AI-Assisted Clinical Decision Support Systems for Rehabilitation and Chronic Disease Management

The global rise in chronic diseases and the complexity of rehabilitation protocols necessitate a shift from reactive to proactive, personalized clinical decision-making. Traditional models for prognosis often lack the power to handle the high-dimensional, time-series nature of modern Electronic Health Record (EHR) data, leading to suboptimal risk prediction . Furthermore, these models typically operate as "black boxes," hindering clinical trust and adoption. This paper addresses these challenges by developing and validating an AI-Assisted Clinical Decision Support System (AI-CDSS) framework that integrates advanced health analytics with transparency. Our methodology employs a deep learning architecture, specifically tailored for patient trajectory modeling using longitudinal EHR data, to learn complex disease progression patterns. This model is then used to perform risk prediction for critical outcomes, such as 90-day readmission or functional decline. Crucially, we utilize eXplainable AI (XAI) techniques to implement interpretable ML , translating model outputs into clear, actionable clinical insights. The resulting system not only achieves superior predictive performance (e.g., demonstrating a significant lift in AUC compared to conventional clinical scores) but also provides robust, patient-specific factor importance. These transparent analytics offer a powerful tool for clinicians to identify high-risk patients earlier, understand the driving forces behind individual patient trajectories, and enable timely, personalized rehabilitation and chronic care interventions. This work represents a significant step toward trustworthy and effective AI integration in healthcare.