Predicting Impact on Re-Admission Rates for Patients Hospitalized with Diabetes

For chronic diseases like diabetes, one of the most important issues facing the healthcare industry is estimating the risk of readmission to the hospital. Hospitals can lower the readmission rate by providing unique care for patients who are found to have re-admissions during the primary stage. Our effort involves the use of machine learning to create a new model. Predicting re-admissions in their early stages enables physicians to give patients who are at high risk of readmission significant attention, which in turn optimizes the healthcare system and reduces healthcare spending. Re-admissions also have a negative impact on hospitals’ reputations. Greater prediction accuracy than current methods is made possible by machine learning. In this study, we use logistic regression, decision trees, and XGBoost classifiers to attempt to predict the hospital readmission rate of diabetes patients. When applied and assessed against real-world data, a combination of data engineering and machine learning was proven to perform better than conventional machine learning techniques. We accomplish this using a variety of modules, one of which is feature extraction, which enhances the analytical process and aids in the creation of an effective prediction summary.