Optimizing Health Supply Chains with Decision-Aware Machine Learning

Abstract

A key challenge facing healthcare systems in Low-and Middle-Income is the difficulty allocating scarce resources to healthcare facilities. This problem is complicated by the limited availability of high-quality data, making it hard to apply traditional data-driven techniques. We propose a novel machine learning framework for essential medicines allocation, which leverages a combination of multi-task learning and decision-aware learning to improve sample efficiency. In collaboration with the Sierra Leone national government, our framework has been deployed in Sierra Leone as a decision support tool to help reduce waste and improve essential medicines allocation. Our evaluation based on synthetic difference-indifferences suggests that our framework has increased consumption of essential medicines by 23%, thereby reducing waste and improving access to medicines and medical supplies for approximately 3.7 million women and children under five. In addition, we provide experimental evidence that our approach outperforms several baselines and ablations. Our work demonstrates the real-world impact and promise of machine learning to improve the efficiency of high-stakes decision-making problems in budget-constrained settings.

Keywords: Machine Learning, Healthcare Operations, Global Health, Decision-Aware Learning