Artificial intelligence in ultrasound imaging for polycystic ovary syndrome (PCOS): methodological advances, bias, and clinical translation

Polyendocrine metabolic ovarian syndrome (PMOS), formerly termed polycystic ovary syndrome (PCOS), is a heterogeneous endocrine and metabolic disorder with reproductive and metabolic manifestations. Ultrasound is the primary imaging modality for assessing polycystic ovarian morphology (PCOM) within the historical PCOS diagnostic framework, but its interpretation is affected by operator dependence, interobserver variability, image quality, and subjective assessment. Artificial intelligence (AI), particularly machine learning (ML) and deep learning (DL), has increasingly been applied to ultrasound-based PCOS diagnosis, including automated image analysis, follicle segmentation, feature extraction, and diagnostic classification. This review synthesises evidence from 65 studies published between 2017 and 2026, identified through a structured search of major biomedical and engineering databases. Methodological quality and risk of bias were assessed using the Prediction model Risk of Bias Assessment Tool (PROBAST) in 53 eligible prediction-model studies, with substantial risk-of-bias concerns identified particularly in the analysis domain; results were visualised using ROBVIS. This review focuses specifically on AI-assisted ultrasound imaging, including multimodal studies only when ultrasound formed an integral component of the diagnostic framework. Reported diagnostic performance was generally high across individual studies, but direct comparison was limited by heterogeneity in datasets, imaging protocols, outcome definitions, preprocessing, model development, and validation strategies. Frequent reliance on retrospective, single-centre datasets, limited external validation, and inconsistent reporting constrain confidence in the generalisability and clinical applicability of existing AI models. Additional barriers to clinical translation include interpretability, data governance, regulatory requirements, and workflow integration. Future research should prioritise standardised reporting, multicentre datasets representing diverse populations, prospective external validation, transparent model development, and regulatory-compliant evaluation to establish the reliability and clinical applicability of AI-assisted ultrasound systems for PCOS diagnosis.