A Hybrid Explainable Deep Learning Model for Chronic Disease Risk Stratification using Multivariate Electronic Health Records

This work aims to meet the need for easy risk rating in chronic diseases by proposing a time-based deep-learning technique for analysing large electronic health record, or EHR, data, therefore promoting equal medical access and sustainable healthcare innovation. It maintains openness and transparency while analysing the medical records of patients over an extended period to help predict how a disease will progress and figure out the best approach to break up patient groups according to risk. By involving chronology and attention mechanisms, the model can make precise predictions and offer medically relevant explanations for its decisions. Its efficacy in identifying high-risk patients and facilitating early intervention is demonstrated by research using actual EHR data. Additionally, by supporting AI-driven, accessible, and environmentally friendly medical solutions, it is appropriate for healthcare institutions with limited equipment and aligns with SDG 3 (Good Nutrition and Well-Being), SDG 9 (Industry, Innovation, as well as Infrastructures), as well as SDG 10 (Reduced Inequalities.