Premium
Automatic assessment with EEG‐based brain‐computer interface shows difference between healthy population, mild cognitive impairment and moderate cognitive decline patients
Author(s) -
Maimon Neta,
Molcho Lior,
Loterrman Tomer,
Pressburger Narkiss,
Sasson Ady,
Intrator Nathan
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.043499
Subject(s) - cognitive decline , audiology , cognition , biomarker , electroencephalography , resting state fmri , population , effects of sleep deprivation on cognitive performance , psychology , brain activity and meditation , medicine , dementia , neuroscience , disease , biology , environmental health , biochemistry
Background Cognitive decline is still under‐diagnosed due to the lack of objective biomarkers. Brain‐ Computer Interface (BCI) combined with inexpensive mobile EEG devices have been used to assess mental workload revealed different neurological biomarkers to detect cognitive load. However, they are still scarcely implemented among the elderly and cognitively impaired populations. We introduce novel cognitive biomarkers obtained from a single‐EEG‐channel (by Neurosteer Aurora) to indicate cognitive decline. The assessment is fully automated using auditory verbal and musical stimulation. Methods Healthy participants (N=40) and mild‐to‐moderate cognitive decline (MMSE score 17‐23, N=10; and 24‐27, N=14) were tested in three conditions: an auditory detection task with two difficulty levels and a resting state task. Two biomarkers, previously extracted from healthy participants data through machine learning, VC9 and T2, were used in the statistical analysis. Results For the healthy population, VC9 activity was significantly increased for higher cognitive load (p<.001); for the mild‐to‐moderate cognitive decline group, VC9 activity was significantly higher than healthy group (p=.004), and remained similar for all tasks and levels (see figure 1 for the mean activity of VC9 biomarker per each condition: detection level 1, level 2 and resting state, for the healthy participants and cognitively impaired group). The biomarker showed a difference in mean activation during the cognitive and resting tasks (p=.022), separating between the two groups of cognitive decline (MMSE<24 & MMSE>24) (see figure 2 for the mean activity of T2 biomarker comparing the two cognitively impaired groups and the healthy group). Furthermore, the T2 feature was negatively correlated with participants’ MMSE score (p=.013, see figure 3 for correlation between MMSE score and T2 activity). Conclusions These initial results show two novel biomarkers of cognitive activity, which are able to differentiates between patient groups and detect MCI. The study shows different patterns of cognitive functions for patients with cognitive impairment, as well as difference in total activity between healthy subjects and mild cognitive impairment patients, to patients with moderate cognitive impairment. Results were obtained with an inexpensive BCI combined with an easy and automatic cognitive assessment, which provides more objective evaluation and decreases the need of highly trained personnel.