Reliability, validity, and feasibility of a smartphone‐based cognitive assessment for preclinical Alzheimer disease

Abstract Background Smartphone‐based ecological momentary assessments (EMA) have the potential to capture subtle changes in cognition associated with preclinical Alzheimer disease (AD) in older adults. The Ambulatory Research in Cognition (ARC) smartphone application is based on EMA principles and administers brief tests of associative memory, processing speed, and working memory ∼4 times per day over 7 consecutive days. Participants perform these 7‐day cycles every ∼6 months, longitudinally. ARC was designed to be performed unsupervised, on participants’ personal devices, and in their everyday environments in order to increase accessibility, testing, and reliability. Method We evaluated the reliability, validity, and feasibility of ARC in a large, well‐characterized sample of cognitively normal older adults (ages 65‐97) and individuals with very mild dementia (ages 61‐88). Participants completed at least one 7‐day cycle of ARC testing along with conventional cognitive assessments. Most participants also had cerebrospinal fluid, amyloid and tau PET, and structural MRI data available. The reliability, validity, and feasibility of ARC as an unsupervised, high‐frequency cognitive assessment tool using participants’ personal smartphones was tested by examining (1) between‐subjects and test‐retest reliability, (2) correlations with age, conventional cognitive measures, and AD‐related biomarkers, and (3) adherence and attrition rates of older adults with varying technology familiarity. Result First, ARC tasks demonstrated good reliability such that between‐person reliability across the 7‐day cycle and test‐retest reliabilities at 6‐month and 1‐year follow‐ups all exceeded intraclass correlations of 0.85. Second, ARC demonstrated construct validity as evidenced by correlations with conventional cognitive measures. Third, ARC measures correlated with AD biomarker burden at baseline to a similar degree as conventional cognitive measures. Finally, high adherence rates and low attrition rates indicated that ARC was feasible and, to some extent, enjoyable, for older adult participants. Conclusion Ultimately, these results suggest that ARC is a reliable and valid measure of cognition in older adults at risk for AD and is a feasible tool for assessing subtle cognitive changes associated with the earliest stages of AD.