Convergent and clinical validity of a remote smartphone‐based self‐assessment of cognition, function, and behavior in early Alzheimer’s disease

Abstract

AbstractBackgroundTrials in early Alzheimer’s disease (eAD) require the quantification of cognitive and functional progression. Compared with traditional in‐clinic assessments, remote assessments via digital health technologies (DHT) enable the collection of a broader palette of cognitive and functional measures at higher frequency and in patients' home environments. However, these must demonstrate validity against standard clinical outcome measures and relevant biomarkers. To this end, the present study aims to determine the reliability, convergent clinical, and functional neuroanatomic validity of a novel DHT, the Alzheimer’s Disease Digital Assessment Suite (AD‐DAS), for individuals on the AD continuum.MethodIn total, 123 participants from 2 countries across 5 sites (3 USA, 2 Spain) participated (https://www.isrctn.com/ISRCTN17035495): 32 healthy controls (HC), 31 amyloid‐PET negative and 30 amyloid‐PET positive subjective cognitive decline (SCDn, SCDP, respectively), and 30 eAD. All participants completed standard in‐clinic neuropsychological assessments and remotely performed the AD‐DAS tasks daily for 28 days without supervision. AD‐DAS comprised 9 active tasks of cognitive/motor functioning and 4 survey questionnaires. Convergent and clinical validity analysis were performed using linear and proportional odds logistic regression modeling with age, sex, education, and site as covariates, respectively. Functional neuroanatomic validity was tested with voxel‐based morphometry (VBM) analyses with anatomic MRIs.ResultOutcome metrics from all AD‐DAS cognitive/motor tasks correlated with their corresponding clinical comparators (ρpartial (range) = 0.16 ‐ 0.52), supporting their convergent validity. Similarly, outcome metrics from all tasks significantly differentiated the eAD from all the other groups in the expected directions (odds ratio (range) = 0.01 ‐ 61), supporting their clinical validity. The reliability of the outcome metrics ranged from good to excellent (interclass correlation coefficient (range) = 0.71 ‐ 0.91). VBM analyses provided independent and largely confirmatory results, further validating the AD‐DAS.ConclusionThe AD‐DAS remote, smartphone‐based assessments of cognition show good to excellent test‐retest reliability, and preliminary analyses indicate that they may show convergent, clinical and exploratory functional neuroanatomic validity. Thus, these results provide a foundation for building towards the overarching goals of remote screening, prediction of progression, and monitoring of cognitive decline for future clinical trials and clinical care for preclinical and eAD.