Prediction of Diabetic Patients with High Risk of Readmission using Smart Decision Support Framework

Abstract

Patients with diabetes are more likely to be readmitted to the hospital than those who are nondiabetic. The earlier patients with a strong probability of readmission are monitored and cared for, the better. The goal of this research is to develop a decision - making framework that can identify diabetes patients who are at risk of early readmission. Many data analysis approaches have been employed to perform this. Computer vision is used to create a novel model in this study. Individuals at high risk of complications to be readmitted are prioritized in the early stages, which in turn reduces healthcare costs and improves the reputation of the hospital, thus enhancing the health service and saving money. Predictions made using machine learning are more accurate than those made using traditional methods. In this study, patients' hospital readmissions may be predicted by utilizing a standard scaler, a decision tree, and random forests for classification, CATboost for categorical features, and XGBoost classifiers. When applied to real-world data, a machine learning method that incorporates deep learning technique has outperformed the other methods. As a response to a number of modules, including extracting features, the analysis has been enhanced and a more useful framework has been created.