscape×Grounding + Random Effects, then compare the standardized β coefficients. While statistically valid, this approach almost certainly fails in interpretation for five reasons.

    First, “variance decomposition” is not a single algorithm but a family of algorithms yielding different proportions. Part $R^2$ (removing a variable to see how much explained variance disappears) calculates unique contribution; structural coefficient conversion for inclusive $R^2$ calculates total contribution (including shared parts); Shapley value decomposition, relative weighting methods, advantage analysis, and LMG methods each follow distinct rules for allocating shared variance. With the same dataset, these methods may rank similarly yet differ by tens of percentage points (PMID 42375422). The claim “landscape accounts for sixty percent, grounding forty percent” could shift to “seventy percent versus thirty percent” simply by changing the decomposition method, without altering a single byte of data.

    Second, the true black hole is “shared variance,” which belongs to neither landscape nor grounding. Auspicious locations almost invariably satisfy both “good landscape” and “effective grounding” simultaneously due to a common cause: selecting a good site. Thus, their covariance is not noise but the subject itself. The tool for this is commonality analysis, which slices total explained variance into three blocks: “unique to A,” “unique to B,” and “shared by A and B” (PMID 40889234, 31891314). In Feng Shui contexts, the shared block is most interesting: it represents an effect jointly produced by two pathways and unassignable to a single pathway, serving as the upper bound for operationalizing “Qi”—the portion that can be decomposed is already categorized as “landscape” or “grounding,” no longer “Qi.”

    Third, measurement error systematically depresses the proportion attributed to a specific pathway, pushing it toward zero. Landscape measurement is relatively straightforward: photo ratings, green view index, NDVI, visible sky proportion; these are objective and repeatable. Grounding measurement is far rougher—barefoot vs. shod, grounding pad continuity, sole resistance, ground humidity—often compressed into a binary variable. Measurement error in regression causes regression dilution bias: lower reliability leads to greater underestimation of coefficients (PMID 37336529). The consequence is concrete: “landscape contribution exceeds grounding” may stem entirely from differences in measurement precision between the two pathways, not from true effect differences. To exclude this possibility, one must report the test-retest reliability (ICC) of the grounding variable and apply regression calibration to correct the attenuated coefficients.

    Fourth, the most overlooked layer: treating the “outcome” as a parallel “cause,” rendering the decomposition meaningless. Why are landscape and grounding collinear? Because they are both downstream products of “selecting a good site.” Treating grounding as an independent predictor parallel to landscape assumes they are parallel inputs, whereas grounding might actually be a result of site selection (good soil, dense vegetation, willingness to go barefoot). This is explicitly discussed in comparative biology: collinearity between predictors and response variables is not the primary source of bias; misalignment between model specification and true causal structure is (PMID 35784454). Translated into Feng Shui language: if grounding itself is a byproduct of a “good place,” asking “how much of comfort comes from grounding” is equivalent to asking “how much of the fever is caused by the thermometer.”

    Fifth, the “proportion” figure is far more sensitive to sample structure than intuition suggests. Multilevel model $R^2$ has multiple versions (marginal, conditional, level-specific), and estimation precision heavily depends on the number of clusters: more level-2 predictors require more clusters (PMID 41848595, 28904005, 34113487). The reality of Feng Shui research is having only a dozen or so locations with thousands of observations; this “few clusters, many observations” structure makes level-2 variance components extremely unstable—the calculated “grounding proportion” might be half sampling noise.

    So what is the correct approach?

    Do not rely on statistical separation if design separation is possible. The entire significance of the 2×2 crossover design lies in orthogonalizing the two pathways at the design level rather than relying on regression to solve a collinear equation; simultaneously, carry-over effect tests and washout period settings must be reported (PMID 38563531, 25115725).

    Report three numbers, not one: Unique contribution (part $R^2$) + Total contribution (inclusive $R^2$) + Shared contribution. Only viewing all three together provides information—if the shared block is large, it indicates that the two pathways are fundamentally inseparable in your data, at which point any claim of “who accounts for what percent” should be rejected.

    Proportions are for ranking, not partitioning. The practical question in Feng Shui is not “what percent does each account for,” but “if only one change can be made, which one should be chosen?” This corresponds to interventability and cost, not variance proportion. A site’s landscape may be unchangeable, yet grounding might only require changing footwear. Furthermore, evidence strength is inherently asymmetric: grounding literature consists mostly of small-sample pilot studies (12-person sleep cortisol trial, PMID 15650465), non-randomized physiological experiments (PMID 21469913), or narrative reviews (PMID 31831261, 25848315); forest environments already have systematic reviews and meta-analyses supporting them (salivary cortisol MD = −0.05 μg/dl, PMID 31001682; systolic BP −3.44 mmHg, diastolic BP −3.07 mmHg, PMID 36612777). This means in variance decomposition, the confidence interval for the grounding term will be far wider than that for landscape.

    Finally, return to the most easily misinterpreted number. Variance decomposition never provides a “table of truth shares”; it shows how variance can be sliced under your specific model, measurement precision, and causal assumptions. Change the model specification, and the slicing changes. Therefore, when encountering statements like “landscape accounts for seventy percent, grounding thirty percent,” do not ask “is this true?” but rather: What is the degree of collinearity? What is the reliability? Who was designated as the cause? If these three questions cannot be answered, that percentage is merely an attractive empty shell.

    Evidence and Limitations

    Verified: Variance decomposition yields multiple estimators with different results (PMID 42375422); multilevel model $R^2$ estimation precision depends on cluster count (PMID 41848595); measurement error causes regression dilution bias correctable via regression calibration (PMID 37336529); collinearity is not the primary bias source, but misalignment of model causal structure is (PMID 35784454); physiological effects in forest environments are supported by systematic reviews/meta-analyses (PMID 31001682, 36612777).

