Evidence level: A
两千年前的周期预测模型遇上百年一遇的变暖趋势——哪些验证成功,哪些正在失效
正文(1000-1500字)
五运六气理论出自《黄帝内经》,用天干地支推演 5 年、6 年、30 年、60 年等多重气候周期,并据此预测”某年某气主事,易发某病”。它本质上是一台基于周期叠加的”气候-疾病预测机”。但现代气候科学给出一个棘手的事实:过去 100 多年全球平均气温上升超过 1℃,极端天气的频率和强度都在改变。当一个建立在稳定周期假设上的预测模型,遇到百年一遇的趋势性漂移,它的准确性还剩下多少?这不是理论之争,是可以拿气象站数据直接检验的问题。
先看”周期假设”本身的验证
要回答变暖背景下的预测准确性,第一步是检验运气理论的周期是否真实存在于气象数据里。北京中医药大学团队用复 Morlet 小波分析了北京 1948—2007 年 60 年的五项气象要素(气温、降水、湿度、水汽压、风速),结果好坏参半:年平均值上找不到完整的 5 年或 10 年周期;但晚冬(终之气末段)的气温、水汽压和湿度确实呈现 5 年周期,与”五运”周期一致;早夏(三之气)存在 3 年和 9 年周期,却不匹配六气理论的 6 年周期。也就是说,周期假设在部分时段、部分要素上成立,但远不是全局性的——这为”变暖环境下周期会不会漂移”埋下了伏笔。
运气-疾病预测模型的实际准确率
更具说服力的检验来自”医学气象预测模型”这一系列研究。北京地区 1990—2004 年肺结核发病率与当年及前 1—3 年的气象因子(温度、风速、降水、相对湿度等)显著相关,其中风速贡献最强(标准化显著性 100%),用反向传播神经网络建立的预测模型全部成立,且”含前三年气象数据”的模型优于单年数据模型——这正好呼应运气理论”三年化疫”的滞后思想(PMID 26525546)。
同样的思路复制到其他传染病:风疹发病率与当年及前三年气象变量密切相关,融合三年气象数据的预测模型准确率比单年模型提升 76%(PMID 30467697);细菌性痢疾与降水、湿度、水汽压、温度显著正相关、与风速负相关,预测模型准确率较高(PMID 22903445);猩红热发病周期的小波分析发现 5 年主周期与五运周期吻合、28 年次周期与六气”准周期、多周期叠加”的描述一致(PMID 23713246)。
四组独立研究指向同一个结论:运气理论框架下的气象-疾病关联是真实存在的,滞后期和周期结构基本能被数据复现。这是它作为预测模型最硬的底牌。
变暖背景下,问题出在哪
但”相关性成立”不等于”预测准确”。把这套模型放到气候变暖的语境下,至少有三个短板:
- 趋势项没建模。运气理论是纯周期模型,没有趋势项。全球变暖带来的平均态抬升(比如冬季平均气温升高)不属于任何干支周期,模型天然无法刻画。当极端高温、暴雨这些”破纪录”事件增多时,周期预测最容易翻车。
- 地域差异大。运气理论诞生于黄河流域,无锡 60 年气象数据与运气推演的符合率约 65%—91%(天干符合率 80%、司天在泉 83%—92%、六主气 100%),但广州研究显示岁运预测符合率可达 90%、司天之气预测上半年气象仅 25% 符合。同一套干支在不同气候带效果天差地别。
- RCT 层面证据薄弱。系统综述显示,运气指导治疗的 13 项 RCT 方法学质量普遍偏低,无法得出确定性结论——预测的”相关性”尚未转化为干预的”有效性”。
如何让这套模型在变暖时代继续有用
承认局限不等于放弃这套框架。把五运六气重新定位成”带趋势校正的周期模型”,它依然有价值:第一,用分段小波分析(把变暖前、变暖后两个时段分开做周期检测)可以量化周期是否漂移、漂移多少,这是直接可做的实证;第二,把干支周期当作宏观背景变量,与现代流行病学的季节 ARIMA、机器学习模型叠加,而不是互斥竞争——运气理论提供的 3 年滞后结构,恰恰是现代模型常忽略的慢变量;第三,极端气候事件库(高温、暴雨、寒潮)与干支”太过不及”年份做匹配检验,能圈出模型最擅长和最不擅长的气候类型。这样,五运六气就从”预测天气”降级为”刻画气候背景的慢周期”,准确性边界反而更清晰。
证据与局限
已验证:
– 北京 15 年肺结核、风疹、痢疾回顾研究均证实气象因子-发病率显著相关,”三年化疫”滞后结构可复现(PMID 26525546, 30467697, 22903445)
– 猩红热 5 年主周期与五运周期吻合(PMID 23713246)
– 运气推演与实际气象的符合率在部分地域(无锡)达 65%—91%
部分验证:
– 周期假设仅部分成立:晚冬 5 年周期与五运一致,早夏周期与六气 6 年不符
– 地域符合率差异悬殊(广州司天之气预测仅 25%),北方优于南方
– “相关性”成立但”预测精度”未达到现代流行病学模型的标准
待验证:
– 气候变暖导致的周期漂移:运气周期在长期趋势叠加下是否发生相位/频率偏移
– 极端气候事件(高温、暴雨、寒潮)与干支”太过不及”年份的符合度
– 运气预测与现代统计模型(如季节性 ARIMA、机器学习)在同期数据上的头对头比较
相关阅读
证据等级: A(有PubMed直接支撑)
参考文献: PMID 26525546, PMID 30467697, PMID 22903445, PMID 23713246
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。
Predictive Accuracy of Five Elements and Six Qi in the Context of Modern Climate Warming
Evidence level: A
Two-thousand-year-old cyclical prediction models meet century-scale warming trends—what validates and what is failing
Main Text (1000-1500 words)
The theory of Wu Yun Liu Qi (Five Movements and Six Gases), originating from the Huangdi Neijing, utilizes the Heavenly Stems and Earthly Branches to derive multiple climatic cycles ranging from 5 years to 60 years, subsequently predicting that “a specific Qi governs a certain year, predisposing to specific diseases.” At its core, it functions as a “climate-disease prediction machine” based on the superposition of cycles. However, modern climatic science presents a challenging fact: over the past century, global average temperatures have risen by more than 1°C, altering both the frequency and intensity of extreme weather events. When a prediction model built on stable cyclical assumptions encounters a century-scale trend-driven drift, how much of its accuracy remains? This is not merely a theoretical debate; it is a question that can be directly tested using meteorological station data.
