Coronary Heart Disease Prediction Using CNN, RNN, LSTM and Attention DNN Model

Coronary Heart Disease popularly referred to as CHD is one of the leading causes of death and illness across the global population making it imperative for the identification of an effective approach to its predictive modeling for early detection. The purpose of this research includes studying a dataset containing a variety of clinical characteristics in order to forecast CHD using sophisticated deep learning structures. Specifically, we implemented and evaluated four distinct models: Convolutional Neural Networks, Long Short-Term Memory networks, Recurrent Neural Networks and Attention Deep Neural Networks (DNN). All models are trained and evaluated on data set and multiple performance measures like accuracy, recall, precision, and F1score were used for the purpose. The experimental outcome showed that the Attention-based DNN had the highest accuracy rate of 91 % followed by the RNN with accuracy rate of 89%. CNN and LSTM models gave moderate accuracy but they did not outperform the best models among options.