"A Multimodal Explainable AI Framework for Early Prediction of Chronic Diseases Using EHR and Wearable Data"

"Early prediction of chronic diseases is critical for reducing healthcare costs and improving patient outcomes. Traditional machine learning models rely primarily on structured clinical data and often fail to incorporate continuous physiological signals from wearable devices. In this study, we propose a multimodal explainable artificial intelligence framework that integrates electronic health records (EHR) with wearable time-series data for early disease prediction. The proposed hybrid architecture combines XGBoost for structured data and Long Short-Term Memory (LSTM) networks for temporal data, followed by a fusion layer for final prediction. Additionally, SHAP-based explainability is employed to interpret model decisions. Experimental results demonstrate that the proposed model outperforms baseline methods in terms of accuracy and ROC-AUC score. The framework also provides clinically interpretable insights, enabling better decision-making and early intervention strategies."