    Partially Verified: Physiological effects of grounding—multiple studies and reviews report positive results, but samples are generally small, blinding and randomization descriptions are insufficient, and effect sizes/evidence levels are clearly lower than the landscape pathway (PMID 31831261, 21469913, 15650465, 25848315). Therefore, the question of “what proportion does each pathway account for” currently has no direct empirical data specifically for the Feng Shui scenario; this paper provides a methodological framework and executable protocol, not an empirical proportion.

    To Be Verified: (1) In 2×2 crossover data, what is the estimation difference between part $R^2$, inclusive $R^2$, and Shapley decomposition regarding “landscape/grounding” proportions? If rankings differ, this metric should be abandoned in this context. (2) Is the proportion of the “shared block” in commonality analysis consistently greater than the unique contributions of both pathways? If so, the two pathways are essentially inseparable in this data type. (3) After upgrading grounding from a binary variable to a continuous dose variable (skin-ground potential difference in μV range, contact resistance in kΩ range), does the part $R^2$ for the grounding pathway increase significantly? If so, prior claims of “low grounding contribution” should be reinterpreted as measurement artifacts. (4) In a structure with 10–20 clusters and ≥500 observations per cluster, what is the bootstrap confidence interval width for level-2 variance components? If the width exceeds the estimate itself, the conclusion is unusable.

    Related Reading

    Evidence Level: B
    References: PMID 42375422, PMID 40889234, PMID 31891314, PMID 37336529, PMID 35784454, PMID 41848595, PMID 28904005, PMID 34113487, PMID 31001682, PMID 36612777, PMID 38563531, PMID 25115725, PMID 15650465, PMID 21469913, PMID 31831261, PMID 25848315


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 控制之后还剩什么:风水评级的信息增量,能不能跨地区复制

    Evidence level: B

    一个变量在控住两个物探变量之后”还有增量”,这句话要成立,得先过四道关:增量指标本身会不会被骗、控制变量是不是躺在评级的下游、跨地区方向不一致算不算失败、以及评级有没有一把跨地区共用的尺子。

    正文

    “信息增量”在统计上不是一个量,而是一族量:判别力增量、再分类改善、综合判别改善、临床净收益。它们对同一份数据可以给出互相矛盾的答案。Kerr 的系统评述指出,再分类改善的类别边界是人为选的,对模型校准高度敏感,而且在新变量完全无用时期望也不为零(PMID 24240655)。Pepe 更早就说明,优势比与曲线下面积之间是非线性关系:一个优势比 3 的强危险因素,加进基线面积已经 0.80 的模型,可能只把面积推到 0.82(PMID 15105181)。反过来,一个优势比只有 1.2 的弱变量,在低基线模型里能推出更大的增量。也就是说,”增量有多大”不说明变量有多强,只说明它与现有模型重叠多少。Steyerberg 的框架把判别、校准、再分类、净收益拆成四件事,明确说不可互相替代(PMID 20010215);决策曲线给的是”在哪个阈值区间净收益变正”(PMID 17099194),它不产出单一数字,也就没那么容易被挑出一个好看的值。

    第二关更要命:把电导率和室内氡放进模型当控制,隐含假设是它们在因果图上位于评级的上游或旁路。可如果”凶地”在传统判据里本就包含潮湿、背阴、通风不畅,那么室内氡与湿度就不是中立的物探代理,它们身上带着评级的成分。此时控制它们等于控制了处理的一部分,Schisterman 把这叫过度校正(PMID 19525685);若控制变量同时受评级与另一个共同原因影响,它就是个对撞点,控制它反而会凭空造出关联(PMID 29040562)。判定办法很朴素:在测氡之前,把”该宅是否符合潮湿与通风类传统判据”单独编码,看它与氡的相关有多强。相关明显,氡就只能进敏感性分析,不能进主模型。

    第三关是”复制”这个词本身。最常见的错误做法是每个地区单独检验、数几个显著。交互检验的功效本来就低,检测一个交互所需的样本量常是检测主效应的数倍(PMID 17578572),六个地区各自做交互检验,有一个显著已属侥幸。更稳的判据是符号:预设交互项方向,只看六个地区同向的有几个。若真实无交互,六地全部同向的概率是 (1/2)⁵,约 3.1%——同样样本下比逐个检验稳得多。但它有前提:方向必须在看数据之前写死,否则事后从六个方向里挑一个,3.1% 会变成 100%。子群分析的系统评估反复印证这一点,试验方案里预设的子群与论文里报告的子群经常对不上(PMID 25030633)。

    第四关最容易被跳过:评级可能根本没有跨地区共用的刻度。盲法评级时,专家在一个地区内同时看二十处宅,他实际做的是地区内的相对排序,不是往一把全国通用的尺子上放。后果有两个:模型里放地区固定效应,评级的组间变异被抽走,只剩组内那一小段;不放,评级就部分充当地区身份的代理,与地质背景共线,”信息增量”测到的其实是地区差异。同时,评级者间一致性本身随边际分布浮动——一致性系数的分母含期望一致率,同样的判断行为在两个吉凶比例不同的地区会算出不同数值(分级基准见 PMID 843571)。所以”两地都是 0.5″并不意味着评级质量相同。

    判决是分三层报,不要合成一个分数。前置门槛:先报评级者间一致性与吉凶比例,跨地区只比”是否落在同一档”,不比数值。主判据:预设唯一主模型,报判别力增量配自助法区间与全流程置换零分布(重跑整个挑选过程),并报决策曲线净收益,不报再分类改善。复制判据:报六地区交互符号一致率(单侧符号检验),加个体资料合并的交互估计与异质性方差(PMID 28057641),样本量按外部验证要求反算(PMID 34031906)。

    《管子·地员》讲”凡草土之道,各有谷造,或高或下,各有草土”——同一套判据,在高处和低处给出不同的具体答案;《葬书》讲”土欲细而坚,润而不泽”,判据是可以现场捏、现场看的物性,不是可以邮寄的分数。古典地理从来没有给过一把跨区域通用的吉凶刻度,”上吉、中吉、下凶”这类分级是后世为教学与交易硬化的。所以六个地区交互项方向不一致,在古人的模型里不是复制失败,而是模型的正常输出——它本来就说”因地而异”。

