An optimized cardiac risk levels classifier based on GMM with min- max model from photoplethysmography signals

Abstract

As per a latest study, coronary artery disease and hemorrhagic stroke are the predominant factors contributing to over 80% of cardiovascular diseases (CVDs). To reduce the mortality rate due to CVDs, researches are proposing the techniques for early detection of these CVDs. For the preliminary investigation on cardiovascular disease Photoplethysmography (PPG) can be used. Using PPG signals, it is possible to infer the risk levels like CVD with low risk, CVD with medium risk and respiratory disorder. To classify the risk levels of CVD, a model incorporating Gaussian mixture model (GMM) classifier with min-max decision model has been implemented. The proposed model resulted in better performance than existing classifiers like Logistic regression-GMM (LR-GMM), Detrend fluctuation analysis (DFA) and Cuckoo search algorithm (CSA) using min-max model. Based on the results GMM reflects a peak 95.9% classification accuracy with minimal false alarm of 7.1% and 0.99% miss classification when compared to other post classifiers.