Heart Attack Prediction with Deep Learning

Thumbnail Image
Syasya Putri binti Mohamad Razauddin
Journal Title
Journal ISSN
Volume Title
Research Projects
Organizational Units
Journal Issue
A heart attack, medically known as myocardial infarction happens when there are plaques at the coronary arteries, which is the blood path to the cardiac muscle. This plaque that blocks the artery and prevents blood flow, causes a sudden heart attack. Up until now, heart attack is one of the global leading cause of death mostly due to lack of effective tools to discover hidden relationships in e-health data and a late diagnosis. Therefore, the ability to predict and impending heart attack would save many lives. Hence, this study proposes a deep-learning-based diagnosis system for heart attack prediction using Artificial Neural Network algorithm. The simulation was done by using Jupyter Notebook accessed through Anaconda Navigator where the model is created based on Python code with the implementation of TensorFlow and Keras library. The datasets used are obtained from University of California Irvine (UCI) Machine Learning Repository and Open Machine Learning (OpenML). The algorithms are used to classify the datasets features in predicting the presence of heart attack. The model’s performance was evaluated based on three evaluation metrics such as accuracy, recall and precision. The Artificial Neural Network algorithm was compared with four popular machine learning algorithms such as Logistic Regression, Random Forest, K Nearest Neighbors and Support Vector Machine. The results proved that Artificial Neural Network could be utilised for prediction of heart attack due to its great performance with accuracy of 79%, recall of 75%, precision of 44% for Cleveland Heart Disease Dataset and accuracy of 77%, recall of 93% and precision of 77% for South Africa Heart Disease Dataset. In Artificial Neural Network, the number of hidden layers are proportional to model’s capacity and increment of 20 neurons to 26 neurons of South Africa Dataset recorded a constant value on all evaluation metrics involved.
FYP 2 SEM 2 2019/2020
Deep learning , Artificial neural network