Accepted for/Published in: Journal of Medical Internet Research
Date Submitted: Nov 15, 2023
Date Accepted: Mar 9, 2024
Integrating biomarkers from virtual reality and magnetic resonance imaging for early detection of mild cognitive impairment using a multimodal learning approach: Validation study
ABSTRACT
Background:
Early detection of mild cognitive impairment (MCI), a transitional stage between normal aging and Alzheimer’s disease (AD), is crucial for preventing dementia progression. Identifying subtle deficits in instrumental activities of daily living (IADL), such as challenges in using a food-ordering kiosk, can effectively indicate early MCI. While virtual reality (VR) has been successful in capturing behavior related to IADL for early MCI detection, the relationship between VR biomarkers and observable structural brain changes via magnetic resonance imaging (MRI) requires further exploration. Moreover, the integration of VR and MRI biomarkers for early MCI detection remains an open question.
Objective:
This study aims to enhance the performance of early MCI detection by leveraging multimodal learning to integrate VR and MRI biomarkers.
Methods:
The study included 44 participants, consisting of 19 healthy controls and 25 MCI patients. Participants completed a virtual kiosk test to collect four VR biomarkers (hand movement speed, scanpath length, time to completion, and number of errors), and T1-weighted MRI scans were performed to collect 16 MRI biomarkers from both hemispheres.
Results:
Correlation analysis showed a significant association between MRI-observed brain atrophy and impaired IADL performance in VR. The support vector machine (SVM) using only VR biomarkers showed high specificity in identifying healthy controls, whereas the MRI-based model exhibited high sensitivity in distinguishing true MCI patients. These findings led us to propose a novel clinical approach, employing VR biomarkers for initial screening of healthy controls and MRI biomarkers for MCI confirmation. Furthermore, a multimodal SVM trained using both VR and MRI biomarkers surpassed unimodal SVM models in terms of accuracy (97.73%), sensitivity (100.00%), specificity (97.73%), precision (96.15%), and F1-score (98.04%).
Conclusions:
Overall, the multimodal learning approach presented in our study offers significant insights into enhancing the performance of early MCI detection by integrating diverse biomarkers.
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.