DETECTION OF HEART DISEASES BASED ON EXTRACTED FEATURES

Cardiac rhythms that deviate from the normal are known as dysrhythmias, heart murmurs; they are the outcome of any of a number of disorders that involve the heart's electrical conduction system. These abnormalities can progress to life-threatening disorders like Irregular atrial rhythm, ventricular arrhythmias, or sudden death of cardiac cause. Thus, the rapid and accurate diagnosis of such arrhythmias is important for early clinical intervention.
In this paper, we propose an inter figure approach for high-speed inter figure AI based on features extracted from The MIT-BIH Arrhythmia Database was used to classify whether a given heart rhythm was normal or abnormal. electrical heart activity. Preprocessing of data consisted of filtering of the signal and R-peak for R-point peak detection routine with the Pan-Tompkins algorithm, and of analysis features extraction from both time-domain (for example, average and standard deviation) and frequency-domain (for example, dominant frequency and spectral entropy).
A model using the extracted parameters was conducted by XGBoost, and strong data performance was observed: Achieved a precision rate of 96.20%, a true positive rate of 95.40%, a true negative rate of 97.00%, and an F1 metric value of 96.10%." To better understand the decision making of the model, a feature importance evaluation was also preformed, revealing that the most effective predictors were dominant frequency and spectral entropy. The outcomes validate the proposed strategy to be used in real-time automatic ECG tracking systems.