AI-Driven Business Analytics for Early Prediction and Prevention of High-Cost Healthcare Utilization

The readmissions to the hospital within 30 days following the discharge are an important indicator of care quality and one of the key elements in increasing healthcare spending, especially in patients with chronic illnesses, such as diabetes mellitus. The readmission rates of diabetic patients are also much higher than the average rate of the hospitalized patients because of the complicated comorbidities, unstable glycemic conditions, difficulties in managing medications, and extensive healthcare use. This study is set to pinpoint the most important indicators related to 30-day readmissions among diabetic patients and to create an Artificial Intelligence (AI)-driven predictive model that will be able to identify the risks early. This study demographic features, clinical indicators, laboratory outcomes, medication history, and previous healthcare use measures using the Diabetes 130-US hospitals for years 19992008 dataset with more than 100 000 hospital encounters. Following thorough data preprocessing and feature engineering, several machine learning models, namely, Logistic Regression, Random Forest, Support Vector Machine, and Gradient Boosting are implemented and assessed in terms of performance measures, such as ROC-AUC and precision, recall, and F1-score to solve the problem of class imbalance. The results emphasize that the number of previous inpatient visits, the number of diagnoses, emergency visits, insulin usage, discharge disposition and change of medications are some of the most significant predictors of early readmission. AI-based models have a high predictive accuracy and give practical information to health practitioners. Situated in the business analytics perspective, it is possible to identify high-risk patients early on and focus on their specific needs, including better discharge planning, medication reconciliation, and subsequent follow-ups. The interventions can decrease avoidable readmissions, better patient outcome, hospital performance measures, and health care costs can be greatly decreased. This study adds to the increased amount of research in the field of AI-based healthcare analytics and helps to make decisions on the quality improvement and cost optimization of the hospital system based on data.