Predictive Model for Hospital Readmission of Diabetic Patients

Abstract

Diabetes is recognized as one of the world's most prevalent health problems. As diabetic patients grew, so did the percentage of diabetic hospital readmissions. Early readmissions can impact patient well-being, operational efficiency, and financial burden. This study uses machine learning approaches to predict hospital readmissions among diabetes patients. Data was collected from 130 US hospitals. CRISP-DM is used for analysis. Logistic regression (LR) and random forest (RF) classifiers were implemented. The classifier performance was compared. Random Forest outperformed the other model, with an accuracy of 0.89. The model was chosen to enable practical deployments. Researchers used a web-based interface to get data and receive real-time predictions. The results showed that the predictive model used alongside an interface creates a clear and understandable prediction platform. However, the research might involve various datasets and Deep Learning to improve models and findings, in future studies. Furthermore, the model could explore the integration of machine learning interpretability approaches to increase transparency and promote better comprehension of the model's predictions by healthcare practitioners.