    证据与局限

    已验证: 室内氡与肺癌的剂量-反应关系在 13 项欧洲病例对照研究的个体资料合并中稳定成立(PMID 15613366);氡的空间分布受地质单元强烈约束,地质类别可作为空间建模的外部漂移项(PMID 17720284),国家级调查也显示室内氡的空间变异具有明确区域结构(PMID 30645746);季节性修正因子会系统性改变同一住宅的氡读数(PMID 30165499)。方法学侧,再分类改善指标的边界依赖与零假设下非零期望、判别力增量与优势比的非线性、交互检验的低功效、过度校正与对撞点偏倚均已确立(PMID 24240655、15105181、17578572、19525685、29040562)。

    部分验证: “风水评级在控制物探变量后是否保留独立信息增量”这一问题本身尚无三变量同步盲法队列的直接答案;跨区域交互方向一致性也没有任何实测数据。本文给出的是判据与设计,不是结论。

    待验证: ① 传统”凶地”判据与室内氡、湿度、通风的耦合强度有多大——这决定氡该进主模型还是只进敏感性分析;② 盲法评级在同一地区内的组内变异是否足以支撑任何增量估计;③ 跨地区交互符号一致率在真实队列中能否超过 3.1% 的零假设线。

    相关阅读

    证据等级: B
    参考文献: PMID 24240655, PMID 15105181, PMID 20010215, PMID 17099194, PMID 19525685, PMID 29040562, PMID 17578572, PMID 25030633, PMID 843571, PMID 28057641, PMID 34031906, PMID 15613366, PMID 17720284, PMID 30645746, PMID 30165499


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    What Remains After Control: Information Increment in Feng Shui Ratings and Cross-Regional Replicability

    Evidence level: B

    For a variable to show “incremental value” after controlling for two geophysical variables, four hurdles must be cleared: whether the incremental metric itself is susceptible to bias, whether control variables lie downstream of the grading rather than upstream, whether inconsistent directions across regions constitute failure, and whether a common scale exists for cross-regional grading.

    Main Text

    “Information gain” is not a single quantity in statistics but a family of metrics: discrimination increment, reclassification improvement, overall discrimination improvement, and net benefit. They can yield contradictory answers for the same dataset. Kerr’s systematic review notes that reclassification improvement relies on arbitrarily chosen class boundaries, is highly sensitive to model calibration, and retains non-zero expectations even when new variables are completely useless (PMID 24240655). Pepe had earlier demonstrated that the relationship between odds ratios and the area under the curve is nonlinear: a strong risk factor with an odds ratio of 3 added to a baseline model with an area already at 0.80 might only push it to 0.82 (PMID 15105181). Conversely, a weak variable with an odds ratio of only 1.2 can yield a larger increment in a low-baseline model. Thus, “how large the increment is” does not indicate variable strength but rather the degree of overlap with the existing model. Steyerberg’s framework decomposes discrimination, calibration, reclassification, and net benefit into four distinct tasks, explicitly stating they are not interchangeable (PMID 20010215); decision curves provide the “threshold range where net benefit becomes positive” (PMID 17099194), do not produce a single number, and are therefore less susceptible to cherry-picking for aesthetic appeal.

    The second hurdle is more critical: including indoor radon and electrical conductivity as controls implies the assumption that they lie upstream or on a parallel path in the causal diagram. However, if “adverse sites” (形煞) traditionally include criteria for dampness, lack of sunlight, and poor ventilation, then indoor radon and humidity are not neutral geophysical proxies but carry the imprint of the grading itself. Controlling for them effectively controls part of the treatment, a phenomenon Schisterman terms overcorrection (PMID 19525685); if a control variable is simultaneously influenced by the grading and another common cause, it acts as a collider, and controlling for it can artificially induce an association (PMID 29040562). The determination method is straightforward: before measuring radon, code separately whether the property conforms to traditional criteria for dampness and ventilation, then assess the strength of correlation with radon. If the correlation is strong, radon should be excluded from the main model and included only in sensitivity analyses.

    The third hurdle concerns the word “replication” itself. A common error is to test each region separately and count significant findings. The power of interaction tests is inherently low; the sample size required to detect an interaction is often several times that needed for a main effect (PMID 17578572). Performing interaction tests across six regions means finding one significant result is merely luck. A more robust criterion is the sign: preset the direction of the interaction term and count how many of the six regions align. If there is no true interaction, the probability of all six regions showing the same direction is $(1/2)^5$, approximately 3.1%—significantly more stable than individual testing under the same sample size. However, this has a prerequisite: the direction must be fixed prior to data inspection; otherwise, selecting one direction from six post-hoc inflates the probability to 100%. Systematic evaluations of subgroup analyses repeatedly confirm this point, where subgroups pre-specified in trial protocols often do not match those reported in the final papers (PMID 25030633).

    The fourth hurdle is most easily overlooked: the grading may lack a common scale across regions. In blinded grading, when experts evaluate twenty properties simultaneously within one region, they perform relative ranking within that region rather than placing them on a single national standard ruler. This leads to two consequences: first, including region fixed effects in the model removes between-group variance in grading, leaving only within-group variation; second, omitting them causes the grading to partially act as a proxy for regional identity, becoming collinear with geological background, meaning the measured “information gain” reflects regional differences rather than universal predictive power. Furthermore, inter-rater reliability fluctuates with marginal distributions—the denominator of the agreement coefficient contains the expected agreement rate, so identical judgment behaviors yield different values in regions with different proportions of good and bad sites (see grading benchmarks in PMID 843571). Therefore, “0.5 in both regions” does not imply equivalent grading quality.

    Judgments should be reported in three layers without synthesizing a single score. Prerequisite thresholds: first report inter-rater reliability and the ratio of good to bad sites; compare across regions only on whether they fall into the same category, not on specific values. Primary criterion: preset a single main model, report discrimination increment with bootstrap confidence intervals and full permutation zero distributions (re-running the entire selection process), and report decision curve net benefit; do not report reclassification improvement. Replication criterion: report the consistency rate of interaction signs across six regions (one-sided sign test), add individual data pooled interaction estimates and heterogeneity variance (PMID 28057641), and calculate sample size requirements based on external validation standards (PMID 34031906).

