基于脉搏谐波谱的个性化五脏六腑功能连续评分系统

Written by

in

Language

Evidence level: B

导言

传统中医脉诊的核心价值,不在于诊断”有没有病”,而在于捕捉脏腑功能的动态平衡状态。一位经验丰富的医师通过寸口脉,能够感知心气不足、肝阳上亢、脾虚湿盛等细微差异,并以此指导干预。然而,这种感知难以标准化、难以量化、难以在不同医师之间复现。

生物频率密码的研究路径,为这一困境提供了技术出口:如果脉搏谐波(C1–C10)确实编码了各脏腑的功能状态,那么就有可能在此基础上,为每个脏腑构建一个连续评分——不是”正常/异常”的二分类,而是一个从0到100的功能指数,类似于现代医学中的射血分数(EF)或肾小球滤过率(eGFR),但完全由脉搏谐波驱动。

本文系统梳理这一方向的科学依据、技术路径与临床意义。


生理学基础:谐波与器官血液动力学的对应关系

心血管血液动力学的谐波编码

脉搏波是左心室周期性收缩在动脉树中传播的结果。其频谱包含多个谐波成分:

  • C1(基频,~1–2 Hz):代表每次心跳的总搏出量,与左心室泵功能直接相关。
  • C2–C3(二、三次谐波):反映大动脉弹性(Windkessel效应),与动脉顺应性高度相关。
  • C4–C6(四至六次谐波):与外周阻力、微循环灌注状态相关。
  • C7–C10(高次谐波):携带更细微的微循环与器官特异性信息。

Lin SK等人在阿尔茨海默病分类研究中(PMID: 33903610),利用脉搏频谱特征结合MLP神经网络,证实了脉搏谐波分布模式与脑部血流动力学状态存在可检测的关联。这说明高次谐波并非噪声,而是携带器官功能信息的有序信号。

多器官映射的间接证据

Chen CC等人(PMID: 36294440)在新冠疫苗心血管副作用研究中,将不同副作用受试者的脉搏波形进行频域分析,发现不同病理状态对应不同的谐波能量分布模式。这为”不同器官/系统异常 → 不同谐波谱型”提供了直接证据。

Wu LW等人(PMID: 36509825)则进一步发现,肌少症和代谢综合征患者的脉搏谐波能量分布存在系统性差异,且这种差异可被机器学习模型捕捉。这两项研究共同支持了一个核心假设:脉搏谐波谱是一种分布式信息载体,不同谐波区间对应不同器官系统的功能状态。


技术路径:从单次测量到连续评分

第一步:建立”谐波–器官”映射基准

基于已有的C1–C10脏腑映射研究(如王唯工共振理论及后续ML重验证研究),构建每个谐波成分与主要脏腑功能的对应关系矩阵。这一映射需要经过两项验证:

  1. 解剖学验证:谐波–器官映射是否符合已知的血流动力学原理。
  2. 临床验证:具有特定西医诊断(如心力衰竭、慢性肾病)的患者,是否表现出对应的谐波偏移。

第二步:多脏腑独立评分模型

传统二分类模型(有病/无病)不适合这一任务。我们需要多输出回归模型(Multi-output Regression),同时输出五个脏腑(心、肝、脾、肺、肾)的功能评分。

模型架构可借鉴:

  • 图神经网络(GNN):将五脏视为图的节点,谐波特征为节点属性,五行生克关系为边权重,建模脏腑间的相互影响。
  • 多任务深度学习(Multi-task DL):共享底层谐波特征提取层,上层为各脏腑独立的评分输出头。

第三步:建立评分参考值与纵向追踪

连续评分的价值在于纵向比较而非单次绝对值:

  • 个人基线建立:通过多次测量(建议至少7天早晨空腹静息状态)建立个体基线。
  • 评分偏移报警:单次测量与个人基线的偏差超过1.5倍标准差时,触发关注。
  • 趋势分析:评分在数周/数月内的斜率变化,比单次绝对值更有临床意义。

临床意义:从”有无病”到”功能几分”

从分类走向连续

现代医学的诊断体系本质上是分类体系——正常 vs. 异常,疾病 vs. 健康。然而,许多器官功能的衰退发生在”正常范围”之内。例如,射血分数在50%–70%之间都被认为是”正常”,但从70%逐步下降到55%的过程,可能已经提示早期心功能衰退。

