INS NYC 2024 Program

Remote cognitive assessment has been recognized as a cost-effective, practical method for capturing longitudinal data in clinical research. Test My Brain (TMB) from the Many Brains Project presents a platform for self-administered, performance-based online neuropsychological assessments. While TMB has been deemed useful as a clinical tool under supervised administration, its utility has yet to be established for self-guided remote cognitive assessment. Variables such as device and input type (i.e., mouse versus touchscreen), user interface, screen size, and operating system (OS) have been shown to adversely affect performance outcomes. Further, it is challenging to discern performance attributes concerning engagement or other testing behaviors in unsupervised online settings. Thus, remote cognitive assessment may present interpretive barriers related to these issues, particularly among impaired individuals. Here, we aim to evaluate the feasibility of remote administration for TMB cognitive measures in individuals with a history of stroke.

Adult patients with a history of stroke, confirmed by ICD-10 codes within medical records, were recruited from the Johns Hopkins Hospital for participation in a prospective observational research study. Participants completed remote serial cognitive assessments [Verbal Paired Associates, Trails A/B, Fast Choices] via TMB. Only raw score variables were analyzed for cognitive measures, given that the original normative data were unstratified. Participants had four opportunities to complete cognitive assessments over ten months (at baseline and every three months thereafter). Data analysis consisted of reporting descriptive statistics regarding participant device type, completion rates, score distribution, measures of variability, and central tendency over time.

Eighty-three adults with a history of stroke were enrolled. Of these, 38 (45.78%) completed at least one round of cognitive testing (Mean months post-stroke=24.18, SD=42.58; Mean age=60, SD=12.6). We observed within-participant missingness; only 11 (28.95%) completed assessments across all time points. Of n=38, there were no statistically significant differences in participant demographics. Regarding user device and input types, 57.9% completed assessments with touchscreen devices using MacOS or iOS. Means and standard deviations for each task are as follows: Verbal Paired Associates=12.26(4.29); Fast Choices=1,802.68(648.84) milliseconds; Trails A=73,813.78(91,778.73) milliseconds; Trails B=150,674.52 (253,969.3) milliseconds.

Results indicate poor adherence with remote self-guided cognitive assessment in this population and significant variability in scores obtained. These findings underscore the challenges of using TMB for remote self-guided cognitive assessment, particularly among post- acute stroke survivors. Future studies should examine longitudinal, self-administered TMB data to elucidate factors influencing assessment performance, such as cognitive impairment and user-technology relationships. Additionally, prioritizing the development of TMB performance validity indices may quantify the effects of effort. Investigating these variables could further characterize the feasibility of TMB for remote cognitive assessment and inform data quality standards related to this instrument.