    The Guanzi · Diyuan states, “All grass and soil paths have their own valleys; some are high, some low, each with its own grass and soil”—the same set of criteria yields different specific answers at different elevations; the Zang Shu* states, “Soil should be fine and firm, moist but not shiny”—criteria are physical properties observable on-site, not scores that can be mailed. Classical geography never provided a cross-regional scale of auspicious or inauspicious ratings; categories like “upper auspicious, middle auspicious, lower inauspicious” were rigidified later for teaching and transaction purposes. Therefore, inconsistent interaction directions across six regions are not failures of replication in ancient models but normal outputs—the model itself states “varies by location.”

    Evidence and Limitations

    Verified: The dose-response relationship between indoor radon and lung cancer holds stably across individual data pooled from 13 European case-control studies (PMID 15613366); the spatial distribution of radon is strongly constrained by geological units, allowing geological categories to serve as external drift terms in spatial modeling (PMID 17720284), and national surveys confirm that indoor radon exhibits clear regional spatial structure (PMID 30645746); seasonal correction factors systematically alter radon readings for the same residence (PMID 30165499). Methodologically, the dependence of reclassification improvement boundaries on arbitrary choices, non-zero expectations under the null hypothesis, the nonlinear relationship between discrimination increment and odds ratios, low power of interaction tests, overcorrection, and collider bias are all established (PMID 24240655, 15105181, 17578572, 19525685, 29040562).

    Partially Verified: There is currently no direct answer from a three-variable synchronous blinded cohort regarding whether “Feng Shui grading retains independent information gain after controlling for geophysical variables”; furthermore, no empirical data exists regarding cross-regional interaction direction consistency. This paper provides criteria and design, not conclusions.

    To Be Verified: ① The coupling intensity between traditional “adverse site” criteria and indoor radon, humidity, and ventilation—this determines whether radon enters the main model or only sensitivity analysis; ② Whether within-group variation in blinded grading within a single region is sufficient to support any increment estimation; ③ Whether cross-regional interaction sign consistency rates can exceed the 3.1% null hypothesis line in real-world cohorts.

    Related Reading

    Evidence Level: B
    References: PMID 24240655, PMID 15105181, PMID 20010215, PMID 17099194, PMID 19525685, PMID 29040562, PMID 17578572, PMID 25030633, PMID 843571, PMID 28057641, PMID 34031906, PMID 15613366, PMID 17720284, PMID 30645746, PMID 30165499


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 断层「150米」能不能预测风水「凶地」:一条连续的地质标尺,怎么去量一张离散的吉凶标签

    Evidence level: B

    150 米这个数字不是风水说的,是从氡研究里借来的。拿一张印着断层线的地图去对一张吉凶标签,首先要问的不是”对不对得上”,而是”这把尺子的刻度,比要量的东西还粗”。

    正文

    一、两个不同物种的变量

    风水评级是离散的、序数的、由人给出的;断层距离是连续的、以米为单位的、由地图或实测给出的。把它们放进同一个检验,第一步就已经做了两次”切”:把距离切成”150 米内/外”,把评级切成”吉/凶”。

    PMID 16217841 说得很直白:把连续预测变量二分,代价是明显的功效损失和残余混杂,而用数据自己找出来的”最优切点”会带来严重偏倚。PMID 16675816 的标题更直接——”二分化连续变量的代价”。两次切叠加,损失不是相加,是相乘。

    二、150 米是怎么”长”出来的

    回到 PMID 26630035 原文:640 处地下室氡实测,用 Haversine 公式算到最近断层的距离,然后用 Kruskal-Wallis 秩和检验逐段扫描,找出断层”开始有显著影响”的距离——150 米。结果是真的:150 米内氡超标率(>200 Bq/m³)22.8%,150-700 米降到 15.2%;只看富铀基岩上的点,是 30.7% 对 17.5%。方向是真的,梯度也是真的。

    但”150″这个数字本身,是扫描出来的一个最优分箱边界,不是物理常数。换一批房子、换一种基岩筛选、换一个分箱粒度,它会移动。把它当成一个可以拿去验证的”阈值”,是把估计值当成了真值——这正是 PMID 16217841 警告的那类偏倚的产生机制。

    三、断层线在图上本来就有几十米误差

    这是本篇最要紧的一句。1:50000 地质图上的断层线,位置精度通常在图上 0.5-1 毫米量级,折算到地面就是 25-50 米;再算上断层带本身有宽度、倾角不明、隐伏段只能推断,实际不确定度更大。

    于是”150 米阈值”这个数字,和”这条线到底画在哪儿”的误差,是同一个量级。阈值的效应宽度,可能还小于测量噪声的宽度。 这不是说效应不存在,是说用这个精度去分辨 150 米内外的差别,尺子先不够用。PMID 16286689 讲的正是同一件事:氡潜势在地质单元之间和之内都有变异,只按单元贴一个标签,会漏掉单元内部的大半信息。

    四、真要验断点,工具不是”分两组比均值”

    断点存在性有专属工具:断点回归(RDD)与分段回归。PMID 25579639 系统综述了 RDD 在医学与公共卫生中的用法,指出这类设计长期被低估,并列出五项必须报告的内容,包括有效性条件检验与稳健性检查;PMID 25061922、27547695、41190057 是同一件事的不同版本教程。

    RDD 最怕的前提在这里反而是满足的:一个宅子无法把自己”安排”到 149 米或 151 米,运行变量不可操纵。但 RDD 对断点附近的样本密度极度敏感——如果 140 到 160 米之间只有三五个宅点,断点真假你都测不出来。所以正确的产出不是一个 p 值,而是断点位置的置信区间,以及”这个断点是否可识别”的判断。如果剖面似然曲线是平的,结论就是”没有证据支持存在阈值”——这本身是一个结果,不是失败。

