Evidence level: B
导言
脉搏谐波分析已成为中医脉诊客观化的核心技术路径。王唯工的”谐波-脏腑映射”假说提出,C1至C10十个谐波分量分别对应心、肝、脾、肺、肾五个脏腑的共振频率——C1/C6对应心、C2/C7对应肝、C3/C8对应脾、C4/C9对应肺、C5/C10对应肾。
然而,这一映射框架面临一个关键的方法学挑战:各谐波分量之间存在高度共线性(C1至C10之和等于原始脉搏波),且年龄、性别、体质指数(BMI)、心血管危险因素等混杂变量同样影响脉搏谐波谱。已有研究多停留在相关性报告层面,未能控制这些混杂因素,因而无法证明某个谐波分量对特定证候具有独立于其他谐波和混杂变量的解释力。
本课题旨在填补这一空白:建立多变量回归模型,在控制年龄、性别、BMI、血压、血糖等协变量后,系统评估C1-C10各分量对心/肝/脾/肺/肾虚实证候的独立预测贡献(incremental R²、标准化回归系数β、独立效应量)。
背景:谐波-脏腑映射的现状与局限
理论基础
中医脉诊的”三部九候”体系将寸关尺三部分别对应不同脏腑:左手寸关尺对应心、肝、肾;右手寸关尺对应肺、脾、命门。脉搏波的频域分解为这一古老框架提供了现代生理学诠释:每个谐波分量代表动脉压力波在特定共振频率上的能量权重。
王唯工实验室的早期工作通过脉搏波频谱分析发现,肝经对应者的第二谐波(C2)幅值显著高于非对应者,脾经对应者的第三谐波(C3)亦然。这一发现在小样本(每组约30-60人)研究中具有统计显著性,但存在若干方法学缺陷:样本量不足、未进行多重比较校正、缺乏外部验证集、混杂变量控制不充分。
现有证据的局限性
近年来的机器学习研究为脉诊客观化提供了新证据。Qiao等(2025,PMID 41087424)利用卷积神经网络整合脉搏波形与临床特征,对妊娠和冠心病进行识别,准确率达到80%以上。Yeh等(2024,PMID 38927797)开发了基于深度学习的糖尿病脉诊辅助诊断系统。Wu等(2026,PMID 42487941)则证明谐波分析可以区分乙肝和丙肝感染患者的脉搏特征差异。
然而,这些研究的共性局限在于:将脉搏波形作为整体特征输入黑箱模型,未将谐波分量解耦为独立预测变量,无法回答”C3对脾虚证是否有独立于C2和C4的解释力”这一关键问题。换言之,这些研究验证了”脉搏波形整体与证候相关”,但未验证”特定谐波与特定脏腑证候存在独立映射关系”。
独立相关性建模的方法学框架
研究设计与统计模型
独立相关性建模的核心思路是:对于每个目标证候(如脾虚证),以该证候为因变量(0/1二分类),以C1-C10十个谐波分量加上所有混杂变量作为自变量,拟合多元 logistic 回归模型(或 penalized logistic 回归处理共线性)。
模型结构如下:
logit(P(Y_脾虚=1)) = β₀ + β₁·C1 + β₂·C2 + ... + β₁₀·C10 + β₁₁·age + β₁₂·sex + β₁₃·BMI + β₁₄·SBP + β₁₅·FPG + ε
关键指标:
– 独立解释力(incremental R²):在控制协变量后,逐个加入C1-C10,计算每个谐波对模型R²的增量贡献(以Nagelkerke R²计量)
– 标准化回归系数(β):反映控制其他变量后,该谐波每改变一个标准差单位,证候优势比(OR)的对数变化
– 方差膨胀因子(VIF):检测C1-C10之间的多重共线性,VIF>5时需采用岭回归(Ridge)或主成分回归
混杂变量控制
混杂变量的选择需基于领域知识与相关性分析:
| 混杂变量 | 纳入理由 | 测量方式 |
|---|---|---|
| 年龄 | 心率变异性、动脉僵硬度随年龄非线性变化 | 岁 |
| 性别 | 性激素影响动脉顺应性和谐波谱 | 男/女 |
| BMI | 肥胖改变中心动脉压和反射波 timing | kg/m² |
| 收缩压(SBP) | 血压直接调制脉搏波幅值和反射 | mmHg |
| 空腹血糖(FPG) | 血糖波动影响动脉壁弹性和波速 | mg/dL |
| 心率 | 心率改变基频,直接影响谐波频率轴 | bpm |
多重比较校正
由于需对5个脏腑×2个证型(虚/实)= 10个独立模型进行假设检验,显著性阈值需采用 Bonferroni 校正(α = 0.05/10 = 0.005)或 Benjamini-Hochberg FDR 控制(q < 0.05)。
预期发现与解读
预期的独立相关模式
若谐波-脏腑映射假说成立,在控制混杂变量后,研究者应观察到以下独立相关模式:
- 心虚证:C1(第一谐波,对应心经基频)和C6(第六谐波,心经二次谐波)的标准化系数β显著为负,表明心虚证患者的C1/C6能量相对衰减
- 肝郁证(实证):C2/C7的β显著为正,提示肝气郁结者的第二谐波能量增强
- 脾虚证:C3/C8的β显著为负,脾失健运者的第三谐波能量降低
- 肺虚证:C4/C9的β显著为负,肺气不足者的第四谐波能量减弱
- 肾虚证:C5/C10的β显著为负,腰膝酸软者的第五谐波能量降低
阴性结果的科学价值
若某一谐波对相应证候未能显示独立相关性,可能的解释包括:
- 谐波映射理论需要修正:C1-C10并非简单的一一对应脏腑,而可能是脏腑联动的整体模式
- 证候界定标准不统一:不同医师对”脾虚证”的判定一致性不足(Kappa < 0.6),稀释了真实效应
- 非线性关系被线性模型遗漏:谐波与证候之间可能存在阈值效应或交互作用
- C1-C10共线性过强:需用降维方法(PCA、因子分析)重新提取独立特征
临床意义与研究价值
独立相关性建模一旦成功,其临床转化价值远超描述性相关研究:
- 精准辨证:若C3对脾虚证具有独立于其他谐波的解释力(ΔR² > 5%),可将C3作为脾虚证的客观生物标志物,辅助年轻医师减少辨证误差
- 证候量化评分:基于独立回归系数,可为每个脏腑构建连续的”谐波功能评分”(0-100分),实现从”有病/无病”二分类向连续功能评估的跨越
- 疗效监测:在中药干预前后,追踪C3/C8等目标谐波的变化轨迹,为”证候改善”提供可量化的生理学终点
