Optimizing Healthcare Resources for Chronic Non-Communicable Disease Patients: A Data-Driven Approach

Currently, hospitals are often overcrowded, and patients with more severe conditions (e.g. cardiovascular diseases and diabetes) do not receive timely treatment. This is because traditional resource allocation methods suffer from some shortcomings, one of which is the inappropriate illness prediction leading to inadequate look-ahead resource allocation with weak response capability for varying medical demand. As a result, the capacity cannot perfectly meet the demand, i.e., either a shortage of medical resources due to inadequate advance allocation or excessive surplus due to over-allocation. To handle this, we propose a proper prediction method serving illness monitoring or warning for different patient populations, facilitated with various customized features to strengthen the prediction accuracy. On this basis, we construct a medical resource allocation system that combines prediction and optimization. More precisely, a mixed integer programming model is formulated to determine which patients should be accepted for medical treatment under limited medical resources, aiming at prioritizing the severe patients. Numerical experiments are conducted to demonstrate our approach and validate the contributions.