Application of an Artificial Neural Network for Heart Disease Diagnosis

Abstract

In this paper, machine learning algorithms were applied for the diagnosis of cardiovascular diseases with the objective of proposing a tool for medical assistance. Artificial neural networks were used to classify patients as having or not having a cardiovascular disease, and a decision support system proposes to the clinician explanations of why such classification was made. An Amazon EC2 cloud server was used to test hyperparameters combinations, and an optimal configuration was used in the training section. As a result, the network achieved a 90.74% recognition rate of the test data and presented a score of 0.91 in the recall, precision, and f-1 score metrics, and 0.94 in the precision-recall and ROC curves.