Detecting Alzheimer's disease using digital virtual reality cognitive tests

Digital biomarkers for detecting Alzheimer's disease (AD) remain controversial due to inconsistent accuracy and unclear relationships with AD core biomarkers.

This study recruited 428 Chinese older adults to determine the associations of artificial intelligence (AI)-assisted virtual reality (VR) cognitive scores with plasma amyloid beta (Aβ)₄₂/Aβ₄₀, phosphorylated tau (p-tau)₂₁₇, glial fibrillary acidic protein (GFAP), neurofilament light chain (NfL), Aβ positron emission tomography (PET), tau PET, neurodegeneration, and cognition using generalized linear models, including age, sex, education, and apolipoprotein E ε4 as covariates.

The AI-VR cognitive scores were significantly correlated with plasma Aβ₄₂/Aβ₄₀, p-tau₂₁₇, GFAP, NfL, Aβ PET, tau PET, hippocampal atrophy, temporal meta-region of interest cortical thinning, and cortical mean diffusivity increases, accurately identifying cognitively impaired patients from unimpaired individuals (area under curve [AUC]: 0.908-0.969) as well as classifying Aβ PET-positive individuals from negative group (AUC: 0.771-0.991).

This study comprehensively examined the capacity of a 5-minute AI-VR cognitive test for identifying abnormal cognition and AD biomarkers, offering a reliable digital marker for early AD detection in community settings.

Digital virtual-reality cognitive tests correlate with AD biomarkers Digital virtual-reality cognitive tests can identify cognitive impairment Digital virtual-reality cognitive tests can classify those who are amyloid beta positron emission tomography positive from negative.