Performance of the CANTAB ‐ Paired Associates Learning task in an unsupervised online setting: Results from the Brain Health Registry.

Background: The aim of this study was to evaluate the feasibility and usability of collecting data from an unsupervised online version of the Paired Associates Learning (PAL) task from the Cambridge Neuropsychological Test Automated Battery (CANTAB™) in adults enrolled in the Brain Health Registry (BHR). Method: BHR is an online cognitive aging registry for referral and longitudinal unsupervised assessment. The PAL assesses visual associative learning and memory and was administered in BHR between 03/2021‐07/2022. PAL variables included in this analysis were 'Memory Score' (patterns correctly identified at first attempt for each difficulty level) and 'Total Errors', adjusted for level reached. We analyzed PAL task completion rate and characteristics of PAL completers and used linear regressions to assess whether cross‐sectional PAL Memory Score and PAL Total Errors were associated with participant data (age, gender, education, self‐ and study partner‐reported cognitive complaints (Everyday Cognition scale (ECog)), self‐reported memory concern, and self‐reported depressive symptom severity (PHQ‐9), self‐report of Mild Cognitive Impairment (MCI)). Result: As of July 2022, 14,520 participants aged 18+ years have completed PAL at least once. Participants were on average 66.3 years old (SD = 11.3), 72.4% female gender, 92.8% self‐identified as White, and had a mean of 16.6 years of education (SD = 2.27). BHR received 374 requests for study support regarding PAL, of which 36% (135) were related to device compatibility. Mean PAL scores were similar to those of previous, supervised studies (Figure 1). Worse PAL performance was significantly associated with being older, female gender, having less education, more self‐ and study partner‐reported ECog, self‐reported memory concern, more depressive symptoms, and self‐report of MCI (Table 1). Conclusion: Our findings provide evidence for the feasibility of collecting PAL data in an unsupervised online setting and construct validity of cross‐sectional BHR PAL. In terms of usability, there was a minimal need for support. Future studies will investigate the predictive value of BHR PAL in identifying adults with self‐reported MCI, longitudinal BHR PAL performance and associations with in‐clinic assessments and Alzheimer's disease biomarkers. Our sample lacked diversity. Therefore, a future direction is to increase accessibility of PAL to ethnoculturally and socioeconomically diverse communities. [ABSTRACT FROM AUTHOR]