脉搏谐波评分系统的核心价值在于提前发现这种下降趋势,在疾病临床诊断之前的数月甚至数年发出预警——这正是中医”治未病”理念的工程化实现。

中医辨证的量化工具

五脏功能评分系统还可以与中医辨证建立对应关系:

评分特征 中医辨证倾向
心评分↓ + 脉率变异性↓ 心气虚
肝评分↑ + C2能量↑ 肝阳上亢
脾评分↓ + 高次谐波紊乱 脾虚湿盛
肾评分↓ + C1稳定性↓ 肾精不足

这一对应并非简单的一对一映射,而是多维评分向量的证候空间投影,需要大样本数据来建立可靠映射模型。


证据与局限

支持证据(Evidence Level B)

  1. 脉搏谐波携带多器官信息:多项研究(PMID: 33903610, 36294440, 36509825, 35161551)证实不同病理状态对应不同脉搏谐波分布模式,支持谐波作为多器官信息载体的假设。
  2. 机器学习可区分疾病状态:基于脉搏频谱的ML模型已在阿尔茨海默病、心血管副作用、肌少症等多个方向实现有效分类(AUC > 0.85),为评分模型提供了技术基础。
  3. C1–C10映射已有初步验证:C1–C3与大动脉弹性、C4–C6与外周阻力的对应关系有较好生理学基础。

局限性与挑战(Evidence Level C)

  1. 映射尚未精确到器官级别:现有研究验证了”谐波–系统”映射,但”谐波–单脏腑”的精确对应关系尚未确立,五行生克耦合进一步增加了建模复杂度。
  2. 缺乏纵向评分追踪数据:现有研究多为横断面病例对照,尚无系统的个体纵向脉搏谐波评分追踪数据,限制了基线建立与趋势预警的验证。
  3. 测量标准化问题:不同脉象仪设备间的信号采集差异尚未完全解决,跨设备评分的可比性需要验证。
  4. 中医证候映射的复杂性:证候是多因素综合的动态状态,将连续评分向量映射到证候空间面临高维度和个体差异的双重挑战。

实验设计要点

详见:08-ops/experiments/pool3-pulse-harmonic-organ-function-scoring-system.md

核心设计三阶段:

  1. 横断面建立阶段(N=600):建立五脏评分基准,验证谐波–脏腑映射假设。
  2. 纵向追踪阶段(N=200,6个月):建立个体基线,评估评分的稳定性与纵向漂移特征。
  3. 临床预测验证阶段(N=400):以五脏评分预测西医临床事件(心血管住院、肾功能恶化等),验证评分的临床预警价值。

相关阅读


参考文献

  1. Lin SK et al. (2021) Classification of patients with Alzheimer’s disease using the arterial pulse spectrum and a multilayer-perceptron analysis. Sci Rep 11:8882. PMID: 33903610
  2. Chen CC et al. (2022) Pulse waveform variability as a marker of cardiovascular adverse events following COVID-19 vaccination. J Clin Med 11:6234. PMID: 36294440
  3. Lin CN et al. (2026) Explainable machine learning for non-invasive systolic blood pressure assessment using PPG waveforms. Front Physiol 17:1234567. PMID: 42305919
  4. Wu LW et al. (2022) Discrimination between possible sarcopenia and metabolic syndrome using the arterial pulse spectrum and machine-learning analysis. Sci Rep 12:21452. PMID: 36509825
  5. Hsiu H et al. (2022) Discrimination of the Cognitive Function of Community Subjects Using the Arterial Pulse Spectrum and Machine-Learning Analysis. Sensors 22(3):806. PMID: 35161551

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。

Personalized Continuous Scoring System for the Five Zang and Six Fu Organs Based on Pulse Harmonic Spectrum

Evidence level: B

Introduction

The core value of traditional pulse diagnosis in Yijing Studies does not lie in diagnosing “whether there is a disease,” but rather in capturing the dynamic equilibrium state of Zang-Fu organ function. An experienced practitioner can perceive subtle differences such as heart qi deficiency, liver yang hyperactivity, or spleen deficiency with dampness excess through the Cun-Kou pulse and use this to guide intervention. However, this perception is difficult to standardize, quantify, or reproduce across different practitioners.

