AI Subtyping Suggests Polycystic Ovarian Syndrome Is Four Distinct Conditions

Polycystic metabolic ovarian syndrome (PMOS), the most common endocrine disorder among women of reproductive age, affecting an estimated 8 to 12 percent of that population, may be far less uniform than its diagnostic label suggests. A review published in Molecular Biology Reports argues that traditional symptom-based diagnostic criteria do not fully capture the underlying biological diversity of the condition, and that data-driven analysis points to a syndrome composed of distinct subtypes rather than a single disease. The condition is also associated with elevated risks of infertility, type 2 diabetes, and heart disease.

The review, by Sristy Pradhan and Abilash Valsala Gopalakrishnan, synthesizes recent work combining multi-omics profiling with artificial intelligence. When clinical variables are fed into data-driven clustering algorithms, the authors report, four major subtypes consistently emerge: HA-PCOS, OB-PCOS, SHBG-PCOS, and LH-PCOS. Each subtype carries its own hormonal, metabolic, and genetic signature, including distinct patterns of lipid, amino acid, and energy metabolism.

The subtypes also differ in immune activity. The review highlights immune system disruption and impaired breakdown of blood clots within the ovarian environment as notable features of the condition, adding to the picture of PMOS as a systemic metabolic disorder rather than a purely reproductive one.

Among the more tentative findings is a possible role for lactate metabolism. The authors suggest that altered lactate metabolism and changes in histone lactylation may contribute to granulosa cell dysfunction in the ovary. They note, however, that further studies are needed before these mechanisms can be considered established.

On the diagnostic side, the review reports that machine learning models integrating clinical, metabolic, and genetic data have achieved high diagnostic accuracy, with the added advantage of producing interpretable results rather than opaque predictions. The authors also propose potential subtype-targeted therapies, including GLP-1 receptor agonists, SGLT2 inhibitors, N-acetylcysteine, and NK3R antagonists. These remain proposals: none has yet been validated in clinical trials for PMOS subtypes, and the review does not claim efficacy for any of them.

Significant limitations temper the promise. The studies synthesized in the review rely on small and relatively homogeneous cohorts, long-term data are limited, and barriers to care persist in low-resource settings. Even so, the review’s central claim is a reframing with practical consequences: if PMOS is not one disorder but several, then AI-powered subtyping could point toward more personalized and more equitable management, provided the subtypes hold up in larger and more diverse cohorts.

Subject of Research: Multi-omics and AI subtyping of polycystic metabolic ovarian syndrome heterogeneity

Article Title: Dissecting polycystic metabolic ovarian syndrome heterogeneity (PMOS): multi-omics, AI subtyping, and precision management

Article References: Pradhan, S., & Gopalakrishnan, A. V. (2026). Dissecting polycystic metabolic ovarian syndrome heterogeneity (PMOS): multi-omics, AI subtyping, and precision management. Molecular Biology Reports, 53(1), Article 1788. https://doi.org/10.1007/s11033-026-12967-5

Image Credits: AI Generated

DOI: 10.1007/s11033-026-12967-5

Keywords: PMOS, PCOS, multi-omics, machine learning, endocrinology, subtypes, granulosa cells, histone lactylation, precision medicine, Molecular Biology Reports, Dissecting, polycystic