AI in Hypertensive Disorders of Pregnancy: Review

Area of Science:

• Obstetrics and Gynecology
• Medical Informatics
• Computational Biology

Background:

• Hypertensive disorders of pregnancy (HDP) are a major cause of maternal and fetal mortality globally.
• Early detection and risk stratification are crucial for preventing severe HDP complications.
• Traditional clinical methods often lack the precision for effective risk identification.

Purpose of the Study:

• To systematically review machine learning (ML) applications in hypertensive disorders of pregnancy (HDP).
• To identify trends, methodologies, and research gaps in ML for HDP.
• To guide future research for improved maternal and fetal outcomes in HDP.

Main Methods:

• Scoping review adhering to PRISMA-ScR guidelines.
• Comprehensive search of three databases for English-language publications on ML models in HDP.
• Data extraction using the CHARMS checklist, summarizing study characteristics, outcomes, and ML methods.

Main Results:

• Most studies focused on preeclampsia (75.27%), with limited representation of other HDP phenotypes.
• Machine learning applications for diagnosis/onset detection (62.37%) and risk prediction (27.95%) have increased since 2014.
• Geographic concentration in China and North America; underutilization of omics and imaging data; limited reporting on intervention timing and readmission outcomes.

Conclusions:

• Machine learning shows significant growth in HDP diagnosis and risk prediction.
• Geographic disparities, limited phenotype representation, and underrepresentation of intervention timing models are key barriers.
• Future research should focus on diverse datasets and models to improve intervention timing and risk profiling for HDP.