Deep Learning based Heart Diseases Detection using ResNet50 Architecture

Artificial Intelligence (AI) in medical diagnostics has become an essential topic in research and business. Deep learning (DL) models are included in AI, and these models have been performing spectacularly in intelligent healthcare. About 25.6 million people died from cardiovascular illnesses in 2020, according to the World Health Organization (WHO) (CVD). So, this essay seeks to provide light on cardiovascular, which is often regarded as one of the essential fields in medicine. Cardiac issues might be prevented if the initial diagnosis was more correct. Generally, ECG testing is performed as a diagnosis to check for heart conditions. To detect different heart problems, an ECG records the electric cardiac output. To identify various heart conditions, several algorithms and approaches have been created using a variety of databases. Therefore, the goal of the proposed work is to analyze electrocardiogram (ECG) traces from images based on cardiac conditions such as arrhythmia or irregular heartbeat (AHB), myocardial infarction (MI), and previous MI (PMI). The proposed work trained on publicly available Kaggle datasets for detecting heart diseases. In the first epoch, the training accuracy of the model is 82%, and the validation accuracy is 91.88%. The model’s accuracy on both sets improves as training progresses, reaching validation accuracy high at increasing epoch model can overcome accuracy of 94.27%.