Classification of Heart Disease using Artificial Neural Network

Abstract

Heart disease refers to any unavoidable condition that affects a person's heart. With so much electronic health data available nowadays, computer machines have entered the area to aid in diagnosis using machine learning (ML) methodologies. This study focuses on classification of heart disease that includes or does not include heart attacks and employs an artificial neural network (ANN) namely Multi-layered Perceptron (MLP) trained by Back-propagation (BP) to classify the data. By going through layers of functions, ANN can identify the trends of the data then form a model. This study uses sigmoid activation functions and runs in 1000 numbers of epochs. Different values of learning rate and neuron numbers were tested and yielded with the best result. The findings recorded high accuracy of 80.66% with 0.25 learning rate and 25 neurons. It is learned that ANN can be used for classifying heart disease cases.