Application of artificial intelligence in diagnosis and management of fetal growth disorders: a comprehensive review

TL;DR

This paper reviews how artificial intelligence can improve the diagnosis and management of fetal growth disorders, offering more accurate and efficient prenatal care.

Contribution

The paper provides a comprehensive review of AI applications in fetal growth disorders, highlighting their potential to enhance diagnostic accuracy and accessibility.

Findings

AI models combining maternal, fetal, and imaging data match clinician accuracy while improving efficiency.

AI applications like automated biometry and deep learning on ultrasound scans improve diagnostic precision.

AI has potential to expand access to prenatal care in low-resource settings.

Abstract

Fetal growth disorders, including both fetal growth restriction and macrosomia, remain major contributors to perinatal morbidity and long-term health risks in adulthood. While ultrasound is the most frequently employed technique for the diagnosis of intrauterine growth abnormalities, its efficacy is constrained by the operator’s experience and variable accuracy. This review explores the role of artificial intelligence (AI) in advancing the detection and management of fetal growth disorders. We conducted a comprehensive literature search of major databases to identify original and review articles addressing the use of AI in fetal growth restriction, small-for-gestational-age and large-for-gestational-age fetuses, as well as fetal macrosomia. The available evidence indicates that AI models combining maternal, fetal, and imaging data exhibit a level of accuracy comparable to that of experienced clinicians, while also enhancing operational efficiency and reducing variability. Emerging applications include automated biometry, prediction models based on biomarkers and Doppler indices, as well as deep learning algorithms applied directly to ultrasound scans. These methods not only enhance diagnostic precision but also expand access to high-quality prenatal care, particularly in low-resource settings. Nonetheless, most of the published studies remain limited by retrospective designs, small sample sizes, and a lack of external validation. Addressing these challenges, along with ethical, technical, and regulatory considerations, will be essential for clinical translation. In conclusion, AI has the potential to become a cornerstone of precision perinatal medicine by enabling earlier diagnosis, individualized monitoring, and thus improved outcomes for both mothers and infants.

Figures