- 反向验证经络本质:若特定谐波仅对特定脏腑证候有独立解释力(而对其他证候无显著效应),则为”经络=低电阻传输通道+谐振频率载体”这一整合假说提供关键证据
证据与局限
证据等级评估
- 证据等级:B(中等)
- 现有研究支持脉搏谐波与中医证候之间存在相关性(Wu 2026,Qiao 2025,Yeh 2024),但均未在控制充分混杂变量后验证独立相关性
- 本课题的创新性在于首次系统建立独立相关性模型,填补了这一方法学空白
- 需要多中心、前瞻性验证队列来提升证据等级至A
主要局限
- 混杂变量遗漏:社会经济地位、饮食结构、运动习惯等亦影响脉搏谐波和证候表现,但难以在单一研究中全部控制
- 证候金标准缺失:中医证候判定依赖医师主观判断,不同医师间的Kappa值可能偏低;需采用”三人独立盲法评定多数决”来降低标签噪声
- 横断面设计:无法建立因果关系——独立相关性不等于谐波变化导致证候变化;需要干预研究(前推研究、随机对照针刺/中药研究)来验证方向性
- C1-C10共线性:高度共线性可能掩盖部分真实效应,需结合正则化方法和敏感性分析评估结果稳健性
- 样本代表性:多数研究在亚洲人群中进行,结论向西方人群的外推性有待验证
相关阅读
- Wu CH, Liao KF, Chen JJ, Wu H. Harmonic analysis of radial pulse in traditional Chinese medicine: physiological alterations associated with hepatitis B and hepatitis C infections. Front Med (Lausanne). 2026;13:1848946. PMID: 42487941
- Qiao M, Yan Z, Tan C, et al. Integrating traditional Chinese pulse diagnosis with machine learning: novel approaches for pregnancy and coronary heart disease identification. Sci Rep. 2025;15(1):35857. PMID: 41087424
- Yeh WC, Kuo CY, Chen JM, et al. Pioneering Data Processing for Convolutional Neural Networks to Enhance the Diagnostic Accuracy of Traditional Chinese Medicine Pulse Diagnosis for Diabetes. Bioengineering (Basel). 2024;11(6):561. PMID: 38927797
- Jiang Z, Zhang D, Lu G. Radial artery pulse waveform analysis based on curve fitting using discrete Fourier series. Comput Methods Programs Biomed. 2019;174:25-31. PMID: 29709314
- Moura NG, Cordovil I, Ferreira Ade S. Traditional Chinese medicine wrist pulse-taking is associated with pulse waveform analysis and hemodynamics in hypertension. J Integr Med. 2016;14(2):100-113. PMID: 26988431
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。
Modeling the Independent Correlation Between Pulse Harmonic C1-C10 and Traditional Chinese Medicine Syndromes
Evidence level: B
Introduction
Pulse harmonic analysis has emerged as a core technical pathway for the objectification of Traditional Chinese Medicine (TCM) pulse diagnosis. Wang Weigong’s “Harmonic-Organ Mapping” hypothesis posits that the ten harmonic components (C1 to C10) correspond to the resonant frequencies of the five Zang-fu organs: C1/C6 for the Heart, C2/C7 for the Liver, C3/C8 for the Spleen, C4/C9 for the Lungs, and C5/C10 for the Kidneys.
