Abstract
Congestive heart failure (CHF) is gradually becoming more prevalent due to the stressed lifestyles in modern life. Accurate detection with lower computational complexity and lower cost of diagnosis is a challenge to the researchers in this domain. In this work, I have proposed an approach using naive Bayes algorithm with a lesser number of significantly discriminating features for differentiating the CHF subjects from the normal subjects. The small size of feature sets enhances the computational efficiency and the choice of strong features improves the accuracy. The features are chosen on the basis of p-value of the 2-sample t-test performed between the two types of subjects. Using the p-value, 6 features are selected to train, validate and test the classifier. Publicly available benchmark PhysioNet datasets for congestive heart failure patients and normal subjects are used to carry out the experimentation. This approach is able to provide 100% classification accuracy as well as sensitivity and specificity of 100% in identifying CHF patients employing Gaussian naive Bayes algorithm.
Highlights
Congestive heart failure (CHF) is gradually becoming more prevalent due to the stressed lifestyles in modern life
Regression tree (CART), naive Bayes network (NB), random forest (RF), convolutional neural network (CNN), and various fuzzy classifiers are widely used in many works to detect different types of heart diseases [1,2,3,7,8,9,10,11,12,13]
Wijbenga et al [8] have performed an analysis on CHF patients by using parameters including left ventricular ejection fraction (LVEF), heart rate variability (HRV) triangular index (HTI), systolic pressure, HR and found significant change in these parameters for CHF patients
Summary
The statistical parameters, based on the differences between RR intervals are calculated to observe the pattern of changes in HRV for CHF patients. DWT is a discrete version of a time-frequency analysis technique which basically filters a signal by a set of high pass filters (HPF) and low pass filters (LPF) [20,21]. It is suitable for analyzing non-stationary signals since it provides a more localized time-frequency analysis. The increase in the decomposition level provides the information of more detailed (high frequency) components of a signal. For our analysis ‘db4’ wavelet member of Daubechies family with 4th level of decomposition is chosen considering the smoothness, compact support properties of db. We have calculated the Shannon entropy (ShEn) [22], kurtosis, skewness [21], and the standard deviation of the 3rd level detail component coefficients as the features to be trained in the naive Bayes classifier
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: International Journal of Engineering and Advanced Technology
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
