Mitigating Socioeconomic Disparities in High-Acuity Maternal Care Shortages Across US Healthcare Systems

• Abbas Ahsun

  Texas University

  Author

The United States faces a persistent maternal health crisis characterized by rising severe maternal morbidity and mortality rates, with approximately 36% of counties designated as maternity care deserts and profound racial and socioeconomic disparities in outcomes. While early warning systems and clinical decision support tools have demonstrated promise in obstetric care, existing approaches fail to address the operational gap between risk identification and resource mobilization, particularly in resource-constrained settings. This study presents the design, implementation, and validation of an edge-computing artificial intelligence architecture for real-time obstetric resource triaging that integrates continuous physiological monitoring, machine learning-based risk stratification, and a resource allocation engine to optimize the deployment of scarce maternal care resources. Using a hybrid methodology combining retrospective analysis of de-identified clinical data and prospective simulation across three US healthcare system archetypes (urban tertiary, rural community, and safety-net hospital), the proposed framework achieved 89.4% accuracy in predicting high-acuity obstetric events requiring immediate intervention, with a 73% reduction in median triage-to-intervention time compared to conventional nurse-led triage protocols. The system demonstrated equitable performance across socioeconomic strata, with no statistically significant disparity in alert-to-action times between patients from high-resource and low-resource catchment areas (p = 0.31). This research contributes a replicable, privacy-preserving architectural blueprint for AI-supported obstetric triage that addresses both technical and equity dimensions, offering a pragmatic pathway for mitigating maternal care shortages through intelligent resource orchestration.

• PDF

Copyright (c) 2026 Abbas Ahsun (Author)

This work is licensed under a Creative Commons Attribution 4.0 International License.