However, this mapping framework faces a critical methodological challenge: high multicollinearity among harmonic components (as the sum of C1 to C10 equals the original pulse wave), alongside confounding variables such as age, sex, body mass index (BMI), and cardiovascular risk factors that also influence the pulse harmonic spectrum. Existing research largely remains at the level of correlation reporting, failing to control for these confounders; consequently, it cannot demonstrate the independent explanatory power of a specific harmonic component for a particular syndrome relative to other harmonics and confounding variables.
This project aims to fill this gap by establishing multivariate regression models. After controlling for covariates such as age, sex, BMI, blood pressure, and blood glucose, it will systematically evaluate the independent predictive contribution of each C1-C10 component to Heart/Liver/Spleen/Lung/Kidney deficiency/excess syndromes (measured via incremental R², standardized regression coefficients β, and independent effect sizes).
Background: Status and Limitations of the Harmonic-Organ Mapping Framework
Theoretical Basis
The TCM pulse diagnosis “Three Parts, Nine Responses” system maps the Cun-Guan-Shi positions to specific organs: Left Cun-Guan-Shi corresponds to Heart, Liver, and Kidney; Right Cun-Guan-Shi corresponds to Lung, Spleen, and Mingmen. The frequency-domain decomposition of the pulse wave provides a modern physiological interpretation for this ancient framework, where each harmonic component represents the energy weight of arterial pressure waves at specific resonant frequencies.
Early work by the Wang Weigong laboratory found through pulse wave spectral analysis that the second harmonic (C2) amplitude was significantly higher in Liver meridian correspondents compared to non-correspondents, and similarly for the third harmonic (C3) in Spleen meridian correspondents. While these findings showed statistical significance in small-sample studies (approx. 30-60 per group), they suffered from methodological flaws: insufficient sample sizes, lack of multiple comparison corrections, absence of external validation sets, and inadequate control of confounding variables.
Limitations of Current Evidence
Recent machine learning research has provided new evidence for the objectification of pulse diagnosis. Qiao et al. (2025, PMID 41087424) utilized convolutional neural networks to integrate pulse waveforms with clinical features for identifying pregnancy and coronary heart disease, achieving accuracy rates above 80%. Yeh et al. (2024, PMID 38927797) developed a deep learning-based auxiliary diagnosis system for TCM pulse diagnosis in diabetes. Wu et al. (2026, PMID 42487941) demonstrated that harmonic analysis can distinguish pulse waveform differences between patients with Hepatitis B and Hepatitis C infections.
