DeepHeart: A Deep Learning Model for Predicting Heart Diseases Using Convolutional Neural Network

Abstract

Objectives: Heart disease is a significant and widespread cause of illness and death around the world. Traditionally, heart disease risk assessment relies on clinical variables. The integration of deep learning techniques in this domain is a challenging, yet promising, solution. Methods: In this study, we employed a Deep Convolutional Neural Network (DCNN) to predict the risk of heart disease using clinical cardiology data. The dataset used in the analysis was obtained from the publicly available UCI Heart Diseases Repository consisting of 920 patients’ records comprising 14 attributes. The model was trained and fine-tuned using a grid search approach to optimize the best hyperparameters. Findings: The performance of the model was rigorously assessed through cross-validation techniques. We found that the proposed technique demonstrated remarkable performance, achieving an impressive testing accuracy of 83%. Precision, recall, and F1-score were equally notable at 84%, 83%, and 83%, respectively; showcasing the model's well-balanced classification capabilities. Our research produced promising results while highlighting the potential of the proposed DCNN model as a robust tool for heart disease prediction.