Reliability and Validity of a Gamified Cognitive Assessment for Adults at High Risk of Stroke

Background: Most existing digital cognitive assessment tools are simple adaptations of paper-and-pencil tests, lacking adequate validation in older adults and showing poor sensitivity to domain-specific deficits common in high vascular risk populations. This study aimed to evaluate the reliability, convergent validity, and user experience of a gamified Digital Cognitive Assessment Game (DCG) battery developed for adults at high risk of stroke. Methods: This study recruited community dwellers with high risk of stroke. Participants completed the DCG battery, a standardized neuropsychological battery, and a 2-month retest. Reliability was assessed using intraclass correlation coefficients (ICCs). Convergent validity was examined via Pearson correlations and canonical correlation analysis (CCA). User experience was evaluated using Likert-scale questionnaires. Results: A total of 86 participants (mean age 67 years; 60.2% female; mean education 11.37 years) completed baseline assessments. Furthermore, 12/18 DCG tasks showed acceptable-to-high test-retest reliability (ICC range: 0.795-0.883), while the remaining 6 tasks showed marginally acceptable reliability (ICC range: 0.615-0.737). Convergent validity revealed correlations in five domains (r = 0.373-0.582, p < 0.05), with the exception of visuospatial function (r = 0.260, p = 0.063), and a strong correlation for global cognition (r = 0.734, p < 0.001). CCA demonstrated a strong overall association between DCG and conventional batteries (R = 0.86, p < 0.001). The DCG showed good ability for discriminating battery-defined mild cognitive impairment (AUC = 0.91, 95% CI: 0.85-0.97). Participants reported good usability and high acceptance (80% reported having no difficulties completing DCG). Conclusions: The DCG battery demonstrates satisfactory reliability, validity, and user acceptability in adults at high risk of stroke. As a self-administered tool, it represents a promising approach for cognitive screening, particularly in resource-limited settings.