Heart disease prediction using distinct artificial intelligence techniques: performance analysis and comparison

Consolidated efforts have been made to enhance the treatment and diagnosis of heart disease due to its detrimental effects on society. As technology and medical diagnostics become more synergistic, data mining and storing medical information can improve patient management opportunities. Therefore, it is crucial to examine the interdependence of the risk factors in patients' medical histories and comprehend their respective contributions to the prognosis of heart disease. This research aims to analyze the numerous components in patient data for accurate heart disease prediction. The most significant attributes for heart disease prediction have been determined using the Correlation-based Feature Subset Selection Technique with Best First Search. It has been found that the most significant factors for diagnosing heart disease are age, gender, smoking, obesity, diet, physical activity, stress, chest pain type, previous chest pain, blood pressure diastolic, diabetes, troponin, ECG, and target. Distinct artificial intelligence techniques (logistic regression, Naïve Bayes, K-nearest neighbor (K-NN), support vector machine (SVM), decision tree, random forest, and multilayer perceptron (MLP)) are applied and compared for two types of heart disease datasets (all features and selected features). Random forest using selected features has achieved the highest accuracy rate (90%) compared to employing all of the input features and other artificial intelligence techniques. The proposed approach could be utilized as an assistant framework to predict heart disease at an early stage.