AI-driven Early Detection Warning System For Obstetric Emergencies In LMICs

AI-driven Early Detection Warning System For Obstetric Emergencies In LMICs

Tinatin Kutchukhidze
New Vision University, Tbilisi, Georgia

Introduction:Postpartum hemorrhage (PPH) and eclampsia remain the leading causes of maternal mortality worldwide, with 30-50% of maternal deaths in the sub-Saharan region, where timely diagnosis and intervention are challenging. Artificial intelligence offers potential to enhance early detection of high-risk patients by analysing clinical, diagnostic, and Vital sign data in real time.

Aim: Our aim is to evaluate the effectiveness of an AI-based early warning system in predicting and preventing maternal mortality from postpartum hemorrhage in LMICs by assessing diagnostic accuracy and, impact on clinical decision-making.

Methods: We conducted a prospective, multicenter, observational study across 3 district-level sites in rural Sub-Saharan Africa from January to December 2024. Each site served a catchment area with limited specialist obstetric support. An AI algorithm was developed using a retrospective training dataset of 1000 anonymised obstetric records from multiple LMIC hospitals, and predictive features included maternal vital signs, laboratory values, obstetric history, and intrapartum data. The system was generating the real-time risk scores for eclampsia every 15-minute interval, and clinical staff were receiving the automatic alerts.

Results: The primary functions of the system identified a sensitivity of 93% specificity of 85% compared to standard assessments with 74% and 79% ( p< 0.01) of sensitivity and specificity, respectively.

Conclusion: An AI-driven early warning system can significantly improve the timely detection of PHH and eclampsia in LMIC Settings, and its integration can strengthen the obstetric safety and reduce preventable maternal death.

Keywords: Postpartum hemorrhage; Eclampsia; Maternal mortality; Artificial intelligence; Early warning system; Predictive analytics; Low- and middle-income countries (LMICs); Diagnostic accuracy; Real-time monitoring; Obstetric safety; Clinical decision support; Sub-Saharan Africa