Publication

Abstract : Cardiovascular disease (CVD) stands as the most widespread severe illness impacting human health on a global scale. Forecasting CVDs in advance becomes more and more crucial as CVDs increase exponentially every day. Deep Learning (DL) algorithms are self-adaptive to recognize patterns and analyze data more effectively in CVD prediction. Over the past few decades, many researchers and practitioners have examined different predictive algorithms, but most of those studies are based on small-sized datasets like less than 10,000 patient records. The major shortcomings of earlier research lie in its reliance on small-sized datasets, elevating the risk of overfitting. In contrast, our study addresses this limitation by utilizing Kaggle’s cardiac dataset encompassing 70,000 patients and 11 features. The primary objective of this study is to minimize the risk of overfitting and accurately predict CVD by showcasing the effectiveness of using comprehensive datasets. This paper proposes a hybrid DL methodology by utilizing a Extra Tree Classifier with Artificial Bee Colony optimized Recurrent Neural Network (ETC-ABC-RNN) for accurate classification of CVDs with 96% accuracy. By measuring accuracy, precision, recall, and F1, the efficiency of the system is demonstrated. The outcomes demonstrated that the suggested methodology surpassed various methods in predicting heart disease.