Comparison of Neural Biomarker Assessment Methods for Early Detection of Alzheimer's Disease

With growing age, the cognitive ability degrades gradually as an aging factor. For a portion of people, the cognitive capability diminishes to a great extent, which will eventually result in Alzheimer's disease (AD). Mild cognitive impairment (MCI) is considered as an intermediate stage of AD. Diagnosis of AD patients at an early stage can reduce the chance of developing into a severe condition for cognition. This study aims to investigate the MCI assessment methods (statistical analysis and individual classification) for distinguishing the healthy control (HC) and MCI patients via functional near-infrared spectroscopy (fNIRS). This study evaluated ten digital biomarkers from three brain regions and three mental tasks ($N$-back, Stroop, and verbal fluency task). Among these three tasks, the $N$-back task achieved the best accuracy (76.67 %) with biomarker 2 (HbO mean from 5 to 25 sec) and 7 (HbO slope from 0 to peak value) in the middle prefrontal cortex by linear discriminant analysis (LDA). Additionally, the statistical analysis results also indicated that a significant difference ($p$-value < 0.05) existed between MCI and HC. However, the biomarkers, which achieved an individual classification accuracy more than 70%, could not be consistent with the biomarkers with $p$-value < 0.05. It reveals that statistical analysis technique still should be improved for diagnosing MCI individuals. Machine learning (LDA) can contribute as a tool by early prediction of AD via analyzing digital biomarkers using a non-invasive technique.