    五、kappa 会被”凶地太少见”毁掉

    一致性系数也埋着坑。PMID 8501467 早已把 kappa 拆成偏倚、患病率、观察一致三部分,明确建议:只报 kappa 会误导。PMID 29238424 与 38283651 展示的是同一个悖论——当某一类极度常见(比如九成宅点都是”吉”),两位评级者几乎全说”吉”时,观察一致率可以很高,而 kappa 趋近于零。

    所以”kappa=0.3,说明风水评级不可靠”是错误推论,它可能只说明”凶地太稀有”。PMID 33145661 提出的信息一致性指标,就是把患病率的影响从一致性核心里剥离出来的一种做法。要报,就同时报偏倚指数、患病率指数、观察一致率,三者一起看。

    六、那到底该怎么做

    同一批宅点:①断层距离用同一版地质图加野外校核,把地图精度作为显式误差项写进模型;②风水评级不二分,用 5 级序数,盲法进行(评级者不知物探与氡结果);③主分析用分段回归估计断点位置与置信区间,RDD 做稳健性检查;④一致性同时报加权 kappa、偏倚/患病率分解与信息一致性;⑤把”150 米”作为待估参数放进模型,而不是已知条件。

    一句话:这条线真正能回答的不是”150 米能不能预测凶地”,而是”断层距离与吉凶评级之间,到底有没有一个数据分辨得出的转折点”。前者是先射箭再画靶,后者才有被证伪的可能。

    证据与局限

    已验证: 断层距离与室内氡水平的存在梯度关系,150 米附近为数据驱动的显著边界(PMID 26630035);连续预测变量二分会导致功效损失与残余混杂,数据驱动切点引入严重偏倚(PMID 16217841、16675816);kappa 受患病率与偏倚影响,存在高一致低 kappa 的悖论(PMID 8501467、29238424、38283651);RDD 用于因果推断的有效条件与报告规范(PMID 25579639、25061922)。

    部分验证: 断层带、富铀基岩与室内氡升高的空间关联在多个地质背景中复现(PMID 40819484、42322771、41012418),但各研究的”影响半径”不一致,提示阈值本身随地质背景变化;氡潜势在地质单元内部亦存在大幅变异(PMID 16286689、33992858、26042833)。

    待验证: 断层距离与风水”吉/凶”评级之间是否存在可识别的断点,及其位置置信区间;1:50000 地质图断层线位置误差量级是否已接近或超过阈值效应宽度,导致阈值不可识别;风水评级的一致性在控制患病率后是否仍低于可接受水平。

    相关阅读

    证据等级: B
    参考文献: PMID 26630035, PMID 16217841, PMID 16675816, PMID 21062456, PMID 8501467, PMID 29238424, PMID 38283651, PMID 33145661, PMID 25579639, PMID 25061922, PMID 27547695, PMID 41190057, PMID 16286689, PMID 33992858, PMID 26042833, PMID 40819484, PMID 42322771, PMID 41012418


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    Can a 150-meter fault predict Feng Shui \”unlucky sites\”: How to measure discrete labels of good or bad fortune with a continuous geological scale

    Evidence level: B

    The figure “150 meters” is not derived from Feng Shui; it is borrowed from radon research. When overlaying a map marked with fault lines against a label of auspiciousness or inauspiciousness, the first question should not be “does it match,” but rather “is the scale of this ruler coarser than the object being measured.”

    Main Text

    I. Variables from Two Different Species

    Feng Shui ratings are discrete, ordinal, and human-assigned; fault distances are continuous, measured in meters, and derived from maps or field measurements. Placing them into the same test immediately involves two “cuts”: slicing distance into “within/without 150 meters” and slicing ratings into “auspicious/in Auspicious.”

    PMID 16217841 states plainly: dichotomizing a continuous predictor incurs obvious loss of power and residual confounding, while using data-driven “optimal cut-points” introduces severe bias. The title of PMID 16675816 is even more direct—”The Cost of Dichotomizing a Continuous Variable.” With two cuts stacked, the loss is not additive but multiplicative.

    II. How the “150 Meters” Was “Grown”

    Returning to the original text of PMID 26630035: 640 basement radon measurements were taken, distances to the nearest fault were calculated using the Haversine formula, and a Kruskal-Wallis rank-sum test scanned segment by segment to find the distance where faults “begin to have a significant effect”—150 meters. The result is real: the rate of radon exceedance (>200 Bq/m³) within 150 meters is 22.8%, dropping to 15.2% between 150–700 meters; looking only at points on uranium-rich bedrock, it is 30.7% versus 17.5%. The direction is real; the gradient is real.

    However, the number “150” itself is an optimal binning boundary scanned out of the data, not a physical constant. Changing a batch of houses, changing a bedrock filter, or changing the binning granularity would shift it. Treating it as a verifiable “threshold” treats an estimate as truth—precisely the mechanism of bias warned against in PMID 16217841.

    III. Fault Lines on Maps Already Carry Tens of Meters of Error

    This is the most critical sentence in this piece. On 1:50,000 geological maps, fault line positions typically have a precision on the order of 0.5–1 millimeters on the map, translating to 25–50 meters on the ground; adding the width of the fault zone, unknown dip angles, and inferred hidden segments increases the actual uncertainty further.

    Thus, the “150-meter threshold” number and the error in “where exactly this line is drawn” are of the same magnitude. The width of the threshold effect may be smaller than the width of the measurement noise. This does not mean the effect is absent; it means using this precision to distinguish differences within versus outside 150 meters renders the ruler insufficient. PMID 16286689 addresses exactly this: radon potential varies both between and within geological units; assigning a single label to a unit misses a large portion of the information within that unit.

    IV. If Verifying the Discontinuity, the Tool Is Not “Comparing Means in Two Groups”

    There are specialized tools for discontinuity existence: Regression Discontinuity Design (RDD) and Piecewise Regression. PMID 25579639 provides a systematic review of RDD usage in medicine and public health, noting that such designs have been long underestimated and listing five required reporting items, including validity condition tests and robustness checks; PMID 25061922, 27547695, and 41190057 are different versions of tutorials on the same topic.

