基于知识图谱与机器学习的FGIDs可解释性证候演化模型研究

The complexity and abstraction of syndrome theory has become an obstacle to the modernization of TCM research and practice. For the individual syndrome differentiation and systematic evaluation of disease, it is essential to establish the bridge of interpretation between the biological mechanism of disease and theoretical definition of TCM syndromes for the promotion of TCM research and practice in a modernization paradigm. This study proposes an approach for multi-source heterogeneous data integration and modeling with application of knowledge graph and machine learning methods on basis of the syndrome differentiation theoretical logic. With multi-scale evidences such as etiological and pathogenesis characteristics, syndrome distribution characteristics, pathological mechanism and pharmacological network of functional gastrointestinal disorders integrated in the knowledge graph, the interpretation framework for syndrome model would be established and further used for strengthening the interpretability of dynamic evolution model of syndromes that estimated with interpretable machine learning methods. Through the integration and application of multidisciplinary approaches, this study provides a new methodological example for the modern application of TCM theories.

中医证候理论的高维复杂性及逻辑抽象化特征是中医规范化实践的重要阻碍。如何结合疾病的系统化认知，建立中医证候理论与生物学基础机制的深层桥接，实现辨证逻辑的系统化阐释是推进中医理论现代化发展的关键问题。因此，本研究扎根于辨证论治的核心特色，利用知识图谱与机器学习在多源异构数据整合与复杂逻辑提取建模方面的优势。通过FGIDs病因病机特点、证候分布特征、生物学基础机制及优势方药作用网络等宏微观证据整合，形成FGIDs证候辨识的系统化解释框架。同时，以核心病机演化路径和分子功能网络为线索，引入可解释性机器学习进行证候辨识逻辑和动态演化特征的解析，建立FGIDs证候演化模型及其宏微观解释体系。通过多学科、跨领域技术的融合应用，探索中医证候宏微观解释体系的构建方法，为中医理论的现代化阐释与应用转化提供新的方法学范例。