A Comparative Study between Neural Network Models and Standard Machine Learning Models in Heart Disease Prediction

No Thumbnail Available
Uttarkar, Dhanush Vinay
Issue Date
MSc in Data Analytics
Dublin Business School
Items in eSource are protected by copyright. Previously published items are made available in accordance with the copyright policy of the publisher/copyright holder.
Cardiovascular diseases are the leading cause of death across the world. There has been a gradual increase in cardiovascular diseases. It is also called a ‘Silent Killer’ because most people who have it do not have obvious symptoms. An artificial neural network is a part of machine learning with a potential solution to identify or detect the onset of this disease effectively. Artificial neural network (ANN) is a technological advancement in the field of machine learning that is gaining a lot of traction because of its design which allows it to solve many complex problems. ANNs are playing a vital role in many sectors of the industry, it is used for financial data analysis, speech recognition, emotion detection, disease prediction, image generation, and many other such applications. Artificial neural networks are made to imitate the human brain and function like one. It computes data like the human brain. The node of the artificial neural network resemble the individual neurons of the human brain. Our report analyses and compares the prediction of heart diseases between a few previously implemented standard machine learning models such as the logistic regression model, an ensemble model like random forest and decision trees versus neural networks like Long Short Term Memory Networks (LSTM), which is a type of recurrent neural network (RNN), a simple Convolutional Neural Network (CNN), a simple recurrent neural network, and a feed-forward neural network. To perform this experiment of comparative analysis we use the Cleveland heart disease dataset available at the UCI (University of California, Irvine) machine learning repository donated by Peter Turney, we compare their precision, accuracy, sensitivity, and specificity. The results we obtained were 72.49 % accuracy for the LSTM model, 75.73% accuracy for the CNN model, 74.43% accuracy with the feed forward neural network, 72.17% accuracy for the RNN model.