An adaptive heart disease diagnosis via ECG signal analysis with deep feature extraction and enhanced radial basis function

Heart Disease (HD) has become a major disease that leads to death worldwide. Here, identifying heart disease in the early stage is essential for the patient's survival. While diagnosing heart disease with the ECG signal, it may cause an error due to the small amplitude variation. The highly accurate recognition process is the only option to save human lives. Initially, the gathered signals are decomposed using Adaptive DWT. Then, these decomposed signals are fed to the Deep Convolutional Neural Networks (DCNN) features to get the first set of features. Secondly, the R-R interval is analyzed and taken from the gathered signals and subjected to a deep feature extraction stage, where the DCNN gets the second set of features from R-R intervals. Thirdly, QRS waves are analyzed from the gathered signals and given to the feature extraction stage, where the third set of features are gathered from QRS waves using DCNN. Finally, all these three sets of features are fused for processing heart disease detection, where the ERBF is utilized for getting the classified outcomes. From the overall analysis, the F1-score of the designed approach is 96.03%. Thus, the experimental outcome has ensured the performance with accuracy, F1-score, and AUC.