An up-to-date, evidence-based section for healthcare professionals. It examines applications of artificial intelligence in obstetrics and perinatology—from fetal imaging to risk prediction, and from large language models to ethics and regulation—through a candid appraisal of the evidence.
⚕️ Target audience: This section is intended for physicians and healthcare professionals for educational and current-information purposes. The AI tools discussed provide clinical decision support and do not replace clinician responsibility. The content is kept balanced; alongside positive results, null/negative findings and limitations are deliberately emphasized.
Machine learning, deep learning and large language models; obstetric practice map, opportunities, limitations, and reporting standards (TRIPOD-AI).
Automated biometrics, standard plane detection, anomaly and cardiac screening support, FDA-approved tools and ISUOG perspective; limits of generalization.
Computerized CTG, deep learning classification, and null result of the INFANT study; observer variability and still open questions.
FMF competing-risks model, 1st trimester combined screening, machine learning models and external validation/transferability issue.
Cervical imaging, electrohysterography, and multivariable models; Honest assessment of the reality of data imbalance and poor external validation.
Time-lapse deep learning, automatic blastocyst grading, iDAScore; and honest presentation of RCTs that did not show superiority in live birth rate.
Clinical decision support, patient communication, documentation and ambient AI; hallucination, evidence-timeliness, confidentiality (KVKK) and bias risks.
Algorithmic bias, external/prospective validation, explainability; FDA/CE framework, TRIPOD-AI/CONSORT-AI and FIGO/ISUOG/WHO stances.
Last updated: .