Personalized metformin treatment in PCOS: Integrating Artificial intelligenceand Biomarkers

Abstract

Polycystic ovary syndrome (PCOS) is a heterogeneous endocrine disorder associated with reproductive dysfunction, insulin
resistance, metabolic syndrome, and increased long term cardiovascular risk. Metformin remains a cornerstone therapy in PCOS
management because of its beneficial effects on insulin sensitivity, ovu lation, menstrual regularity, and metabolic parameters.
However, substantial inter individual variability in therapeutic response limits its effectiveness, highlighting the need for precision
medicine approaches. This narrative review synthesizes current e vidence regarding biomarkers and artificial intelligence (AI) based
strategies for personalizing metformin therapy in PCOS. Key metabolic biomarkers such as HOMA IR, fasting insulin, and HbA1c,
alongside hormonal markers including anti Müllerian hormone (A MH), testosterone, luteinizing hormone/follicle stimulating
hormone ratio, and sex hormone binding globulin, demonstrate potential in predicting treatment response. Inflammatory biomarkers and emerging molecular markers, particularly microRNAs and pharmaco genomic variants, may further refine patient stratification.
The review also evaluates the growing role of AI and machine learning in PCOS diagnosis, risk stratification, and treatment
prediction. Machine learning models integrating clinical, biochemical, genomic, and lifestyle data could identify likely responders to
metformin before treatment initiation, thereby reducing empirical prescribing and adverse effects. A conceptual AI assisted clinical
decision support framework is proposed to guide future tran slational research and clinical implementation. Despite promising
advances, challenges related to dataset quality, model interpretability, algorithm bias, and external validation remain signi ficant
barriers. Future progress will require multi center collab orations, standardized phenotyping, explainable AI models, and integration of multi omics and digital health technologies to enable truly personalized metformin therapy in PCOS.