Diagnosis of Polycystic Ovary Syndrome With Predictive Modeling of Select Clinical Features - PubMed
5 hours ago
- #PCOS
- #Women's Health
- #Diagnostic Modeling
- The study aimed to predict PCOS diagnosis using a limited set of ultrasonographic, biochemical, and clinical features.
- Participants included 101 with PCOS and 50 controls, diagnosed per 2023 International Evidence-Based Guideline.
- Key features analyzed: demographic, ultrasonographic (ovarian volume, follicle count), biochemical (testosterone, AMH), and clinical (menstrual cycle length, hirsutism score).
- AMH alone showed good diagnostic accuracy (AUROC 0.884, F1 score 0.807).
- Combining AMH and ovarian volume improved performance (AUROC 0.906, F1 score 0.811).
- A model with all features achieved excellent accuracy (AUROC 0.991, F1 score 0.811).
- A refined model using AMH, ovarian volume, hirsutism score, and maximum menstrual cycle length performed strongly (AUROC 0.982, F1 score 0.805).
- Conclusion: A minimal combination of ovarian volume, AMH, and clinical history can accurately predict PCOS, streamlining diagnosis.