Abstract

Heart disease holds a main place in public health issue globally, making millions of deaths annually. The hike in the number of cases for cardiovascular disease highlights the need for effective and automated smart diagnostic methods. This work brings an intelligent E-healthcare system for heart disease detection using hybrid deep learning and ensemble classification method. The model proposed here joins a 1D Convolution Neural Network (CNN), a Bidirectional Long Short-Term (BiLSTM) network, and an XGBoost classifier to improve both features such as age, gender, blood pressure, cholesterol, glucose level and heart rate, seizing high order nonlinear relationships in the data. The BiLSTM netwok cathes temporal and sequential dependencies in patient health records to point out risk patients such as “older + smoker + high glucose = High risk”. Data preprocessing methods like normalization, missing value during input, feature selection are used to make sure that the datasets are consistent and dependable. Experimental resultants proved that the proposed hybrid CNN-BiLSTM-XGBoost hybrid algorithm has achieved accuracy of 98.4%, which is better than the previous models, the proposed model also performed with precision 98.6%, recall 98.5% and AUC scores 98.9% which proves its strength and prediction efficiency.