Cognition 'In the Wild': using smartphones to assess cognitive variability in healthy adults and individuals with Parkinson's disease
OA Version
Citation
Abstract
The prevalence of neurodegenerative disorders affecting cognition is rising as the population ages, underscoring the need for accessible methods of assessment. Repeated remote testing in one’s context may reveal cognition’s dynamics and determinants and advance ecologically valid precision medicine.
The hypotheses of Study 1 were that cognition would vary with context, and that individual differences would predict patterns of cognitive variability. 177 young adults (mean age 19.2, 100 women) were prompted via smartphone app 5x/day for 10 days to complete brief cognitive tasks and surveys about their context. Psychometric analyses showed that the smartphone working memory task, more than the executive function task, had greater between-subject reliability and within-subject variability in performance. Multi-level modeling revealed no contextual variables that were directly predictive of smartphone performance. There was indication of significant between person differences in the association between momentary motivation and working memory, but no between-person moderators of this relation were identified.
Study 2 proposed the same hypotheses as Study 1 while assessing the feasibility of smartphone cognitive assessment in Parkinson’s disease (PD) and convergent validity between performance on the smartphone tasks and traditional neuropsychological tests. Twenty-seven participants with mild-moderate PD (mean age 63.2, 13 women) completed surveys and games 5x/day for 10 days. The results supported both hypotheses. Response rate to prompts was high, demonstrating feasibility of the approach. Between-subject reliability was high on both games. Within-subject variability was higher for working memory. Multi-level modeling indicated that performance was better on smartphone working memory when participants reported being “home” and having recently exercised, particularly in individuals who endorsed exercise less frequently. Multi-level modeling also revealed that less self-reported daytime sleepiness, and lower PD symptom burden, predicted a stronger association between time of day and smartphone test performance. Strong convergent validity was seen between traditional tests and smartphone working memory but not executive function.
Together these findings support the use of repeat smartphone assessments to understand how context affects cognitive performance. Further development of this assessment method could be useful in increasing sensitivity and specificity regarding cognitive dysfunction, and ultimately lead to personalized treatment recommendations.