When PCOS Meets Eating Disorders: An Explainable AI Approach to Detecting the Hidden Triple Burden

Researchers have developed small, open-source language models to detect the co-occurrence of body image distress, disordered eating, and metabolic challenges in women with polycystic ovary syndrome (PCOS) through social media analysis. The study utilized 1,000 PCOS-related posts from six subreddits, achieving a 75.3% accuracy rate with robust explainability in identifying these interconnected issues.

This advancement in natural language processing represents a significant step towards understanding and addressing the hidden triple burden faced by women with PCOS, potentially leading to improved support and interventions for this population.