JMIR Publications

ABSTRACT

Background:

Traditional paper-and-pencil cognitive scales for older adults with cognitive impairment suffer from subjectivity, monotony, and poor ecological validity. Virtual reality (VR) combined with eye-tracking offered objective, ecologically valid assessment approaches based on daily living activities. However, localized VR cognitive screening tools tailored to Chinese older adults’ domestic dietary routines remain scarce.

Objective:

Focusing on the cognitive traits and eating habits of elderly people in China, we combined eye-tracking technology to develop a virtual kitchen cognitive screening system tailored to local needs.

Methods:

This two-phase study adopted a mixed-methods design. Phase 1 centered on system development: a multidisciplinary team formulated an 11-step standardized tomato scrambled egg cooking task, capturing seven real-time eye movement indicators. Two rounds of Delphi expert consultation (16 experts total) were performed to screen indicators using thresholds of mean importance >3.50 and coefficient of variation <0.25. Usability testing was conducted in Phase 2 with 17 older adults aged 72–92 recruited from a nursing home in Fuzhou; participants were stratified into three groups: normal cognition (NC, n=6), MCI (n=6), and mild-to-moderate AD (n=5). Quantitative measurements included the System Usability Scale (SUS), Simulator Sickness Questionnaire (SSQ), and clinical cognitive scales (MMSE, MoCA, CDR, ADL). Semi-structured interviews were audio-recorded, transcribed, and analyzed via conventional qualitative content analysis until thematic saturation.

Results:

The VK-CS system was divided into five major modules: subject registration and login, eye-tracking device calibration, virtual kitchen familiarization, cognitive screening module, and data management. Two rounds of expert consultation yielded high expert enthusiasm (93.75% and 87.5%) and an expert authority coefficient of 0.897; rising Kendall’s W demonstrated converged expert consensus. Among all participants, the mean SUS score reached 89.50±2.52, indicating excellent usability, while the total SSQ score was only 1.21±1.08, reflecting minimal VR cybersickness. Qualitative themes demonstrated high perceived ease of use attributed to lifelike kitchen scenes, realistic cooking workflows, and clear voice prompts. Key barriers to widespread use included heavy VR head-mounted displays, complex controller operations, undersized on-screen fonts, and motion discomfort induced by rapid scene transitions. Advanced age, low educational attainment, and pre-existing chronic illnesses further compromised task completion performance.

Conclusions:

The culturally localized VK-CS system integrating VR and eye-tracking exhibited outstanding usability and low VR sickness burden for Chinese older adults. It overcomes major limitations of conventional cognitive assessments by embedding ecologically valid household cooking tasks and objective eye-tracking biomarkers. Identified hardware and interface limitations require targeted iterative optimization. Multi-center large-sample trials are warranted to verify its diagnostic sensitivity and specificity against clinical gold-standard scales, facilitating standardized deployment in community and institutional aged care settings for routine early cognitive screening. Clinical Trial: none