J Korean Med Assoc.  2025 Mar;68(3):169-174. 10.5124/jkma.2025.68.3.169.

Applications and challenges of artificial intelligence in obstetrics and gynecology: a narrative review

• Lee YK

• KMID: 2566571
• DOI: http://doi.org/10.5124/jkma.2025.68.3.169

Abstract

Purpose
The integration of artificial intelligence (AI), especially deep learning techniques, is revolutionizing obstetrics and gynecology. AI algorithms can analyze complex medical data, including imaging and genomics, to manage high-risk conditions such as placenta accreta spectrum (PAS) and to improve the outcomes of assisted reproductive technology. Despite significant advancements, challenges persist, including issues with data standardization, ethical concerns, and the complexity of implementing deep learning models.
Current Concepts
Deep learning has demonstrated high sensitivity and specificity in diagnosing conditions like PAS and fetal cardiac anomalies. It optimizes embryo selection, tailors hormonal therapy, and predicts preterm birth risks using advanced tools such as CADXpert. Additionally, radiomics combined with deep learning enhances magnetic resonance imaging-based diagnosis and prognosis, particularly in endometrial cancer. However, barriers such as limited model interpretability and regulatory compliance continue to restrict widespread implementation.
Discussion and Conclusion
A multidisciplinary approach involving clinicians, data scientists, and policymakers is crucial to addressing these challenges. Efforts should concentrate on developing standardized datasets, creating explainable models, and establishing robust educational frameworks. Overcoming these barriers will enable precise diagnoses and personalized treatments, ultimately improving patient care and outcomes in obstetrics and gynecology.

Keyword

Artificial intelligence; Obstetrics; Gynecology; Diagnosis; Therapeutics; Deep learning; 인공지능; 산과; 부인과; 진단; 치료; 딥러닝