    The premise that RDD fears most is actually satisfied here: a house cannot “arrange” itself to be 149 meters or 151 meters away; the running variable is non-manipulable. However, RDD is extremely sensitive to sample density near the discontinuity—if there are only three or four houses between 140 and 160 meters, the truth of the discontinuity cannot be measured. Therefore, the correct output is not a p-value, but a confidence interval for the discontinuity position and a judgment on whether “this discontinuity is identifiable.” If the profile likelihood curve is flat, the conclusion is “no evidence supports the existence of a threshold”—which itself is a result, not a failure.

    V. Kappa Can Be Destroyed by “Too Few Inauspicious Sites”

    The consistency coefficient also hides pitfalls. PMID 8501467 has long decomposed kappa into bias, prevalence, and observed agreement, explicitly advising: reporting only kappa misleads. PMID 29238424 and 38283651 demonstrate the same paradox—when one category is extremely common (e.g., 90% of sites are “auspicious”), two raters may almost always say “auspicious,” resulting in high observed agreement while kappa approaches zero.

    Therefore, the inference “kappa=0.3 indicates Feng Shui ratings are unreliable” is a false deduction; it may only indicate “inauspicious sites are too rare.” PMID 33145661 proposes an information consistency index, which is a method of stripping the prevalence influence from the core of consistency. If reporting, simultaneously report bias index, prevalence index, and observed agreement; view all three together.

    VI. So What Should Actually Be Done

    For the same batch of sites: ① Use the same geological map with field verification for fault distance, explicitly writing map precision as an error term in the model; ② Do not dichotomize Feng Shui ratings; use a 5-level ordinal scale, conducted blindly (raters unaware of geophysical and radon results); ③ Use piecewise regression for the main analysis to estimate discontinuity position and confidence intervals, using RDD for robustness checks; ④ Report weighted kappa, bias/prevalence decomposition, and information consistency simultaneously for consistency; ⑤ Treat “150 meters” as a parameter to be estimated in the model, not a known condition.

    In one sentence: The line truly answers not “Can 150 meters predict inauspicious sites,” but “Is there a data-resolvable turning point between fault distance and Feng Shui ratings.” The former is shooting the arrow before drawing the target; only the latter has the possibility of being falsified.

    Evidence and Limitations

    Verified: Existence of a gradient relationship between fault distance and indoor radon levels, with a data-driven significant boundary near 150 meters (PMID 26630035); dichotomizing continuous predictors leads to loss of power and residual confounding, while data-driven cut-points introduce severe bias (PMID 16217841, 16675816); kappa is affected by prevalence and bias, exhibiting a paradox of high agreement with low kappa (PMID 8501467, 29238424, 38283651); RDD validity conditions and reporting standards for causal inference (PMID 25579639, 25061922).

    Partially Verified: Spatial associations between fault zones, uranium-rich bedrock, and elevated indoor radon are replicated across multiple geological backgrounds (PMID 40819484, 42322771, 41012418), but inconsistent “influence radii” across studies suggest the threshold itself varies with geological context; radon potential also exhibits substantial variation within geological units (PMID 16286689, 33992858, 26042833).

    To Be Verified: Whether an identifiable discontinuity exists between fault distance and Feng Shui “auspicious/in Auspicious” ratings, and its position confidence interval; whether the magnitude of error in fault line positions on 1:50,000 geological maps has approached or exceeded the width of the threshold effect, rendering the threshold unidentifiable; whether Feng Shui rating consistency remains below an acceptable level after controlling for prevalence.

    Related Reading

    Evidence Level: B
    References: PMID 26630035, PMID 16217841, PMID 16675816, PMID 21062456, PMID 8501467, PMID 29238424, PMID 38283651, PMID 33145661, PMID 25579639, PMID 25061922, PMID 27547695, PMID 41190057, PMID 16286689, PMID 33992858, PMID 26042833, PMID 40819484, PMID 42322771, PMID 41012418


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 一条链子,两个病:得气→网络→疗效的中介,能不能跨病种比

    Evidence level: B

    骨关节炎说”没关”,痛经说”有关”。这不是两个病给出了两个答案,而是两个研究根本不是同一个问题。

    正文

    得气强度→脑网络调制→临床疗效,这条三段链是针刺机制研究里最漂亮的一个假设:刺激产生感觉,感觉改变网络,网络改变疼痛。它已被源文立为待验证的完整中介模型。现在有人拿两个病种来对照:骨关节炎(OA)里得气与镇痛无关(p=0.49),痛经里得气与镇痛相关。于是很容易得出一个结论——中介强度有疾病特异性,痛经这条链通,OA 不通。

    这个结论下得太快。两篇放在一起先要问”可比吗”,答案几乎是否定的,理由四层,一层比一层基本。

    第一层:阳性/阴性是两套对照强度的产物。 2018 年 Vickers 的个体患者数据元分析(29 试验、20 827 人)给出过一句很硬的话:试验之间效应量的差异,主要由对照组接受的治疗决定,而不是由针刺本身的特征决定(PMID 29198932)。这直接决定了 OA 那个”阴性”的分量。Hinman 2014 的 JAMA 试验里,毫针相对假针 12 周时疼痛只差 −0.4 单位(95%CI −1.2~0.4),但相对不治疗对照差 −1.1(−1.8~−0.4)(PMID 25268438)——针不是没用,是它赢不了那个假针。而痛经那篇的两组是”深刺粗针加手法”对”浅刺细针不手法”(PMID 31223331),后者是天然偏弱的对照,组间差必然拉大。一个是针打不过强假针,一个是拿自己设计的弱对照比。这两个标签比的不是中介强弱,是假针对照做得像不像。

