AUTHOR=Shima Sayuri , Ohdake Reiko , Mizutani Yasuaki , Tatebe Harutsugu , Koike Riki , Kasai Atsushi , Bagarinao Epifanio , Kawabata Kazuya , Ueda Akihiro , Ito Mizuki , Hata Junichi , Ishigaki Shinsuke , Yoshimoto Junichiro , Toyama Hiroshi , Tokuda Takahiko , Takashima Akihiko , Watanabe Hirohisa 

TITLE=Virtual reality navigation for the early detection of Alzheimer’s disease

JOURNAL=Frontiers in Aging Neuroscience

VOLUME=Volume 17 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/aging-neuroscience/articles/10.3389/fnagi.2025.1571429

DOI=10.3389/fnagi.2025.1571429

ISSN=1663-4365

ABSTRACT=ObjectiveThe development of non-invasive clinical diagnostics is paramount for the early detection of Alzheimer’s disease (AD). Neurofibrillary tangles in AD originate from the entorhinal cortex, a cortical memory area that mediates navigation via path integration (PI). Here, we studied correlations between PI errors and levels of a range of AD biomarkers using a 3D virtual reality navigation system to explore PI as a non-invasive surrogate marker for early detection.MethodsWe examined 111 healthy adults for PI using a head-mounted 3D VR system, AD-related plasma biomarkers (GFAP, NfL, Aβ40, Aβ42, and p-tau181), Apolipoprotein E (ApoE) genotype, and demographic and cognitive assessments. Covariance of PI and AD biomarkers was assessed statistically, including tests for multivariate linear regression, logistic regression, and predictor importance ranking using machine learning, to identify predictive relationships for PI errors.ResultsWe found significant positive correlations between PI errors with age and plasma GFAP, p-tau181, and NfL levels. Multivariate analysis identified significant correlations of plasma GFAP (t-value = 2.16, p = 0.0332) and p-tau181 (t-value = 2.53, p = 0.0128) with PI errors. Predictor importance ranking using machine learning and receiver operating characteristic curves identified plasma p-tau181 as the most significant predictor of PI. ApoE genotype and plasma p-tau181 showed positive and negative PI associations (ApoE: coefficient = 0.650, p = 0.037; p-tau181: coefficient = −0.899, p = 0.041). EC thickness exhibited negative correlations with age, mean PI errors, and GFAP, NfL, and p-tau181; however, none of these associations remained significant after adjusting for age in linear regression analyses.ConclusionThese findings suggest that PI quantified by 3D VR navigation systems may be useful as a surrogate diagnostic tool for the detection of early AD pathophysiology. The hierarchical application of 3D VR PI and plasma p-tau181, in particular, may be an effective combinatorial biomarker for early AD neurodegeneration. These findings advance the application of non-invasive diagnostic tools for early testing and monitoring of AD, paving the way for timely therapeutic interventions and improved epidemiological patient outcomes.