Identifying preclinical Alzheimer’s disease in the community with brief automated hand movement analysis: validation against plasma ptau181 and asymptomatic episodic memory decline

Abstract

AbstractBackgroundDementia prevention and drug development is hindered by the lack of low‐cost population‐level tests to help identify preclinical Alzheimer’s disease (AD) in the community. Recent evidence suggests that precise analysis of hand movements may detect motor changes indicative of preclinical AD. The objective was to evaluate how TAS Test, a new online automated hand movement test, predicts preclinical AD biomarkers (plasma ptau181 and subtle episodic memory decline) in a cognitively asymptomatic cohort of older adults.MethodParticipants completed TAS Test online at home using their own computer without researcher assistance: a series of 10‐30 second index finger‐key and index finger‐thumb tapping tests recorded with a keyboard and/or webcam. Movement features including frequency, rhythm, pauses and accuracy were extracted. Participants also completed online tests of episodic memory, CANTAB Paired Associate Learning (CANTAB). A subset provided blood samples for ptau181 analysis. Accuracy of linear regression models comprising hand motor features to predict PAL scores and ptau181 levels, adjusted for confounding, was compared to null models (with only confounders: age, gender, education level, anxiety and depression) using R2adj and AIC. ΔAIC > 2 denotes statistical difference.Result1,228 adults (mean (SD) age, 65.8 (7.4) years; 73.0% female) completed TAS Test and CANTAB; 459 underwent ptau181 analysis. All keyboard tests improved prediction of asymptomatic episodic memory decline; the 3 step‐ (ΔAIC = 11.2; R2adj = 8.1%) and alternate‐ key (ΔAIC = 3.3; R2adj = 7.5%) tests ranked highest and were the only keyboard tests to improve prediction of ptau181 (3 step ΔAIC = 7.0; R2adj = 17.8%; alternate key ΔAIC = 3.4; R2adj = 17.4%). All finger‐thumb motor measures improved predictions of CANTAB scores and ptau181 levels; the highest performing tests were dominant hand tapping (CANTAB ΔAIC = 2.9; R2adj = 8.2%; ptau181 ΔAIC = 2.4; R2adj = 12.9%) and both hands dual‐task tapping (CANTAB ΔAIC = 3.0; R2adj = 6.8%; ptau181 ΔAIC = 8.7; R2adj = 11.9%).ConclusionTAS Test provides an accessible brief home‐based test for identifying preclinical AD risk. This novel approach holds potential as a pre‐screening/enrichment tool for identifying at risk cohorts for further investigation.