AUTHOR=Li Aoyu , Li Jingwen , Zhang Dongxu , Wu Wei , Zhao Juanjuan , Qiang Yan 

TITLE=Synergy through integration of digital cognitive tests and wearable devices for mild cognitive impairment screening

JOURNAL=Frontiers in Human Neuroscience

VOLUME=Volume 17 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/human-neuroscience/articles/10.3389/fnhum.2023.1183457

DOI=10.3389/fnhum.2023.1183457

ISSN=1662-5161

ABSTRACT=The rapid development of mobile computing platforms and wearable devices has enabled continuous monitoring of mild cognitive impairment (MCI) patients and their daily activities. These rich data could reveal more subtle variations in patients' behavioral and physiological features, providing new ways to detect MCI anytime, anywhere. In this study, we present a convenient, non-invasive framework for MCI classification that combines tablet cognitive tests and wearable sensors. Photoplethysmography (PPG), electrodermal activity (EDA) and electroencephalogram (EEG) signals were collected from 120 participants at rest and during the task. Features extracted from these physiological signals involved the time domain, frequency domain, time-frequency domain and statistics. However, time and score features during the cognitive test were recorded automatically by the system. Five different classifiers were used to classify the selected features for all modalities by ten-fold cross-validation. The experimental results show that using the weighted soft voting strategy achieved the highest accuracy rate of 88.9%. Notably, patients' classification performance improved when combining features from multiple modalities compared to using only tablet parameters or physiological features, indicating that our scheme could reveal MCI-related discriminative information. Finally, by comparing the classification results across cognitive tasks, we found the best overall performance on the digital span test, suggesting that MCI patients may have deficits in attention and short-term memory that came to the fore earlier.