Validating the “Cyclical Assumption” Itself
To address predictive accuracy under warming conditions, the first step is to verify whether the cyclical patterns inherent in the theory genuinely exist within meteorological data. A team from Beijing University of Traditional Chinese Medicine employed complex Morlet wavelet analysis on 60 years of meteorological data from Beijing (1948–2007) across five variables (temperature, precipitation, humidity, vapor pressure, and wind speed). The results were mixed: no complete 5-year or 10-year cycles were found in annual averages; however, temperature, vapor pressure, and humidity in late winter (the final segment of the Terminal Qi) did exhibit a 5-year cycle consistent with the “Five Movements.” Conversely, early summer (Third Qi) displayed 3-year and 9-year cycles that did not align with the Six Gases’ theoretical 6-year cycle. Thus, while the cyclical assumption holds true in specific timeframes and variables, it is not universal—planting the seed for whether cycles might drift under warming environments.
Actual Accuracy of the Qi-Disease Prediction Model
More compelling validation comes from a series of studies on “medical meteorological prediction models.” In Beijing, between 1990 and 2004, tuberculosis incidence rates showed significant correlations with meteorological factors from the current year and the preceding 1–3 years (temperature, wind speed, precipitation, relative humidity, etc.), with wind speed contributing most strongly (standardized significance 100%). Prediction models built using backpropagation neural networks were all successful, and models incorporating the three years of prior meteorological data outperformed single-year models—resonating precisely with the Wu Yun Liu Qi concept of “three-year epidemic latency” (PMID 26525546).
This logic was replicated across other infectious diseases: measles incidence correlated significantly with meteorological variables from the current and preceding three years, where models integrating three years of data improved accuracy by 76% compared to single-year models (PMID 30467697); bacterial dysentery showed significant positive correlations with precipitation, humidity, vapor pressure, and temperature, but negative correlation with wind speed, yielding high prediction accuracy (PMID 22903445); wavelet analysis of scarlet fever outbreak cycles revealed a primary 5-year cycle matching the Five Movements and a secondary 28-year cycle consistent with the Six Gases’ description of “quasi-cycles and multi-cycle superposition” (PMID 23713246).
