Heart disease classification based on combination of PCA /ANFIS model

This study aims to create an intelligent system for the classification of electrocardiogram (ECG) signals using a combined approach of Principal Component Analysis (PCA) and the Adaptive Neuro-Fuzzy Inference System (ANFIS). Due to the high mortality rate of cardiac arrhythmia, ECG signals can be classified into four categories: normal sinus rhythm(NRS), premature ventricular contractions (PVC), supraventricular tachycardia (SVT), and atrial fibrillation (AFib) becomes important. This research article demonstrates an intelligent system implemented through the utilization of ANFIC hybrid approach for electrocardiogram (ECG) signal to categorize via ECG obtaining Features, which include the mean amplitude of the QRS complex, standard deviation (std) of QRS amplitude, mean of QRS time, std of QRS time, mean of the PR interval, std of the PR interval, mean of QT interval, std of QT interval, and statistics of heart rate variability. The output of the PCA was used as input for the ANFIS classifier.

The results prove that the provided feature extraction method and PCA technique can be implemented to improve the ANFIS system's ability to detect heart disease. The outcome shows that the tools used are highly efficient, with an accuracy level of more than 98%, an F score of 99.2%, and a sensitivity of 100%.

To get high accuracy a new feature extraction technique was used in the first stage, which produced 13 features, followed by PCA in the second stage to reduce the dimensionality of the obtained features. Based on these findings, our combined PCA/ANFIS model can significantly improve the detection of various heart conditions. Moreover, our results indicate that our strategy is superior to previously published methods.