Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Machine Learning and Deep Learning for Diagnosis of Polycystic Ovary Syndrome: Systematic Review and Meta-Analysis
ABSTRACT
Background:
Polycystic ovary syndrome (PCOS) is a prevalent endocrine disorder affecting reproductive-aged women worldwide, yet its diagnosis remains challenging due to heterogeneous clinical presentation and subjective assessment criteria. While machine learning (ML) and deep learning (DL) models have shown promising results in automating and improving PCOS diagnosis, a comprehensive quantitative synthesis of their diagnostic performance is still lacking.
Objective:
To evaluate the overall diagnostic accuracy of ML and DL models for PCOS and identify factors influencing their performance.
Methods:
This study was conducted according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. A systematic search was performed in PubMed, Web of Science, Embase, and the Cochrane Library from inception to August 7, 2025. Studies evaluating the diagnostic value of DL or ML algorithms for PCOS were included, while those with duplicate or unavailable data were excluded. The Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) tool was used to evaluate the risk of bias in each study. Data synthesis and statistical analysis were performed utilizing the MIDAS and METAPROP packages in Stata.
Results:
Forty-five studies comprising 49 datasets were included. The pooled sensitivity and specificity of ML or DL models for diagnosing PCOS were 0.93 (95% CI: 0.90-0.95; I² = 96.77%) and 0.95 (95% CI: 0.93-0.97; I² = 96.28%), respectively. The area under the SROC curve was 0.98 (95% CI: 0.97-0.99). Subgroup analysis revealed that models utilizing DL (sensitivity: 0.98, specificity: 0.97), those trained on large databases, and those using ultrasound image features (sensitivity and specificity both: 0.99) demonstrated superior performance.
Conclusions:
ML or DL models show promise for the diagnosis of PCOS, achieving high summary estimates of sensitivity and specificity. However, the substantial heterogeneity among studies and the scarcity of external validation significantly constrain the certainty of these findings and their generalizability to diverse clinical settings. Future research must prioritize large-scale, multi-center studies with robust external validation to confirm the real-world applicability and facilitate the clinical translation of these models. Clinical Trial: PROSPERO CRD420251137107; https://www.crd.york.ac.uk/PROSPERO/view/CRD420251137107
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.