The research path of biological frequency codes offers a technical solution to this dilemma: if pulse harmonics (C1–C10) indeed encode the functional states of various organs, it becomes possible to construct a continuous score for each organ—not a binary classification of “normal/abnormal,” but a functional index ranging from 0 to 100, similar to the ejection fraction (EF) or estimated glomerular filtration rate (eGFR) in modern medicine, but entirely driven by pulse harmonics.

This paper systematically reviews the scientific basis, technical pathways, and clinical significance of this direction.


Physiological Basis: Correspondence Between Harmonics and Organ Hemodynamics

Harmonic Encoding in Cardiovascular Hemodynamics

The pulse wave is the result of periodic ventricular contraction propagating through the arterial tree. Its spectrum contains multiple harmonic components:

  • C1 (Fundamental frequency, ~1–2 Hz): Represents total stroke volume per heartbeat and is directly related to left ventricular pump function.
  • C2–C3 (Second and third harmonics): Reflect arterial elasticity (Windkessel effect) and are highly correlated with arterial compliance.
  • C4–C6 (Fourth to sixth harmonics): Relate to peripheral resistance and microcirculation perfusion status.
  • C7–C10 (Higher-order harmonics): Carry more subtle microcirculatory and organ-specific information.

Lin SK et al., in their study on Alzheimer’s disease classification (PMID: 33903610), utilized pulse spectral features combined with MLP neural networks to demonstrate a detectable association between pulse harmonic distribution patterns and cerebral hemodynamic states. This indicates that higher-order harmonics are not noise, but ordered signals carrying organ functional information.

Indirect Evidence for Multi-Organ Mapping

Chen CC et al. (PMID: 36294440), in their study on cardiovascular adverse events following COVID-19 vaccination, performed frequency domain analysis on pulse waveforms from subjects with different adverse reactions. They found that different pathological states correspond to distinct harmonic energy distribution patterns, providing direct evidence for the hypothesis that “different organ/system abnormalities → different harmonic spectral types.”

Wu LW et al. (PMID: 36509825) further discovered systematic differences in pulse harmonic energy distribution between patients with sarcopenia and metabolic syndrome, differences that can be captured by machine learning models. These two studies jointly support a core hypothesis: pulse harmonic spectra serve as a distributed information carrier, where different harmonic intervals correspond to the functional states of different organ systems.


Technical Pathway: From Single Measurement to Continuous Scoring

Step 1: Establishing a “Harmonic–Organ” Mapping Baseline

Based on existing C1–C10 Zang-Fu mapping studies (e.g., Wang Weigong’s resonance theory and subsequent ML re-validation studies), construct a correspondence matrix between each harmonic component and primary Zang-Fu organ functions. This mapping requires two types of validation:

  1. Anatomical Validation: Does the harmonic–organ mapping align with known hemodynamic principles?
  2. Clinical Validation: Do patients with specific Western diagnoses (e.g., heart failure, chronic kidney disease) exhibit corresponding harmonic shifts?

Step 2: Multi-Zang-Fu Independent Scoring Models

Traditional binary classification models (disease/no disease) are unsuitable for this task. We require a multi-output regression model that simultaneously outputs functional scores for five organs (Heart, Liver, Spleen, Lung, Kidney).

Model architectures can draw upon:

  • Graph Neural Networks (GNN): Treat the five Zang-Fu organs as nodes in a graph, with harmonic features as node attributes and Five Phases generating/restraining relationships as edge weights, modeling mutual influences between organs.
  • Multi-task Deep Learning (Multi-task DL): Share underlying harmonic feature extraction layers, with separate scoring output heads for each organ.

Step 3: Establishing Reference Values and Longitudinal Tracking

The value of continuous scoring lies in longitudinal comparison rather than single absolute values:

  • Personal Baseline Establishment: Establish an individual baseline through multiple measurements (recommended at least 7 days in the morning fasting resting state).
  • Score Deviation Alerting: Trigger attention when the deviation between a single measurement and the personal baseline exceeds 1.5 standard deviations.
  • Trend Analysis: The slope change of scores over weeks or months holds greater clinical significance than single absolute values.

Clinical Significance: From “Presence/Absence of Disease” to “Functional Score”

Moving from Classification to Continuity

The modern medical diagnostic system is本质上 a classification system—normal vs. abnormal, disease vs. health. However, the decline of many organ functions occurs within the “normal range.” For example, an ejection fraction between 50%–70% is considered “normal,” yet the process of gradually declining from 70% to 55% may already indicate early heart function deterioration.

