Heart Disease Classification Using Different Transfer Learning Models

Heart diseases are becoming more common daily, especially among the older population. Therefore, an automated detection system is necessary to avoid the life-threatening effects of heart disease (HD). Early identification of heart problems may be determined by an automatic system, which specialists in the medical field use. In the past days, different diseases were detected by manually identifying features from the Electrocardiogram signal (ECG). However, the latest study has shown that information extraction and categorization may be done independently by employing deep learning (DL) neural networks. To develop efficient DL models, massive amounts of information are necessary; this is not the case for physiological signals. Therefore, in terms of transfer learning, both the recognition and categorization of heart diseases are considered. In this work, we have used eight transfer learning (TL) models to classify HD based on the information gained from millions of data that are applied to our data set. Various TL models are evaluated based on accuracy, layers, optimizer, and signal-to-image conversion method.