Hybrid Detection Model for Meticulous Presaging of Heart Disease using Deep Learning: HDMPHD

Abstract

Heart diseases that occur due to the blockage of coronary arteries, which causes heart attack, are also commonly known as myocardial infarction. Rapid detection and acute diagnosis of myocardial infarction avoid death. The electrocardiographic test or ECG signals are used to diagnosis myocardial infarction with the help of ST variations in the heart rhythm. ECG helps to detect whether the patient is normal and suffering from myocardial infarction. In blood, when the enzyme value increases, after a certain time pass occurs, heart attack. For ECG images, the manual reviewing process is a very difficult task. Due to advancements in technology, computer-aided tools and software are used to diagnosis myocardial infarction,because manual ECG requires more expertise .so that automatic detection of myocardial infarction on ECG could be done by different machine learning tools. This study detects the normal and myocardial infarction patients by selecting the feature with their feature weights by selecting from the model and by Random forest classifier selecting the index value using DenseNet-121, ResNet_50, and EfficientNet_b0 deep learning techniques .This proposed work used the real dataset from Medanta hospital (India) at the time of covid 19. The dataset is in the form of ECG images for Normal and myocardial infarction (960 samples). With an end-to-end structure, deep learning implements the standard 12-lead ECG signals for the detection of normal and myocardial infarction..The proposed model provides high performance on normal and myocardial infarction detection. The accuracy achieved by the proposed model for Efficientnet_b0 Random Forest to Select from Model Accuracy 84.244792, Precision 84.396532, Recall 84.227410, F-Measure 84.222295.