Prediction of heart disease is more crucial for reducing mortality rates. Many other conditions can contribute to heart disease, including clogged arteries, heart attacks, chest discomfort, and strokes. Bypass surgery or coronary artery surgery is performed to correct issues with the heart's muscles, valves, or rhythm caused by heart disease. Lack of early diagnosis of heart disease in humans is a leading cause of mortality. Using a pre-processing and classification strategy, accurately forecast cardiac issues. In this study, we utilise the UCI heart disease dataset and a real-time dataset to evaluate deep learning algorithms against conventional approaches. As a first step, we examined the interplay and dependency between a variety of clinical characteristics and coronary heart disease. Finally, we use both the whole set of features and a smaller subset to conduct an in-depth examination of the Modified Convolutional Neural Network (MCNN) classification model. In order to enhance the CNN's classification capabilities, an IGHOA model is applied to it. Accuracy, the ROC curve, and the F1-Score were used to classifiers. The classification outcomes demonstrated that pertinent characteristics had a significant effect on organization precision. When compared to replicas trained on a complete feature set, the presentation of classification models trained with a smaller feature set better melodramatically with less training period.