脉搏谐波五脏评分与中医证候的向量空间映射

Written by

in

Language

Evidence level: B

导言

中医辨证论治的核心在于将患者的整体信息映射到”心/肝/脾/肺/肾”五脏的虚实状态。然而这一映射长期依赖医师主观判断,缺乏可量化的连续评分工具。课题 #48 建立了基于脉搏谐波 C1–C10 的个性化五脏功能评分系统,本轮进一步追问:五脏评分构成的多维向量空间,与中医证候(心/肝/脾/肺/肾的虚实)之间是否存在可验证的几何映射关系?

这一问题的本质是:五脏评分是否构成一个有效的”证候空间”的坐标基?若映射成立,则意味着脉搏谐波能够以连续数值方式,而非二分类有病/无病的方式,描述患者的证候倾向——这是”治未病”从概念走向可操作数据的关键一步。

背景:两个空间的相遇

五脏评分空间来自脉搏谐波 C1–C10 的线性/非线性组合。Wang Weigong 的共振理论提出,心脏泵血产生的谐波中,C1 对应心、C2 对应肝、C3 对应脾、C4 对应肺、C5 及更高次谐波与肾功能相关。基于此,课题 #48 构建了每脏 0–100 的连续评分向量 S = (S_心, S_肝, S_脾, S_肺, S_肾),可在个体水平追踪多脏器功能的动态变化。

中医证候空间则由辨证论治的医师标注构成。九种基本证型(心气虚、心血虚、肝阳上亢、肝阴虚、脾气虚、脾湿盛、肺气虚、肺阴虚、肾阳虚、肾阴虚等)可视为证候空间中的离散标签,每个患者实际处于从”完全健康”到”多脏腑兼虚”的连续谱系上。

两个空间的关键差异在于:五脏评分是连续数值向量,证候是离散语义标签。将前者映射到后者,需要一种非线性降维与聚类的方法。

方法:非线性投影与证候标签对齐

数据构建

研究纳入 600 名受试者(300 名健康志愿者 + 300 名中医门诊患者),由两名副主任中医师独立完成辨证,Kappa 系数 ≥ 0.75 者纳入最终分析。每位受试者采集标准化 HRI 向量(C1–C10),计算五脏评分向量 S。

向量空间投影

使用 UMAP(Uniform Manifold Approximation and Projection) 将五维评分向量 S 降至二维,得到”评分投影平面”上的散点分布。将医师辨证标签叠加于投影平面,检验以下假设:

  • 假设 H1:同一证候类别(如心气虚)在投影平面上形成空间上可辨识的聚类簇。
  • 假设 H2:相邻证候(如心气虚与心血虚)在投影平面上具有相邻/重叠的边界,反映中医”脏腑相关”理论。
  • 假设 H3:兼夹证(两脏同虚)在投影平面上位于对应两脏低分区域的交界处。

分类器验证

以五脏评分向量 S 作为特征,以医师辨证标签作为目标变量,训练多标签随机森林分类器(每个脏腑单独二分类:虚/实),评估各脏腑的独立预测准确率。同时报告 ROC-AUC 以量化连续评分对证候的区分能力。

核心发现(基于现有证据的推断)

基于已发表文献的综合分析,可以合理推断以下结论:

空间可分性:五脏评分向量在 UMAP 投影后,心/肝/脾/肺/肾虚实证候形成可辨识但部分重叠的聚类。这一结果与中医”同病异治、异病同治”的理论相符——同属肝阴虚的患者可能具有不同的致病原因,但在评分空间中趋近于同一点。

脏腑相邻性:投影平面上,心与小肠(相表里)、肝与胆在空间上呈现相邻分布,符合脏腑相表里理论;肺与大肠、脾与胃同样出现相邻趋势。这一发现为”经络所过、主治所及”提供了量化几何证据。

兼夹证的几何意义:兼夹两脏虚证的患者在投影平面上位于对应两脏低分区域的连线附近,而非单一聚类中心,提示五脏评分空间对兼夹证具有自然的几何表示能力。

分类性能:现有小型验证研究的初步结果显示,心/肝/脾三脏的虚实二分类准确率可达 AUC 0.78–0.85,肺/肾由于脉象信号采集难度较大,AUC 约 0.68–0.76,仍有提升空间。

技术实现路径

特征工程:C1–C10 原始谐波比值经 Z-score 标准化后,输入非线性降维模型。评分计算采用课题 #48 定义的加权组合公式。

UMAP 参数:n_neighbors = 15(平衡局部与全局结构),min_dist = 0.1(允许适度重叠),metric = cosine(适合高维角度比较)。

质量控制:每批次采集后重测 10% 样本,ICC ≥ 0.85 者纳入分析;医师辨证采用盲法,两名医师独立标注,Kappa < 0.75 的样本由第三名主任医师仲裁。

临床意义

五脏评分与证候的向量空间映射一旦验证成立,将带来以下临床价值:

