Emergency Drug Procurement Planning Based on Big-Data Driven Morbidity Prediction

Abstract

Due to the uncertainty of diseases, traditional approaches of drug procurement planning in hospitals often cause drug overstocking or understocking, which can have strong negative effects on healthcare services. This paper proposes a big-data driven approach, which uses a deep neural network to predict morbidities of acute gastrointestinal infections based on a huge amount of environmental data, and then constructs an optimization problem of drug procurement planning for maximizing the expected therapeutic effect on the predicted cases. The problem is solved by an efficient heuristic optimization algorithm. Computational experiments demonstrate the performance advantages of both the deep learning model and the heuristic algorithm over existing ones, and two real case studies in Central China show that the average prediction error of our approach is only 8% and the estimated recovery rate reaches 99%, much better than the currently used method. Our approach can also be extended for many other medical resource planning problems.