Baseline Reliability and Six‐Month Stability of a Remotely Administered Battery of Smartphone Cognitive Tests

Abstract

BackgroundFrontotemporal dementia (FTD) is a collection of neurodegenerative diseases affecting behavior, cognition, and motor functions. FTD treatment trials will benefit from sensitive assessments capable of detecting and longitudinally tracking cognitive changes. Remote smartphone cognitive tests may improve the sensitivity to detect and monitor cognitive decline via novel methods of data capture and more frequent assessments within ecologically valid environments, while simultaneously reducing patient burden. The first step in determining the utility of these measures is establishing their reliability. We previously reported baseline reliability estimates for the ALLFTD Mobile App cognitive tests in a small pilot sample. Here we extend these results using a larger sample and evaluate six‐month stability.MethodWe enrolled a diagnostically mixed sample of 128 participants (CDR®+NACC‐FTLD = 0 [n = 84]; CDR®+NACC‐FTLD = 0.5 [n = 21]; CDR®+NACC‐FTLD≥1 [n = 23]) from ongoing studies of healthy aging (Hillblom Network) and frontotemporal dementia (ALLFTD). Participants self‐administered five cognitive tests from the ALLFTD Mobile App on their own smartphone; they were asked to complete each test three times within two weeks. Tests included gamified versions of N‐back, Stroop, Flanker, and Go/No‐Go paradigms, and an adaptive short‐term memory test. We evaluated the internal consistency of each test’s first occurrence using Spearman‐Brown split‐half reliability (100 simulations). Test‐retest reliability over the first three baseline replicates (as available) was evaluated by calculating intraclass correlations (ICC). In a preliminary analysis, we tested the six‐month stability of test performance in nine functionally intact participants who completed baseline and follow‐up Flanker, N‐Back, and adaptive memory assessments by calculating Spearman’s rho.ResultSplit‐half reliability estimates were in the good to excellent range (r’s: 0.81–0.97). Test‐retest reliability estimates were in the adequate to excellent range (ICCs: 0.72–0.96). ICC estimates were comparable for participants with and without clinical impairment (CDR®+NACC‐FTLD = 0 (0.74–0.95) and CDR®+NACC‐FTLD >0 (0.67–0.95). Among functionally intact individuals, baseline and 6‐months scores were highly correlated (r’s: 0.76–0.93, p’s<.05).ConclusionOur findings of adequate to excellent internal consistency, test‐retest reliability, and 6‐month stability support the feasibility of remote, smartphone‐based cognitive testing among a diagnostically mixed sample typical of healthy aging and frontotemporal dementia research studies.