    第二层:两个”疗效”不在同一个时间窗。 OA 试验的结局是四周疗程结束时的疼痛变化;痛经那篇测的是起针当时、起针后 10、20、30 分钟(PMID 31223331)。前者是累积疗程效应,后者是即时镇痛。即时镇痛和疗程效应背后大概率不是同一条通路——前者更依赖内源性镇痛与注意调制,后者还要经过组织修复、习惯化与期望重塑。用一个即刻的量去当四周结局的中介,或者反过来,都不是”中介强弱不同”,是链条长度不同。

    第三层:得气不是一个标量,符号甚至会在同一研究内部翻转。 痛经那篇最要命的不是”相关”,而是相关的方向在两组里相反:得气组里得气强度与镇痛负相关,非得气组里正相关(PMID 31223331)。同一份数据、同一个量表、同一个穴位,方向翻了。原因写得也明白——得气含两簇方向相反的东西:酸痛、胀满随强度上升而疗效下降,电窜、放散则相反。文章最终结论是”中度得气比强得气更能延长镇痛“,也就是一条倒 U 曲线。而 OA 那篇用的是 Park 问卷总分加”有没有得气”的二分类(PMID 20964256),用一维总分去平均一个多维、且方向互抵的构念,无论真实关系是正是负,都会被压成”p=0.49″。拿一个倒 U 曲线的左段和一个被压平的总分去比”中介强度”,比的其实是两把不同的尺子。

    第四层:中介效应量本身没有跨研究的共同单位。 自然间接效应的点识别依赖 cross-world independence,这个假设不可检验(PMID 33512846);不做该假设时只能拿到部分识别区间(PMID 38962169)。加上效应量的非可压缩性——在 OR 尺度上调整协变量引起的效应变化并非中介(PMID 33314251)——”中介比例 30%”这类数字一旦换了结局量表就不可移植。更实际的问题是测量误差:单次得气评分是瞬时量,其重测信度若像多数症状量表那样在 0.5-0.7,间接效应会按信度被系统性稀释(回归稀释,PMID 37336529),而两个病种用的量表不同,稀释幅度就不同——连”谁的中介更强”的排序都不可靠,更不用说比较其数值。

    所以本篇的判决是:“跨病种中介强度比较”这个设计,在现有研究之间做不成,只在一个统一样本的试验内部做得成。 可行的做法是把疾病从”两个研究”改成”一个研究里的一个调节变量”:一份方案、一个得气量表(且必须拆成不适簇/放散簇/非痛感簇分别进模型,不合并总分)、一个脑网络指标、一个固定时间窗,按病种分层随机,检验的是”病种×得气”的交互项——而不是拿两篇论文的中介比例相减。中介报告必须同时给出部分识别区间而非点估计,并声明结局尺度。

    易学落点

    《灵枢·终始》讲针刺的次序:”一刺则阳邪出,再刺则阴邪出,三刺则谷气至,谷气至而止。”最后四个字值得停一下——“至而止”。古人没有说”谷气越多越好”,说的是到了就停。痛经那篇的倒 U 曲线,恰好在数据层面重复了这四个字:强得气并不优于中度得气。

    更要紧的是,古典文本里的”得气”从来不是一把连续刻度尺。它至少是三种不同性质的变量:《灵枢·九针十二原》”刺之要,气至而有效”——是有/无的二值事件;《灵枢·终始》”谷气至而止”——是何时该停的时点判断;《灵枢·行针》分四类人”或神动而气先针行,或气与针相逢,或针已出气独行,或数刺乃知”——是时序类型的分类变量(该篇已论)。这三种里,一种都不是”强度 0-10 分”。把得气压缩成一个可以进回归的连续量,是现代研究方法的需求,不是古典概念的结构。 于是当数据给出矛盾结果时,我们其实是在追问一个被自己改造过的变量。

    证据与局限

    已验证:Vickers 2018 的对照驱动结论、Hinman 2014 的针不优于假针、White 2010 的得气与 OA 镇痛无关、Hu 2019 的组内符号翻转与倒 U 结论,均有原文支撑。

    部分验证:得气的多维成分结构在痛经中已被观察到,但在其他病种是否同构未知;测量误差对间接效应的稀释幅度尚无针刺领域的实测值。

    待验证:跨病种中介的调节效应(”病种×得气”交互)从未在同一方案内被检验;得气量表各成分簇的重测信度与最小可检出差值均未标定。

    相关阅读

    证据等级: B
    参考文献: PMID 29198932, PMID 25268438, PMID 20964256, PMID 31223331, PMID 33512846, PMID 38962169, PMID 33314251, PMID 37336529


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    One Chain, Two Diseases: The Mediating Role of Qi Reception, Network, and Efficacy—Can We Compare Across Disease Categories?

    Evidence level: B

    Osteoarthritis (OA) says “no link”; dysmenorrhea says “yes link.” These are not two diseases giving two answers; the two studies address fundamentally different questions.

    Main Text

    The three-stage chain—Qi reception intensity → brain network modulation → clinical efficacy—is arguably the most elegant hypothesis in acupuncture mechanism research: stimulation generates sensation, sensation alters networks, and networks alter pain. It has been established in the original literature as a complete mediator model awaiting verification. Now, two disease categories are used for comparison: Qi reception shows no association with analgesia in OA ($p=0.49$), but a significant association in dysmenorrhea. Consequently, it is easy to conclude that mediator strength exhibits disease specificity—the chain is open in dysmenorrhea but closed in OA.

    This conclusion is drawn too hastily. Before juxtaposing the two studies, one must ask “Are they comparable?” The answer is almost certainly no, for four reasons, each more fundamental than the last.