However, a common limitation across these studies is that they treat the pulse waveform as an overall feature input into black-box models, failing to decouple harmonic components into independent predictors. This prevents answering the critical question: “Does C3 have independent explanatory power for Spleen Deficiency beyond C2 and C4?” In other words, these studies validate the correlation between “overall pulse waveform and syndrome” but do not verify the existence of an “independent mapping relationship between specific harmonics and specific organ syndromes.”
Methodological Framework for Independent Correlation Modeling
Study Design and Statistical Models
The core logic of independent correlation modeling is as follows: for each target syndrome (e.g., Spleen Deficiency), the syndrome serves as the dependent variable (0/1 binary classification), while C1-C10 harmonic components plus all confounding variables serve as independent variables to fit a multivariate logistic regression model (or penalized logistic regression to address multicollinearity).
Model structure:
logit(P(Y_SpleenDeficiency=1)) = β₀ + β₁·C1 + β₂·C2 + ... + β₁₀·C10 + β₁₁·age + β₁₂·sex + β₁₃·BMI + β₁₄·SBP + β₁₅·FPG + ε
Key metrics:
– Independent Explanatory Power (incremental R²): After controlling for covariates, calculate the incremental contribution of each harmonic to the model’s R² upon sequential addition (measured via Nagelkerke R²).
– Standardized Regression Coefficient (β): Reflects the logarithmic change in the odds ratio for the syndrome per one standard deviation change in the harmonic, controlling for other variables.
– Variance Inflation Factor (VIF): Detects multicollinearity among C1-C10; if VIF > 5, Ridge regression or Principal Component Regression should be employed.
Control of Confounding Variables
Selection of confounding variables should be based on domain knowledge and correlation analysis:
| Confounding Variable | Rationale for Inclusion | Measurement Method |
|---|---|---|
| Age | Heart rate variability and arterial stiffness change non-linearly with age | Years |
| Sex | Sex hormones affect arterial compliance and harmonic spectra | Male/Female |
| BMI | Obesity alters central arterial pressure and reflected wave timing | kg/m² |
| Systolic Blood Pressure (SBP) | Blood pressure directly modulates pulse wave amplitude and reflection | mmHg |
| Fasting Plasma Glucose (FPG) | Blood glucose fluctuations affect arterial wall elasticity and wave velocity | mg/dL |
| Heart Rate | Heart rate changes the fundamental frequency, directly affecting the harmonic frequency axis | bpm |
Multiple Comparison Correction
Since hypothesis testing requires 10 independent models (5 organs × 2 syndrome types [deficiency/excess]), the significance threshold must be adjusted using Bonferroni correction (α = 0.05/10 = 0.005) or Benjamini-Hochberg False Discovery Rate (FDR) control (q < 0.05).
Expected Findings and Interpretation
Expected Independent Correlation Patterns
If the harmonic-organ mapping hypothesis holds, researchers should observe the following independent correlation patterns after controlling for confounding variables:
- Heart Deficiency: Standardized coefficients β for C1 (first harmonic, corresponding to the Heart meridian fundamental frequency) and C6 (sixth harmonic, Heart meridian second harmonic) should be significantly negative, indicating relative energy attenuation of C1/C6 in Heart Deficiency patients.
- Liver Stagnation (Excess): β for C2/C7 should be significantly positive, suggesting enhanced second harmonic energy in individuals with Liver Qi stagnation.
- Spleen Deficiency: β for C3/C8 should be significantly negative, indicating reduced third harmonic energy in individuals with impaired Spleen transportation.
- Lung Deficiency: β for C4/C9 should be significantly negative, reflecting weakened fourth harmonic energy in individuals with insufficient Lung Qi.
- Kidney Deficiency: β for C5/C10 should be significantly negative, indicating reduced fifth harmonic energy in individuals with soreness and weakness of the waist and knees.
Scientific Value of Negative Results
If a specific harmonic fails to show independent correlation with its corresponding syndrome, possible explanations include:
- Revision Needed for Harmonic Mapping Theory: C1-C10 may not represent a simple one-to-one correspondence with organs but rather an integrated pattern of organ interaction.