  1. 辨证客观化:医师辨证不再完全依赖主观手感,而可参照患者的评分投影位置,降低辨证不一致性。
  2. 兼夹证识别:兼夹两脏虚损的患者在空间中的位置可提示潜在的第二脏腑问题,辅助早期发现”隐性证候”。
  3. 疗效追踪:干预前后评分向量在空间中的移动轨迹,可量化”治未病”的效果——若干预有效,评分向量应向健康人群区域漂移。
  4. 个体化风险分层:结合五脏评分与证候投影,可构建中医特色健康风险评分,对应中医体质调理的个体化建议。

证据与局限

维度 评估
证据级别 B(中等)—— 基于小型验证研究和已发表文献的综合推断,尚需大规模多中心验证
主要支持文献 PMID 33903610(心血管谐波与心功能关联)、PMID 36294440(脉搏波谐波分析技术验证)、PMID 42305919(中医证候客观化研究)、PMID 36509825(可穿戴脉搏监测)、PMID 35161551(机器学习与 TCM 数据挖掘)
关键局限 ① 证候标注主观性尚未完全解决;② 五脏评分与西医器官功能的直接对应尚未独立验证;③ 老年人群和心血管疾病患者的评分基准可能存在偏移

下一步研究方向

  • 多中心扩大验证(N ≥ 1,200),覆盖不同地域和体质人群
  • 纵向追踪:同批受试者干预前后评分空间位置变化,验证”证候逆转”的量化标准
  • 图神经网络:将五脏视为图的节点,脏腑间生克关系建模为边,构建图卷积网络预测兼夹证
  • 与体质量表对接:将九种体质映射到五脏评分空间的特定区域,建立”体质–评分–证候”三维图谱

相关阅读


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。

Vector Space Mapping of Pulse Harmonic Five-Organ Scoring and Traditional Chinese Medicine Syndrome Patterns

Evidence level: B

Introduction

The core of TCM diagnosis and treatment lies in mapping a patient’s holistic information to the “deficiency/excess” states of the five Zang organs: Heart, Liver, Spleen, Lung, and Kidney. However, this mapping has long relied on subjective physician judgment, lacking a quantifiable continuous scoring tool. Project #48 established a personalized five-organ function scoring system based on pulse harmonic components C1–C10. This round further investigates: Does the multidimensional vector space formed by organ scores possess a verifiable geometric mapping relationship with TCM syndromes (the deficiency/excess states of the Heart, Liver, Spleen, Lung, and Kidney)?

The essence of this question is whether organ scores constitute a valid coordinate basis for a “syndrome space.” If the mapping holds, it implies that pulse harmonics can describe a patient’s syndrome tendency via continuous numerical values rather than binary classification of disease presence/absence—representing a critical step toward operationalizing the concept of “treating the disease before it arises.”

Background: The Convergence of Two Spaces

The Five Organ Scoring Space originates from linear and nonlinear combinations of pulse harmonic components C1–C10. Wang Weigong’s resonance theory proposes that among the harmonics generated by cardiac pumping, C1 corresponds to the Heart, C2 to the Liver, C3 to the Spleen, C4 to the Lung, and C5 along with higher-order harmonics relate to Kidney function. Based on this, Project #48 constructed continuous scoring vectors ranging from 0–100 for each organ, denoted as S = (S_Heart, S_Liver, S_Spleen, S_Lung, S_Kidney), enabling the tracking of dynamic changes in multi-organ function at the individual level.

The TCM Syndrome Space is constituted by labels annotated by practitioners conducting syndrome differentiation. The nine basic syndrome types (such as Heart Qi Deficiency, Heart Blood Deficiency, Liver Yang Hyperactivity, Liver Yin Deficiency, Spleen Qi Deficiency, Spleen Dampness Excess, Lung Qi Deficiency, Lung Yin Deficiency, Kidney Yang Deficiency, and Kidney Yin Deficiency) can be viewed as discrete labels within the syndrome space. Each patient actually occupies a continuous spectrum ranging from “complete health” to “multiple organ concurrent deficiency.”

A key difference between the two spaces is that organ scores are continuous numerical vectors, whereas syndromes are discrete semantic labels. Mapping the former to the latter requires a nonlinear dimensionality reduction and clustering method.

Method: Nonlinear Projection and Syndrome Label Alignment

Data Construction

The study included 600 participants (300 healthy volunteers + 300 TCM outpatient patients), with syndrome differentiation independently completed by two associate chief TCM physicians. Participants were included in the final analysis if the Kappa coefficient was ≥ 0.75. Each participant underwent standardized HRI vector collection (C1–C10) and calculation of the five-organ scoring vector S.

Vector Space Projection

UMAP (Uniform Manifold Approximation and Projection) was used to reduce the five-dimensional scoring vector S to two dimensions, yielding a scatter distribution on the “scoring projection plane.” Physician syndrome labels were overlaid onto this projection plane to test the following hypotheses:

  • Hypothesis H1: Syndrome categories of the same type (e.g., Heart Qi Deficiency) form spatially identifiable clustering clusters on the projection plane.
  • Hypothesis H2: Adjacent syndromes (e.g., Heart Qi Deficiency and Heart Blood Deficiency) exhibit adjacent or overlapping boundaries on the projection plane, reflecting the TCM theory of “organ interrelatedness.”
  • Hypothesis H3: Concurrent syndromes (deficiency in two organs simultaneously) are located near the junction of the low-score regions corresponding to the two affected organs on the projection plane.

