Abstract
Study on the Latent Structure and Rules of Compatibility and Modification in Fan Qu''s Treatment of Polycystic Ovary Syndrome Based on Data-Driven Models
DOI:
En KeyWords: data mining  polycystic ovary syndrome  machine learning  medication rules  latent structure
Fund Project:国家自然科学4项(82575119;82274564;82074476;81874480);浙江省自然科学基金重点项目2项(LZ26H270001;LZ21H270001)。
Author:Zeng Wenshan
View Counts:
PDF Download Counts:
En Abstract:
      Objective: To explore the latent core formulas and the rules of compatibility and modification in Professor Fan Qu''s treatment of polycystic ovary syndrome (PCOS) based on data-driven models. Methods: A retrospective collection was conducted on the traditional Chinese medicine (TCM) prescriptions for PCOS treated by Professor Fan Qu. Non-negative Matrix Factorization (NMF) was employed to extract the latent core formulas, combined with the Louvain complex network community detection algorithm to partition functional modules of targeted medication for cross-validation. Results: The NMF extracted 4 latent core formulas, and the Louvain algorithm identified 4 functional modules, both of which consistently corresponded to four major therapeutic effects: tonifying the kidney and strengthening the spleen, invigorating blood and resolving stasis, cooling blood and stopping bleeding, and soothing the liver and relieving depression. The mapping network further uncovered a cross-module foundational formula centered on Taizishen (Radix Pseudostellariae) and Fupenzi (Fructus Rubi). Conclusion: This study reveals Professor Fan Qu''s clinical strategy for treating PCOS, characterized by a foundational approach of tonifying the kidney and strengthening the spleen, combined with flexible modifications aimed at invigorating blood, stopping bleeding, and soothing the liver. The integrative application of two machine learning algorithms provides a novel methodological framework for inheriting TCM clinical experience.
View Fulltext   Download PDF Viewer
Close