- Inconsistency in Syndrome Definitions: Inter-rater reliability for “Spleen Deficiency” among different physicians may be low (Kappa < 0.6), diluting the true effect.
- Non-linear Relationships Missed by Linear Models: Threshold effects or interactions may exist between harmonics and syndromes that linear models overlook.
- Excessive Multicollinearity among C1-C10: Dimensionality reduction methods (PCA, factor analysis) may be required to extract independent features.
Clinical Significance and Research Value
Once successful, independent correlation modeling offers clinical translational value far exceeding descriptive correlational studies:
- Precise Syndromic Differentiation: If C3 has independent explanatory power for Spleen Deficiency beyond other harmonics (ΔR² > 5%), C3 can serve as an objective biomarker for Spleen Deficiency, assisting junior physicians in reducing diagnostic errors.
- Syndrome Quantification Scoring: Based on independent regression coefficients, a continuous “Harmonic Function Score” (0-100 points) can be constructed for each organ, bridging the gap from binary “present/absent” classification to continuous functional assessment.
- Efficacy Monitoring: Tracking the trajectory of target harmonics (e.g., C3/C8) before and after herbal intervention provides quantifiable physiological endpoints for “syndrome improvement.”
- Reverse Validation of Meridian Essence: If specific harmonics exhibit independent explanatory power only for specific organ syndromes (with no significant effects on others), this provides critical evidence for the integrated hypothesis that “meridians = low-resistance transmission channels + resonant frequency carriers.”
Evidence and Limitations
Evidence Level Assessment
- Evidence Level: B (Moderate)
- Existing research supports correlations between pulse harmonics and TCM syndromes (Wu 2026, Qiao 2025, Yeh 2024), but none have verified independent correlations after fully controlling for confounding variables.
- The innovation of this project lies in systematically establishing independent correlation models for the first time, filling a methodological gap.
- Evidence level should be elevated to A with multi-center, prospective validation cohorts.
Major Limitations
- Confounding Variable Omission: Socioeconomic status, dietary structure, and exercise habits also influence pulse harmonics and syndrome manifestations but are difficult to control entirely in a single study.
- Lack of Gold Standard for Syndromes: TCM syndrome diagnosis relies on subjective physician judgment; inter-physician Kappa values may be low. Adopting a “three-person independent blind assessment with majority rule” is needed to reduce label noise.
- Cross-sectional Design: Cannot establish causality—independent correlation does not imply harmonic changes cause syndrome changes; intervention studies (lagged studies, randomized controlled trials of acupuncture/herbal medicine) are needed to verify directionality.
- Strong Multicollinearity among C1-C10: High multicollinearity may mask certain true effects; results robustness should be evaluated using regularization methods and sensitivity analysis.
- Sample Representativeness: Most studies are conducted in Asian populations; extrapolation to Western populations requires further verification.
Related Reading
- Wu CH, Liao KF, Chen JJ, Wu H. Harmonic analysis of radial pulse in traditional Chinese medicine: physiological alterations associated with hepatitis B and hepatitis C infections. Front Med (Lausanne). 2026;13:1848946. PMID: 42487941
- Qiao M, Yan Z, Tan C, et al. Integrating traditional Chinese pulse diagnosis with machine learning: novel approaches for pregnancy and coronary heart disease identification. Sci Rep. 2025;15(1):35857. PMID: 41087424
- Yeh WC, Kuo CY, Chen JM, et al. Pioneering Data Processing for Convolutional Neural Networks to Enhance the Diagnostic Accuracy of Traditional Chinese Medicine Pulse Diagnosis for Diabetes. Bioengineering (Basel). 2024;11(6):561. PMID: 38927797
- Jiang Z, Zhang D, Lu G. Radial artery pulse waveform analysis based on curve fitting using discrete Fourier series. Comput Methods Programs Biomed. 2019;174:25-31. PMID: 29709314
- Moura NG, Cordovil I, Ferreira Ade S. Traditional Chinese medicine wrist pulse-taking is associated with pulse waveform analysis and hemodynamics in hypertension. J Integr Med. 2016;14(2):100-113. PMID: 26988431
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