A Comparative Study of Two Machine Learning Algorithms for Heart Disease Prediction System

Increasing cases of heart disease cause more people to wait to be diagnosed, leading to the inefficiency of the diagnosing process in terms of time and human labor required. However, this problem can be alleviated if a prediction mechanism is in place. The purpose of this study is to predict whether a person has heart disease or not. The prediction system will use machine learning algorithms to select attributes obtained from the Cleveland dataset. Prediction is made using two machine learning models, Artificial Neural Network (ANN) and Logistic Regression. Different sizes of hidden layers and activation functions are used to find the hyperparameters with optimal performance. The number of inputs and outputs are kept constant at one with a maximum iteration of 500. Logistic Regression is used to classify a discrete data set and return the probability value where the Sigmoid function acts as the cost function. Finally, a confusion matrix was used to compare the performance of both models. ANN resulted in higher accuracy of 92.31% and an F1-score of 93.2% compared to Logistic Regression with 90.11% accuracy and an F1-score of 91.26%.