    First Layer: Positive/Negative findings are products of two different control intensities. Vickers’ 2018 individual patient data meta-analysis (29 trials, 20,827 participants) made a definitive statement: between-trial effect size differences are primarily determined by the control treatment received, not by characteristics of acupuncture itself (PMID 29198932). This directly determines the “negative” component in the OA study. In Hinman’s 2014 JAMA trial, at 12 weeks, pain scores differed by only −0.4 units between fine needles and sham needles (95%CI −1.2~0.4), but by −1.1 units compared to no-treatment controls (−1.8~−0.4) (PMID 25268438)—the needle is not useless; it simply loses against the sham. Conversely, in the dysmenorrhea study, the comparison was “deep insertion with coarse needles and manipulation” versus “shallow insertion with fine needles without manipulation” (PMID 31223331), where the latter represents a naturally weaker control, inevitably widening the group difference. One study compares a needle against a strong sham; the other compares against a weak, self-designed control. These labels do not compare mediator strength; they compare the quality of the control group.

    Second Layer: The two “efficacy” outcomes operate outside different time windows. The OA trial measured pain change at the end of a four-week course; the dysmenorrhea study measured effects immediately upon needle removal, and at 10, 20, and 30 minutes post-removal (PMID 31223331). The former reflects cumulative course effects, while the latter reflects immediate analgesia. Immediate analgesia and course effects likely rely on different pathways—the former depends more on endogenous analgesia and attentional modulation, while the latter involves tissue repair, habituation, and expectation reshaping. Using an immediate measure as a mediator for a four-week outcome, or vice versa, does not indicate “different mediator strengths”; it indicates different chain lengths.

    Third Layer: Qi reception is not a scalar; even the sign can flip within the same study. The most critical issue in the dysmenorrhea study is not “association,” but the direction of association reverses between groups: in the Qi-reception group, intensity correlates negatively with analgesia, while in the non-Qi-reception group, it correlates positively (PMID 31223331). The same data, same scale, same acupoint—yet the direction flips. The authors explain this clearly: Qi reception contains two clusters of opposing effects; efficacy decreases with rising intensity for soreness/distension, but increases for electric dispersal/spreading. The study concludes that “moderate Qi reception extends analgesia better than strong Qi reception,” implying an inverted U-curve. In contrast, the OA study used the total score of the Park questionnaire plus a binary classification of “presence/absence of Qi” (PMID 20964256). Using a one-dimensional total score to average a multi-dimensional construct with opposing directional components compresses both positive and negative relationships into “p=0.49.” Comparing the left segment of an inverted U-curve against a flattened total score is not comparing “mediator strength”; it is comparing two different rulers.

    Fourth Layer: Mediator effect sizes lack a common unit across studies. Natural indirect effects rely on cross-world independence, an assumption that is untestable (PMID 33512846); without it, only partial identification intervals can be obtained (PMID 38962169). Furthermore, effect sizes are non-compressible: changes in effects caused by covariate adjustment on an odds ratio scale are not mediated effects (PMID 33314251). Consequently, figures like “mediator proportion of 30%” are not portable when the outcome scale changes. A more practical issue is measurement error: single-timepoint Qi scoring is an instantaneous measure. If its test-retest reliability is around 0.5–0.7, similar to many symptom scales, the indirect effect will be systematically diluted by regression dilution (PMID 37336529). Since different scales are used in the two disease categories, the dilution magnitude differs—even the ranking of “whose mediator is stronger” is unreliable, let alone comparing their numerical values.

    Therefore, the verdict for this paper is: Comparing mediator strength across disease categories cannot be performed between existing studies; it can only be done within a single study’s trial. A feasible approach is to reframe the disease from “two studies” to “a moderator variable within one study”: a single protocol, a single Qi scale (which must be disaggregated into discomfort clusters/dispersal clusters/non-painful sensation clusters into separate models, not merged into a total score), a single brain network metric, and a fixed time window, with randomization stratified by disease category. This tests the “disease × Qi” interaction term—rather than subtracting mediator proportions from two papers. Mediator reports must provide partial identification intervals rather than point estimates and explicitly declare the outcome scale.

    Yijing Application Points

    The Ling Shu · Zhong Shi discusses the sequence of acupuncture: “One insertion drives out Yang evil; two insertions drive out Yin evil; three insertions bring the valley Qi. Stop when valley Qi arrives.” The final four characters, “arrive and stop,” deserve pause. Ancient texts did not say “more valley Qi is better”; they stated to stop upon arrival. The inverted U-curve in the dysmenorrhea study replicates these four characters at the data level: strong Qi reception is not superior to moderate Qi reception.

    More importantly, “Qi reception” in classical texts is never a continuous ruler. It comprises at least three distinct variable types: Ling Shu · Jiu Zhen Shi Er Yuan‘s “The key to needling is when Qi arrives, efficacy follows”—a presence/absence binary event; Ling Shu · Zhong Shi‘s “Stop when valley Qi arrives”—a timing judgment of when to cease; and Ling Shu · Xing Zhen‘s classification of four types of people (“or spirit moves before needle travel, or Qi meets needle, or needle exits but Qi travels alone, or multiple insertions reveal it”)—a temporal type categorical variable (discussed previously). None of these are “intensity scores from 0 to 10.” Compressing Qi reception into a continuous variable suitable for regression is a demand of modern research methodology, not the structural nature of the classical concept. Thus, when data yields contradictory results, we are actually interrogating a variable that has been transformed by ourselves.

    Evidence and Limitations

    Verified: Vickers’ 2018 conclusion driven by control design, Hinman’s 2014 finding that needles are not superior to sham needles, White’s 2010 finding of no association between Qi and OA analgesia, and Hu’s 2019 finding of intra-group sign reversal and inverted U-conclusion are all supported by original texts.

    Partially Verified: The multi-dimensional component structure of Qi reception has been observed in dysmenorrhea but remains unknown whether it is isomorphic in other conditions; no empirical values exist in the acupuncture field regarding the dilution magnitude of measurement error on indirect effects.

    To Be Verified: The moderating effect of cross-disease mediation (“disease × Qi” interaction) has never been tested within a single protocol; test-retest reliability and minimum detectable differences for each component cluster of the Qi scale remain uncalibrated.

    Related Reading

    Evidence Level: B
    References: PMID 29198932, PMID 25268438, PMID 20964256, PMID 31223331, PMID 33512846, PMID 38962169, PMID 33314251, PMID 37336529


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.