EVALUATION OF THE ROLE OF AI IN CLINICAL DECISION SUPPORT FOR THE DETECTION OF FETAL ANOMALIES VIA ULTRASOUND

Abstract

Artificial intelligence (AI) has been described as a promising tool for improving prenatal ultrasound, particularly in the detection of congenital anomalies. Recent literature shows that deep learning algorithms outperform conventional ultrasound examinations in terms of sensitivity, specificity, and accuracy, with notable applications in screening for congenital heart disease and identifying central nervous system anomalies. The effectiveness of this approach is exemplified by systems such as HeartAssist, which report accuracy rates exceeding 98% in fetal cardiac assessment, consolidating AI as a highly accurate support tool for gestational monitoring. However, considerable limitations persist, such as heterogeneity of training databases, gaps in multicenter validation, and concerns regarding ethics and data protection. In conclusion, AI represents a resource with significant potential for standardization and expansion of intrauterine diagnosis, operating as a complementary instrument to the specialist's expertise and contributing to the optimization of perinatal outcomes.