Volume 17, Number 3
A Prescriptive Analytics Framework for Risk-Integrated Maternal Healthcare Resource Allocation in Zimbabwe
Authors
Ruramai Judith Yotamu, Harare Institute of Technology, Zimbabwe
Abstract
This paper presents a Prescriptive Analytics Framework for Maternal Health (PAMH) for risk-integrated maternal healthcare resource allocation in Zimbabwe. The framework combines machine-learning risk prediction with a mixed-integer linear programming optimisation model to allocate midwives, delivery kits and ambulances across 84 facilities under six budget scenarios. Logistic Regression, Random Forest and Gradient Boosting models were trained on 5,001 patient encounters, with Gradient Boosting achieving the strongest predictive performance (test ROC-AUC 0.913). Facility-level risk scores were embedded as priority weights in the optimisation objective, enabling risk-sensitive allocation under budget and equity constraints. Baseline optimisation achieved 97.9% budget utilisation, while the austerity scenario showed a 157% rise in weighted unmet demand. A six-page decision support dashboard translates the framework into actionable intelligence for district health officers.
Keywords
Prescriptive analytics, MILP, Maternal health, Machine learning, Resource allocation