Diagnostic performance of self-administered unsupervised computerized cognitive tests for the identification of mild cognitive impairment (MCI) and dementia: a systematic review and meta-analysis

Self-administered, unsupervised computerized tools for community-based early cognitive screening are gradually being developed and applied. This study aims to evaluate the diagnostic performance and system usability of those tools in detecting MCI and dementia in older adults. Five electronic databases were systematically searched from inception to 17 August 2023. The included studies were reported using the PRISMA 2020 guideline, the risk of bias and applicability was assessed using QUADAS-2. 28 articles were included in the systematic review and 21 studies in the meta-analysis. The pooled sensitivity and specificity for unsupervised computerized tools were 0.84 (95% CI: 0.81 - 0.87) and 0.86 (95% CI: 0.82 - 0.89) respectively. Meta-regression showed that mean age, cognitive status, cognitive domains and administration time might be responsible for heterogeneity. Notably, tools taking less than 5 min had a pooled sensitivity of 0.85 (95% CI: 0.81-0.89). However, only 3 studies evaluated feasibility and system usability in the intended clinical settings. Unsupervised computerized cognitive screening tools demonstrate good diagnostic performance for MCI/dementia, and tools taking less than 5 min appear particularly suitable for large-scale cognitive screening. Future research should combine various cognitive data to develop multimodal screening tools with better diagnostic performance, evaluate these tools' usability and improve age-friendly designs.