Using Computational Classifiers to Detect Chronic Alcoholism. Issue 2 (April 2016)

• Record Type: Journal Article
• Authors: Kumar, Surendra; Ghosh, Subhojit; Sinha, Rakesh Kumar
• Is Part Of: Journal of clinical engineering. Volume 41:Issue 2(2016)
• Publication Date: 2016-04
• Subjects: Biomedical engineering -- Periodicals; 610.28
• DOI: 10.1097/JCE.0000000000000154
• Languages: English
• ISSNs: 0363-8855

Abstract

Abstract : Addiction to alcohol is one of the commonest problems in human society. The tribes of the Jharkhand state of India are also generally addicted to a homemade alcohol Mahuwa, fermented from the flower of Madura longifolia . In the present work, the effects of Mahuwa on the cerebral cortex of the brain are evaluated. The digital brain signals or electroencephalogram (EEG) was recorded from frontal, central, and parietal cortices of the brain, and the features for classification were extracted with both time and frequency domain tools: the third-order nonautoregressive method and power spectral density, respectively. The dimensions of these features were further reduced by using linear discriminant analysis before the identification of alcoholic EEG data with the help of support vector machine and fuzzy C-means clustering algorithms. Results obtained from third-order nonautoregressive and power spectral density–based features along with support vector machine and fuzzy C-means classifiers revealed that the frontal cortex of the brain and β band of the EEG frequency are most vulnerable to the chronic use of Mahuwa alcohol. On the basis of the results, it can be suggested that EEG has sufficient information for development of a noninvasive online diagnostic tool for chronic alcoholism.