Identifying nutraceutical targets to treat polycystic ovary syndrome using graph representation learning.

Abstract

Polycystic ovary syndrome (PCOS) is a complex, multifactorial, and polygenic disorder. Here, we employed machine learning (ML) techniques to analyze large open-source datasets to identify bioactive molecules in foods and pharmacological agents that interact with genes and biological functions central to PCOS pathophysiology. We selected 13 PCOS-associated genes as targets, and the network propagation algorithm systematically identified bioactive molecules that interact with pathways relevant to PCOS. Among the top-ranked molecules, epicatechin-3-gallate (found in green tea) and 24-methylenecycloartan-3-ol (found in almonds) were newly identified, with green tea and almonds previously demonstrated to have anti-androgenic and anti-inflammatory properties. Validation of the ML pipeline with clinically available drugs revealed significant interactions with gonadotropin-releasing hormone receptor modulators, consistent with their established role in PCOS pathophysiology. These findings identify novel therapeutic targets for further research in precision nutrition and drug repurposing for PCOS treatment.