Artificial intelligence based ECG signal classification of sendetary, smokers and athletes

Abstract

The current study deals with the design of a computer aided diagnosis procedure to classify 3 groups of people with different lifestyles, namely sedentary, smoker and athletes. The ECG Classification based on statistical analysis of HRV and ECG features. The heart rate variability (HRV) parameters and ECG statistical features were used for the pattern recognition in Artificial Intelligence classifiers. The ECG was recorded for a particular time duration using the EKG sensor. The HRV, time domain and wavelet parameters were calculated using NI BIOMEDICAL STARTUP KIT 3.0 and LABVIEW 2010. The important HRV features, time domain and wavelet features were calculated by the statistical non-linear classifiers (CART and BT).the important parameters were fed as input to artificial intelligence classifiers like ANN and SVM. The Artificial Intelligence classifiers like artificial neural network (ANN) and Support vector Machine (SVM) were used to classify 60 numbers of ECG signal. It was observed from result that the Multi layer perceptron (MLP) based ANN classifier gives an accuracy of 95%, which is highest among other the classifiers. The HRV study implies that the time domain parameters (RMSSD and PNN50), frequency domain parameters (HF power and LF/HF peak), Poincare parameter (SD1) and geometric parameters (RR triangular index and TINN) are higher in athlete class and lower in smoker class. The Higher values of HRV parameters indicate increase in parasympathetic activity and decrease in sympathetic activity of the ANS. This indicates that the athlete class has better heath and less chance of cardiovascular diseases where smoker class has high chances of cardiovascular diseases. These HRV parameters of sedentary class were higher than smoker class but lower than athlete class. This indicates less chances of cardiovascular disease in sedentary class as compared to smoker class.