Artificial Intelligence, Obstetricians, and Midwives in Non-Stress Test Interpretation

This prospective observational study aims to evaluate and compare the interpretation of antenatal non-stress tests (NSTs) by three artificial intelligence models, three obstetricians, and three midwives. A total of 300 NST recordings obtained during routine antenatal care will be collected and prepared in PDF format. The recordings will be distributed to the evaluators, who will independently assess NST reactivity and recommend appropriate clinical management within two weeks. Two perinatologists will independently evaluate the NST recordings to establish the reference standard. In cases of disagreement, a third perinatologist will be consulted to determine the final reference assessment. The study will compare the diagnostic performance, agreement with the reference standard, and inter-rater agreement among artificial intelligence models, obstetricians, and midwives.

Inclusion Criteria: * Gestational age of 34+0 weeks or greater * Singleton pregnancy undergoing routine antenatal follow-up * Availability of an interpretable NST recording with a minimum duration of 20 minutes * Complete clinical and demographic data * Provision of written informed consent Exclusion Criteria: * Multiple pregnancy * Major fetal congenital anomaly * Technically uninterpretable NST recording * NST recording shorter than 20 minutes * Incomplete clinical data * Refusal to participate