A computational intelligence tool for the detection of hypertension using empirical mode decomposition. (March 2020)

• Record Type: Journal Article
• Authors: Soh, Desmond Chuang Kiat; Ng, E.Y.K.; Jahmunah, V.; Oh, Shu Lih; San, Tan Ru; Acharya, U. Rajendra
• Is Part Of: Computers in biology and medicine. Volume 118(2020)
• Publication Date: 2020-03
• Subjects: Hypertension -- Masked hypertension -- Computer intelligence tool -- Empirical mode decomposition -- Nonlinear features -- Student's t-test -- 10-fold validation -- Machine-learning; Medicine -- Data processing -- Periodicals; Biology -- Data processing -- Periodicals; 610.285
• DOI: 10.1016/j.compbiomed.2020.103630
• Languages: English
• ISSNs: 0010-4825

Abstract

Abstract: Hypertension (HPT), also known as high blood pressure, is a precursor to heart, brain or kidney diseases. Some symptoms of HPT include headaches, dizziness and fainting. The potential diagnosis of masked hypertension is of specific interest in this study. In masked hypertension (MHPT), the instantaneous blood pressure appears normal, but the 24-h ambulatory blood pressure is abnormal. Hence patients with MHPT are difficult to identify and thus remain untreated or are treated insufficiently. Hence, a computational intelligence tool (CIT) using electrocardiograms (ECG) signals for HPT and possible MHPT detection is proposed in this work. Empirical mode decomposition (EMD) is employed to decompose the pre-processed signals up to five levels. Nonlinear features are extracted from the five intrinsic mode functions (IMFs) thereafter. Student's t-test is subsequently applied to select a set of highly discriminatory features. This feature set is then input to various classifiers, in which, the best accuracy of 97.70% is yielded by the k-nearest neighbor (k-NN) classifier. The developed tool is evaluated by the 10-fold cross validation technique. Our findings suggest that the developed system is useful for diagnostic computational intelligence tool in hospital settings, and that it enables the automatic classification of HPT versus normal ECG signals. Highlights: Classification of normal and masked hypertension (MHPT) classes using ECG signals. Empirical mode decomposition (EMD) and nonlinear methods are employed. k-nearest neighbor (k-NN) classifier is implemented for classification. Obtained accuracy of 97.7% with ten-fold cross-validation. Tested on entire MIT-BIH normal sinus rhythm and SHAREE database of ECG signals.