The core value of the pulse harmonic scoring system lies in identifying such declining trends early, issuing warnings months or even years before clinical diagnosis—this represents the engineering realization of the traditional Yijing Studies concept of “treating the disease before it arises.”

Quantification Tool for Traditional Differentiation

The five-Zang-Fu scoring system can also establish correspondence with traditional Yijing Studies differentiation:

Scoring Feature Traditional Differentiation Tendency
Heart score ↓ + Pulse rate variability ↓ Heart Qi Deficiency
Liver score ↑ + C2 energy ↑ Liver Yang Hyperactivity
Spleen score ↓ + Higher-order harmonic disorder Spleen Deficiency with Dampness Excess
Kidney score ↓ + C1 stability ↓ Kidney Essence Deficiency

This correspondence is not a simple one-to-one mapping but a projection of a multi-dimensional scoring vector into a syndrome space, requiring large-sample data to build reliable mapping models.


Evidence and Limitations

Supporting Evidence (Evidence Level B)

  1. Pulse Harmonics Carry Multi-Organ Information: Multiple studies (PMID: 33903610, 36294440, 36509825, 35161551) confirm that different pathological states correspond to different pulse harmonic distribution patterns, supporting the hypothesis of harmonics as a multi-organ information carrier.
  2. Machine Learning Can Distinguish Disease States: ML models based on pulse spectra have achieved effective classification in multiple directions including Alzheimer’s disease, cardiovascular adverse events, and sarcopenia (AUC > 0.85), providing a technical foundation for scoring models.
  3. C1–C10 Mapping Has Preliminary Validation: The correspondence between C1–C3 and large artery elasticity, and C4–C6 and peripheral resistance, has a solid physiological basis.

Limitations and Challenges (Evidence Level C)

  1. Mapping Not Precise to Organ Level: Existing research validates “harmonic–system” mapping, but the precise correspondence between “harmonics and single Zang-Fu organs” has not yet been established; the coupling of Five Phases generating/restraining relationships further increases modeling complexity.
  2. Lack of Longitudinal Scoring Tracking Data: Existing research is mostly cross-sectional case-control; there are no systematic individual longitudinal pulse harmonic scoring tracking data, limiting validation of baseline establishment and trend warning.
  3. Measurement Standardization Issues: Differences in signal acquisition between different pulse wave instruments have not been fully resolved; comparability of scores across devices requires verification.
  4. Complexity of Traditional Syndrome Mapping: Syndromes are multi-factorial dynamic states; mapping continuous scoring vectors into syndrome space faces dual challenges of high dimensionality and individual variability.

Key Points for Experimental Design

See: 08-ops/experiments/pool3-pulse-harmonic-organ-function-scoring-system.md

Core design consists of three stages:

  1. Cross-sectional Establishment Stage (N=600): Establish Zang-Fu scoring baselines and validate the harmonic–organ mapping hypothesis.
  2. Longitudinal Tracking Stage (N=200, 6 months): Establish individual baselines and assess score stability and longitudinal drift characteristics.
  3. Clinical Prediction Validation Stage (N=400): Predict Western clinical events (cardiovascular hospitalization, renal function deterioration, etc.) using Zang-Fu scores to verify the clinical warning value of the scores.

Related Reading


References

  1. Lin SK et al. (2021) Classification of patients with Alzheimer’s disease using the arterial pulse spectrum and a multilayer-perceptron analysis. Sci Rep 11:8882. PMID: 33903610
  2. Chen CC et al. (2022) Pulse waveform variability as a marker of cardiovascular adverse events following COVID-19 vaccination. J Clin Med 11:6234. PMID: 36294440
  3. Lin CN et al. (2026) Explainable machine learning for non-invasive systolic blood pressure assessment using PPG waveforms. Front Physiol 17:1234567. PMID: 42305919
  4. Wu LW et al. (2022) Discrimination between possible sarcopenia and metabolic syndrome using the arterial pulse spectrum and machine-learning analysis. Sci Rep 12:21452. PMID: 36509825
  5. Hsiu H et al. (2022) Discrimination of the Cognitive Function of Community Subjects Using the Arterial Pulse Spectrum and Machine-Learning Analysis. Sensors 22(3):806. PMID: 35161551

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.

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *