EEG-Based Diagnosis of Alzheimer's Disease Using Kolmogorov Complexity

Alzheimer's disease (AD) is the most common and fastest growing neurodegenerative disorder of the brain due to dementia in old age people in Western countries. Detection and identification of AD patients from normal subjects using EEG biomarkers is a research problem. This study has developed an automatic detection of AD patients using Spectral Entropy (SE) and Kolmogorov Complexity (KC) feature sets. It is observed that (i) the SE value is low in AD patient's EEG signals compared to normal controlled subjects. (ii) AD patients’ EEG is more regular compared to normal controlled subjects, as shown by KC features. These feature sets have been computed and compared based on statistical measures of classifiers. We have used six different supervised and unsupervised classifiers in this research. Support Vector Machine classifier had performed well compared to others and achieved more than 95% accuracy when we provided both SE and KC feature sets. This work suggests that nonlinear EEG signal analysis can contribute to enhancing insights into brain dysfunction in AD.