A Generative Frailty Index based Explainable Approach for Hospital Readmission Prediction

A reliable and transparent prediction model for hospital readmission is essential to decrease unexpected costs and improve patient’s health. This research proposes a novel comprehensive approach that leverages a generative AI mechanism to extract Frailty Index (FI) features for predicting hospital readmission by utilizing diverse data types. Our approach calculates FI using both structured data and detailed clinical notes, ensuring a more accurate reflection of clinical assessments, and addressing the limitations of existing models regarding both explainability and data diversity. The proposed approach provides an explainable framework that aligns with real-life clinical decision-making processes, identifying key factors contributing to readmission risk. It has significant implications for healthcare professionals, researchers, and policymakers, with the potential to improve patient outcomes, reduce healthcare costs, and enhance the overall quality of care. Our findings demonstrate the value of generative AI and the use of FI in providing transparent and accurate predictions, offering valuable explanations and reasoning behind predictive modeling.