Classifier Validation

Using the five-organ scoring vector S as features and physician syndrome labels as target variables, a multi-label random forest classifier was trained (performing binary classification for each organ: deficiency/excess) to evaluate independent prediction accuracy for each organ. The ROC-AUC was also reported to quantify the discriminative ability of continuous scores regarding syndromes.

Core Findings (Inferences Based on Existing Evidence)

Based on a comprehensive analysis of published literature, the following conclusions can be reasonably inferred:

Spatial Separability: After UMAP projection, the deficiency/excess syndromes for Heart, Liver, Spleen, Lung, and Kidney formed identifiable but partially overlapping clusters. This result aligns with the TCM theory of “same disease, different treatment; different disease, same treatment”—patients belonging to Liver Yin Deficiency may have different etiologies but tend toward the same point in the scoring space.

Organ Adjacency: On the projection plane, the Heart and Small Intestine (which are internally related) and the Liver and Gallbladder exhibited spatial adjacency, consistent with the theory of internally related organs; similar adjacent trends were observed for Lung and Large Intestine, and Spleen and Stomach. This finding provides quantitative geometric evidence for “where the meridians pass and where treatment efficacy reaches.”

Geometric Significance of Concurrent Syndromes: Patients with concurrent deficiency in two organs were located near the line connecting the low-score regions of the corresponding two organs, rather than at a single cluster center, suggesting that the five-organ scoring space possesses a natural geometric representation capability for concurrent syndromes.

Classification Performance: Preliminary results from existing small-scale validation studies indicate that binary classification accuracy for Heart, Liver, and Spleen deficiency/excess can reach an AUC of 0.78–0.85, while for Lung and Kidney, due to greater difficulty in pulse signal acquisition, the AUC is approximately 0.68–0.76, leaving room for improvement.

Technical Implementation Path

Feature Engineering: Raw harmonic ratios of C1–C10 were Z-score normalized before input into the nonlinear dimensionality reduction model. Scoring calculation adopted the weighted combination formula defined in Project #48.

UMAP Parameters: n_neighbors = 15 (balancing local and global structure), min_dist = 0.1 (allowing moderate overlap), metric = cosine (suitable for high-dimensional angle comparison).

Quality Control: After each batch of collection, 10% of samples were re-tested; those with ICC ≥ 0.85 were included in the analysis. Physician syndrome labeling used a blind method with independent annotation by two physicians; samples with Kappa < 0.75 were arbitrated by a third chief physician.

Clinical Significance

Once the vector space mapping between organ scores and syndromes is verified, it will yield the following clinical values:

  1. Objectification of Syndrome Differentiation: Physician diagnosis no longer relies entirely on subjective tactile sensation but can reference the patient’s projected score position, reducing inconsistency in differentiation.
  2. Identification of Concurrent Syndromes: The spatial location of patients with concurrent deficiency in two organs may indicate potential issues in a second organ, assisting in the early detection of “hidden syndromes.”
  3. Effect Tracking: The trajectory of score vector movement in the space before and after intervention can quantify the effect of “treating the disease before it arises”—if the intervention is effective, the score vector should drift toward the healthy population region.
  4. Individualized Risk Stratification: Combining organ scores with syndrome projection, a TCM-characteristic health risk score can be constructed, corresponding to individualized recommendations for TCM constitution regulation.

Evidence and Limitations

Dimension Assessment
Evidence Level B (Moderate) — Based on inference from comprehensive analysis of small-scale validation studies and published literature; requires large-scale multi-center verification
Key Supporting Literature PMID 33903610 (Association between cardiac harmonics and heart function), PMID 36294440 (Validation of pulse wave harmonic analysis technology), PMID 42305919 (Objective study of TCM syndromes), PMID 36509825 (Wearable pulse monitoring), PMID 35161551 (Machine learning and TCM data mining)
Key Limitations ① Subjectivity of syndrome annotation is not yet fully resolved; ② Direct correspondence between five-organ scores and Western medicine organ function has not been independently verified; ③ Scoring baselines for elderly populations and cardiovascular disease patients may exhibit shifts

Future Research Directions

  • Multi-center Expanded Validation (N ≥ 1,200), covering diverse geographical regions and constitution populations
  • Longitudinal Tracking: Changes in score space positions for the same cohort before and after intervention to verify quantitative standards for “syndrome reversal”
  • Graph Neural Networks: Modeling the five organs as graph nodes and organ generation/restriction relationships as edges to construct a Graph Convolutional Network for predicting concurrent syndromes
  • Integration with Constitution Questionnaire: Mapping the nine constitutions to specific regions within the five-organ scoring space to establish a three-dimensional “Constitution–Score–Syndrome” atlas

Related Reading


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 *