Four independent studies point to a single conclusion: The meteorological-disease associations within the Wu Yun Liu Qi framework are real, and the latency periods and cyclical structures can generally be reproduced by data. This is its strongest asset as a predictive model.
Where the Model Fails Under Warming Conditions
However, “established correlation” does not equal “accurate prediction.” When applied to the context of climate warming, the model exhibits at least three shortcomings:
- No Trend Term Modeled. The Wu Yun Liu Qi theory is a pure cyclical model without trend terms. The mean elevation caused by global warming (e.g., rising winter average temperatures) does not belong to any Heavenly Stem/Earthly Branch cycle; the model naturally cannot depict it. When “record-breaking” events like extreme heatwaves or heavy rainfall increase in frequency, cyclical predictions are most prone to failure.
- Significant Regional Disparities. Originating from the Yellow River basin, the theory shows varying fit rates across regions. In Wuxi, 60 years of meteorological data matched the Wu Yun Liu Qi derivation at approximately 65%–91% (80% for Heavenly Stems, 83%–92% for Governing Qi and Receiving Qi, 100% for Six Principal Gases). In contrast, Guangzhou studies show age-cycle predictions reaching 90% fit, yet predictions for the first half of the year based on Governing Qi only reached 25% fit. The efficacy of the same Heavenly Stem/Earthly Branch system varies drastically across different climatic zones.
- Weak Evidence at the RCT Level. Systematic reviews indicate that the methodological quality of 13 RCT studies guided by Wu Yun Liu Qi is generally low, precluding definitive conclusions—the “correlation” of prediction has not yet translated into the “efficacy” of intervention.
Making the Model Useful in the Warming Era
Acknowledging limitations does not mean abandoning the framework. Re-positioning Wu Yun Liu Qi as a “cycle model with trend correction” retains its value: First, using segmented wavelet analysis (separating pre-warming and post-warming periods for cycle detection) can quantify whether cycles drift and by how much; this is a direct empirical possibility. Second, treating Heavenly Stem/Earthly Branch cycles as macro-background variables to be overlaid with modern epidemiological seasonal ARIMA models and machine learning algorithms rather than viewed as mutually exclusive competitors; the 3-year latency structure provided by Wu Yun Liu Qi is precisely a slow variable often overlooked in modern models. Third, matching extreme climate event databases (heatwaves, heavy rain, cold waves) against Heavenly Stem/Earthly Branch years of “excess or deficiency” can identify the climate types where the model excels and where it fails. In this manner, Wu Yun Liu Qi downgrades from “weather prediction” to “characterizing slow background cycles,” thereby clarifying its accuracy boundaries.
Evidence and Limitations
Verified:
– Beijing 15-year retrospective studies on tuberculosis, measles, and dysentery confirmed significant correlations between meteorological factors and incidence rates; the “three-year epidemic latency” structure can be reproduced (PMID 26525546, 30467697, 22903445)
– Scarlet fever primary 5-year cycle aligns with Five Movements cycle (PMID 23713246)
– Fit rates between Wu Yun Liu Qi derivation and actual meteorology reached 65%–91% in certain regions (Wuxi)
Partially Verified:
– Cyclical assumption holds only partially: Late winter 5-year cycle matches Five Movements, but early summer cycles do not match Six Gases’ 6-year cycle
– Disparities in regional fit rates are vast (Guangzhou Governing Qi prediction only 25%), with the North outperforming the South
– “Correlation” is established, but “prediction precision” does not meet standards of modern epidemiological models
Pending Verification:
– Cyclical drift caused by climate warming: Whether Wu Yun cycles undergo phase/frequency shifts under long-term trend superposition
– Fit rates between extreme climate events (heatwaves, heavy rain, cold waves) and Heavenly Stem/Earthly Branch years of “excess or deficiency”
– Head-to-head comparison of Wu Yun predictions with modern statistical models (e.g., seasonal ARIMA, machine learning) on concurrent data
Related Reading
- Wu Yun Liu Qi = Climate Cycle Medicine
- Sunspot Cycle and Epidemic Rhythm
- Birth Season Effects and Physiological Imprints
- Research Pool: Spatio-Temporal Frequency Topology
Evidence Level: A (Direct PubMed support)
References: PMID 26525546, PMID 30467697, PMID 22903445, PMID 23713246
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.
Leave a Reply