Background Research investigating treatments and interventions for cognitive decline fail due to difficulties in accurately recognizing behavioral signatures in the presymptomatic stages of the disease. For this validation study, we took our previously constructed digital biomarker‐based prognostic models and focused on generalizability and robustness of the models. Method We validated prognostic models characterizing subjects using digital biomarkers in a longitudinal, multi‐site, 40‐month prospective study collecting data in memory clinics, general practitioner offices, and home environments. Results Our models were able to accurately discriminate between healthy subjects and individuals at risk to progress to dementia within 3 years. The model was also able to differentiate between people with or without amyloid neuropathology and classify fast and slow cognitive decliners with a very good diagnostic performance. Conclusion Digital biomarker prognostic models can be a useful tool to assist large‐scale population screening for the early detection of cognitive impairment and patient monitoring over time.
The use of digital technologies may help to diagnose Alzheimer’s Disease (AD) at the pre-symptomatic stage. However, before implementation into clinical practice, digital measures (DMs) need to be evaluated for their diagnostic benefit compared to established questionnaire-based assessments, such as the Mini-Mental State Examination (MMSE) for cognition and Functional Activity Questionnaire (FAQ) for daily functioning. Moreover, the quantitative and qualitative relationship of DMs to these well understood scores needs to be clarified to aid interpretation. In this work we analyzed data from 148 subjects, 58 cognitively normal and 90 at different stages of the disease, which had performed a smartphone based virtual reality game to assess cognitive function. In addition, we used clinical data from Alzheimer’s Disease Neuroimaging Initiative (ADNI). We employed an Artificial Intelligence (AI) based approach to elucidate the relationship of DMs to questionnaire-based cognition and functional activity scores. In addition, we used Machine Learning (ML) and statistical methods to assess the diagnostic benefit of DMs compared to questionnaire-based scores. We found non-trivial relationships between DMs, MMSE, and FAQ which can be visualized as a complex network. DMs, in particular those reflecting scores of individual tasks in the virtual reality game, showed a better ability to discriminate between different stages of the disease than questionnaire-based methods. Our results indicate that DMs have the potential to act as a crucial measure in the early diagnosis and staging of AD.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.