Heart disease, another name for cardiovascular illness, continues to be the world's leading cause of mortality. In order to save lives, early diagnosis and classification are essential. Electrocardiography (ECG), a popular, affordable, and non-invasive method of tracking the electrical activity of the heart, is essential for identifying heart-related disorders. The application of deep learning (DL) techniques to the prediction of four main cardiac conditions—abnormal heart rhythms, myocardial infarction, myocardial infarction history, and classifications of normal individuals—is examined in this study. The study first investigated transfer learning using pre-trained deep neural networks, namely AlexNet and SqueezeNet, on a small dataset of ECG pictures that were made publically available. In order to assess and forecast cardiac abnormalities, a unique CNN (convolutional neural network) architecture was also put forth. The study further evaluated the performance of the pre-trained models and the proposed CNN by utilizing them as feature extractors for heart condition forecasting.