Learning to Detect Cardiovascular Disease Using Multilayer Perceptron Neural Network

Cardiovascular diseases (CVDs) are the most perilous disease in worldwide. The CVD risk occurs abruptly at any instance with or without symptoms. The World Health Organization data shows CVD is the maximum rated disease. Artificial intelligence plays significant role for early diagnosis of CVD that supports health care professions. Advanced machine and deep learning techniques enhance the prediction of CVD based on clinical data. We propose MLP learning and testing model as an ensemble technique for the prediction of CVD risk. This ensemble model has two phases namely learner phase as level 0 and meta phase as level 1. Training is performed in learner phase using Logistic Regression, Decision Tree, Naïve Bayes, Support Vector Machine, and Random Forest and Multilayer perceptron. Multilayer perceptron is used in meta phase to test the model. The ensemble model is trained using 2215 samples of UCI dataset and tested with multilayer perceptron using 443 samples. The proposed model attains an accuracy rate of 96.1 %. The proposed model is then compared with existing CNN and ANN models and outperforms the other models.