Assessing everyday functioning using a novel smartphone app in cognitively normal adults: A pilot study of the Assessment of Smartphone Everyday Tasks (ASSET)

Background Cognitively complex everyday tasks may be affected in early stages of Alzheimer’s disease. Performance‐based digital assessments could facilitate more precise measurement of subtle changes in everyday functioning. We piloted a novel smartphone app ‘Assessment of Smartphone Everyday Tasks’ (ASSET), to investigate its utility for measuring everyday functioning in cognitively normal older adults. ASSET consists of Patient Portal and Calendar tasks for renewing prescriptions, making payments, and scheduling urgent and recurring appointments (Figure1). It tracks response time and accuracy. Method ASSET was piloted in 34 participants (n = 20 cognitively normal young adults (YN): 21.3±2.6 years, 85%female, median 14.5 years of education; n = 14 cognitively normal older adults (ON), 75.1±4.8 years, 71%female, median 18 years of education). At baseline, participants completed the first version of ASSET and the Preclinical Alzheimer’s Cognitive Composite (PACC5) in the research clinic. ON participants and their study partners completed the Alzheimer’s Disease Cooperative Study Activities of Daily Living – Prevention Instrument (ADCS ADL‐PI). We calculated accuracy rates (correct responses/task duration) for both tasks and computed Pearson’s r correlation coefficients between subtasks (internal consistency) and with PACC5 and ADCS ADL‐PI (construct validity). To investigate test‐retest reliability, participants subsequently completed the first and five alternate versions of ASSET remotely on a fixed, biweekly schedule. Result At baseline, the Patient Portal and Calendar tasks were highly correlated (r = 0.78, 95% confidence interval (95%CI) = [0.60,0.89]). YN performed better on both tasks than ON, completing them faster and making fewer errors (Patient Portal: Cohen’s d = ‐3.0, 95%CI = [‐4.2, ‐1.8]; Calendar: d = ‐1.6, 95%CI = [‐2.4, ‐0.7]). Performance was moderately correlated with PACC5 (Patient Portal: r = 0.59, 95%CI = [0.49,0.67]; Calendar: r = 0.48, 95%CI = [0.37,0.58]). Among ON, performance on ASSET was not significantly correlated with either self‐ or study partner‐reported ADCS ADL‐PI. Over repeated assessments, Patient Portal accuracy increased, while Calendar accuracy increased but dropped at the last assessment among all participants (see Figure2). Conclusion Initial results show that ASSET may have good internal consistency and good ability to distinguish young from older cognitively normal adults. Future research should aim to corroborate