Abstract
BACKGROUND: Diabetic retinopathy (DR), one of the most common complications of diabetes, severely impacts patients' quality of life. The combined use of the traditional Chinese medicines Astragalus, Fructus ligustris, and Cornus officinalis has yielded considerable therapeutic effects in clinical DR treatment. METHODS: In this study, a multimodule framework (BNM) encompassing bioinformatics, network pharmacology, and machine learning (ML) based on molecular fingerprints was innovatively developed to thoroughly investigate the molecular mechanisms of this Chinese medicine in treating DR. RESULTS: A total of 40 active components and 12 core targets were identified. Enrichment analysis identified key pathways such as VEGF signaling pathway, TNF signaling pathway and HIF-1 signaling pathway. Prediction models using key targets, such as PPARG, were constructed from the GEO database and validated via immune infiltration analysis and molecular docking, revealing that PPARG may be a potential target for DR treatment. Moreover, the core component of this Chinese medicine, stigmasterol, was identified using a ML model based on molecular fingerprints. In vivo experiments demonstrated that stigmasterol can regulate glucose and lipid metabolism, improve systemic inflammatory levels, and ameliorate ocular vascular changes in DR by modulating the expression of PPARG. CONCLUSION: The BNM framework suggests that PPARG may be an important target for stigmasterol in the treatment of DR, with its mechanism potentially related to the VEGF/VEGFR pathway.