Building the Research Path for Identifying and Validating Cognition‐related Digital Biomarkers

Abstract

AbstractBackgroundCognitive dysfunction is a core feature of dementia spectrum disorders but simple screening tests are less sensitive to detect the deficits, especially in the very early stage of disease course using most standardized test instruments. This results in delayed discovery of disease onset and opportunity for early interventions that might be more effective in delaying or arresting progression. Digital phenotyping using a combination or hardware and software solutions provide potentially much more effective tools for helping clinicians identify subtle signs of cognition deficits and track them in a compressed timeline to confirm the accuracy of decreasing slope indicative of a neurodegenerative etiology.MethodsBeginning in 2017, 1,866 participants were recruited from the Dementia Center, Shuang Ho Hospital, New Taipei City, Taiwan and will be invited to use a smartphone application to collect digital phenotypes including information processing speed, response accuracy, error detection, and multitasking capacity and vocal patterns, such as speech pitch, tone, rhythm. Additionally, four groups of 100 new participants each will be recruited; cognitive unimpaired (CU), subjective cognitive decline (SCD), mild cognitive impairment (MCI) and mild dementia. In addition to the smartphone applications, these 400 participants will undergo additional testing that include the cognitive abilities screening instrument (CASI), Clinical Dementia Rating (CDR), Cookie Theft Picture Description, Eye Tracking Test, and gait/balance.ResultsDigital phenotyping profiles will be constructed for the cohort and validated against the smaller sample to determine those metrics that can serve as actual digital biomarkers. For the smaller sample, additional multi‐dimensional profiles will be constructed that add gait parameters (step length, stride length, step width, foot angle) and eye‐tracking (visual fixation changes, speed, duration) and determine to what extent they increase differential diagnostic accuracy.ConclusionsDigital assessment tools can be used for detecting changes in cognitively related behaviors associated with preclinical and early stage clinical dementia. Future research is needed to confirm the efficacy of these digital phenotypes to serve as digital biomarkers that have high prognostic and diagnostic accuracy.