Variability in Cognitive Performance on Mobile Devices Is Sensitive to Mild Cognitive Impairment: Results From the Einstein Aging Study

Abstract

Background and Objective: Within-person variability in cognitive performance has emerged as a promising indicator of cognitive health with potential to distinguish normative and pathological cognitive aging. We use a smartphone-based digital health approach with ecological momentary assessments (EMA) to examine differences in variability in performance among older adults with mild cognitive impairment (MCI) and those who were cognitively unimpaired (CU).Method: A sample of 311 systematically recruited, community-dwelling older adults from the Einstein Aging Study (Mean age = 77.46 years, SD = 4.86, Range = 70–90; 67% Female; 45% Non-Hispanic White, 40% Non-Hispanic Black) completed neuropsychological testing, neurological assessments, and self-reported questionnaires. One hundred individuals met Jak/Bondi criteria for MCI. All participants performed mobile cognitive tests of processing speed, visual short-term memory binding, and spatial working memory on a smartphone device up to six times daily for 16 days, yielding up to 96 assessments per person. We employed heterogeneous variance multilevel models using log-linear prediction of residual variance to simultaneously assess cognitive status differences in mean performance, within-day variability, and day-to-day variability. We further tested whether these differences were robust to the influence of environmental contexts under which assessments were performed.Results: Individuals with MCI exhibited greater within-day variability than those who were CU on ambulatory assessments that measure processing speed (p < 0.001) and visual short-term memory binding (p < 0.001) performance but not spatial working memory. Cognitive status differences in day-to-day variability were present only for the measure of processing speed. Associations between cognitive status and within-day variability in performance were robust to adjustment for sociodemographic and contextual variables.Conclusion: Our smartphone-based digital health approach facilitates the ambulatory assessment of cognitive performance in older adults and the capacity to differentiate individuals with MCI from those who were CU. Results suggest variability in mobile cognitive performance is sensitive to MCI and exhibits dissociative patterns by timescale and cognitive domain. Variability in processing speed and visual short-term memory binding performance may provide specific detection of MCI. The 16-day smartphone-based EMA measurement burst offers novel opportunity to leverage digital technology to measure performance variability across frequent assessments for studying cognitive health and identifying early clinical manifestations of cognitive impairment.

Accounting for retest effects in cognitive testing with the Bayesian double exponential model via intensive measurement burst designs

“…Also, consistently across the two tasks, individuals with more years of education exhibited less variability. Paired with our previous findings that intra-individual variation in performance predicts MCI status, this may suggest that day-to-day variation reflects individual differences in cognitive reserve ( Cerino et al, 2021 ).…”

“…While the primary aim of BDEM is to disentangle learning features from peak performance (with the goal of modeling asymptotic change over time), each model parameter may also be of interest for understanding the dynamics of cognitive change. For this reason, we also quantify individual differences, in a multilevel framework, not only in terms of peak performance and changes therein, but also for example in terms of learning rate, and intra-individual variability in performance, and test whether these are linked to cognitive aging ( Lövdén et al, 2007 ) or MCI status ( Cerino et al, 2021 ).…”

Digital Cognitive Biomarker for Mild Cognitive Impairments and Dementia: A Systematic Review

“…Table S5: Diagnostic performance of digital cognitive biomarkers for cognitive impairments (CI). Table S6: Quality assessment based on The Newcastle–Ottawa Scale (NOS) [ 88 , 89 , 90 , 91 , 92 , 93 , 94 , 95 , 96 , 97 , 98 , 99 , 100 , 101 , 102 , 103 , 104 , 105 , 106 , 107 , 108 , 109 , 110 , 111 , 112 , 113 , 114 , 115 , 116 , 117 ]. Figure S1: Between-group mean difference in cognitive outcomes measured by computerized tests in mild cognitive impairments (MCI) and normal controls.…”

Ambulatory Assessment in Concussion Clinical Care and Rehabilitation

“…Of note, Patient A exhibits greater session-to-session, intra-individual variation than Patient B. Inconsistency in speeded performance in other populations (e.g., older adults) has been identified as a risk factor for cognitive impairment ( 31 – 33 ). Recently, in a study using M2C2, we found that older adults with mild cognitive impairment exhibit significantly greater session-to-session and day-to-day variation in Symbol Search task performance than cognitively-unimpaired older adults ( 34 ). A primary aim of the MNCH project is to determine if RT inconsistency is a clinically-relevant marker of cognitive impairment following concussion.…”