JMIR Publications

ABSTRACT

Background:

With the global rise in Alzheimer’s disease (AD), early screening of mild cognitive impairment (MCI), which is a preclinical stage of AD, is of paramount importance. Although biomarkers such as cerebrospinal fluid amyloid level and magnetic resonance imaging have been studied, they have limitations such as high cost and invasiveness. Digital markers that assess cognitive impairment by analyzing behavioral data collected from digital devices in daily life, can be a new alternative. In this context, we developed a ‘Virtual Kiosk Test’ for early screening of MCI by analyzing behavioral data collected when using a kiosk in a virtual environment.

Objective:

Our goal was to investigate key behavioral features collected from a virtual kiosk test that could distinguish MCI patients from healthy controls with high statistical significance. Additionally, we focused on developing and validating a machine learning model capable of early screening of MCI based on these behavioral features.

Methods:

A total of 37 participants, comprising 17 healthy controls and 19 MCI patients, were recruited by two neurologists from a university hospital. The participants performed a virtual kiosk test—developed by our group—during which various behavioral data such as hand and eye movements were recorded. Based on these time series data, we computed the following four behavioral features: hand movement speed, proportion of fixation duration, the time to completion, and the number of errors. To compare these behavioral features between healthy controls and MCI patients, independent samples t-tests were employed. Additionally, we employed these behavioral features to train and validate a machine learning model for early screening of MCI patients from healthy controls.

Results:

In the virtual kiosk test, all four behavioral features showed statistically significant differences between MCI patients and healthy controls. Compared to healthy controls, MCI patients had slower hand movement speed (t(34)=4.88; P=.00), lower proportion of fixation duration (t(34)=2.99; P=.01), longer time to completion (t(34)=-2.58; P=.02), and a greater number of errors (t(34)=-3.08; P=.00). Three of these features (hand movement speed, proportion of fixation duration, and the number of errors) were then used to train a support vector machine to distinguish between healthy controls and MCI patients. The best machine learning model we trained achieved 91.67% accuracy, 100.0% sensitivity, 85.0% specificity, and 91.43% F1 score.

Conclusions:

Our research indicates that analyzing hand and eye movements in the virtual kiosk test can serve as a reliable digital marker for early screening of MCI. In contrast to conventional biomarkers, this digital marker in virtual reality is advantageous as it can collect ecologically valid data at an affordable cost and in a short period (five to 15 minutes), making it a suitable means for early screening of MCI.