Hospital Readmission Risk Mitigation

Hospital readmission is an essential metric for healthcare executives to consider when evaluating the effectiveness of their care. The early identification of patients at higher risk of readmission enables the hospital to allocate resources more effectively and implement an early intervention plan to reduce readmission risk. This project investigated 30 day hospital readmissions using a large clinical dataset containing demographic variables, insurance type, medications, procedures, and hospitalization characteristics. The dataset was subjected to exploratory analysis, followed by various modeling techniques including logistic regression, neural network, and elastic net. The final model used was LASSO, as the area under the curve was similar across methods, and sensitivity (correctly classifying readmissions) was highest. From this model, it was possible to determine which predictor variables are most closely tied to readmission, which can serve as a tool in assisting hospital staff in managing high-risk patients.