A remote digital memory composite to detect cognitive impairment in memory clinic samples in unsupervised settings using mobile devices.

Remote monitoring of cognition holds the promise to facilitate case-finding in clinical care and the individual detection of cognitive impairment in clinical and research settings. In the context of Alzheimer's disease, this is particularly relevant for patients who seek medical advice due to memory problems. Here, we develop a remote digital memory composite (RDMC) score from an unsupervised remote cognitive assessment battery focused on episodic memory and long-term recall and assess its construct validity, retest reliability, and diagnostic accuracy when predicting MCI-grade impairment in a memory clinic sample and healthy controls. A total of 199 participants were recruited from three cohorts and included as healthy controls (n = 97), individuals with subjective cognitive decline (n = 59), or patients with mild cognitive impairment (n = 43). Participants performed cognitive assessments in a fully remote and unsupervised setting via a smartphone app. The derived RDMC score is significantly correlated with the PACC5 score across participants and demonstrates good retest reliability. Diagnostic accuracy for discriminating memory impairment from no impairment is high (cross-validated AUC = 0.83, 95% CI [0.66, 0.99]) with a sensitivity of 0.82 and a specificity of 0.72. Thus, unsupervised remote cognitive assessments implemented in the neotiv digital platform show good discrimination between cognitively impaired and unimpaired individuals, further demonstrating that it is feasible to complement the neuropsychological assessment of episodic memory with unsupervised and remote assessments on mobile devices. This contributes to recent efforts to implement remote assessment of episodic memory for case-finding and monitoring in large research studies and clinical care.

Feasibility of Digital Memory Assessments in an Unsupervised and Remote Study Setting.

Sensitive and frequent digital remote memory assessments via mobile devices hold the promise to facilitate the detection of cognitive impairment and decline. However, in order to be successful at scale, cognitive tests need to be applicable in unsupervised settings and confounding factors need to be understood. This study explored the feasibility of completely unsupervised digital cognitive assessments using three novel memory tasks in a Citizen Science project across Germany. To that end, the study aimed to identify factors associated with stronger participant retention, to examine test-retest reliability and the extent of practice effects, as well as to investigate the influence of uncontrolled settings such as time of day, delay between sessions or screen size on memory performance. A total of 1,407 adults (aged 18-89) participated in the study for up to 12 weeks, completing weekly memory tasks in addition to short questionnaires regarding sleep duration, subjective cognitive complaints as well as cold symptoms. Participation across memory tasks was pseudorandomized such that individuals were assigned to one of three memory paradigms resulting in three otherwise identical sub-studies. One hundred thirty-eight participants contributed to two of the three paradigms. Critically, for each memory task 12 independent parallel test sets were used to minimize effects of repeated testing. First, we observed a mean participant retention time of 44 days, or 4 active test sessions, and 77.5% compliance to the study protocol in an unsupervised setting with no contact between participants and study personnel, payment or feedback. We identified subject-level factors that contributed to higher retention times. Second, we found minor practice effects associated with repeated cognitive testing, and reveal evidence for acceptable-to-good retest reliability of mobile testing. Third, we show that memory performance assessed through repeated digital assessments was strongly associated with age in all paradigms, and individuals with subjectively reported cognitive decline presented lower mnemonic discrimination accuracy compared to non-complaining participants. Finally, we identified design-related factors that need to be incorporated in future studies such as the time delay between test sessions. Our results demonstrate the feasibility of fully unsupervised digital remote memory assessments and identify critical factors to account for in future studies.

Bayesian modeling of item heterogeneity in dichotomous recognition memory data and prospects for computerized adaptive testing.

Most current models of recognition memory fail to separately model item and person heterogeneity which makes it difficult to assess ability at the latent construct level and prevents the administration of adaptive tests. Here we propose to employ a General Condorcet Model for Recognition (GCMR) in order to estimate ability, response bias and item difficulty in dichotomous recognition memory tasks. Using a Bayesian modeling framework and MCMC inference, we perform 3 separate validation studies comparing GCMR to the Rasch model from IRT and the 2-High-Threshold (2HT) recognition model. First, two simulations demonstrate that recovery of GCMR ability estimates with varying sparsity and test difficulty is more robust and that estimates improve from the two other models under common test scenarios. Then, using a real dataset, face validity is confirmed by replicating previous findings of general and domain-specific age effects (Güsten et al. in Cortex 137:138-148, https://doi.org/10.1016/j.cortex.2020.12.017 , 2021). Using cross-validation we show better out-of-sample prediction for the GCMR as compared to Rasch and 2HT model. In addition, we present a hierarchical extension of the model that is able to estimate age- and domain-specific effects directly, without recurring to a two-stage procedure. Finally, an adaptive test using the GCMR is simulated, showing that the test length necessary to obtain reliable ability estimates can be significantly reduced compared to a non-adaptive procedure. The GCMR allows to model trial-by-trial performance and to increase the efficiency and reliability of recognition memory assessments.

Age impairs mnemonic discrimination of objects more than scenes: A web-based, large-scale approach across the lifespan.

Recent findings suggest that the effect of aging on recognition memory is modality-dependent, affecting memory for objects and scenes differently. However, the lifespan trajectory of memory decline in these domains remains unclear. A major challenge for assessing domain-specific trajectories is the need to utilize different types of stimuli for each domain (objects and scenes). We tested the large sample required to cover much of the adult lifespan using a large stimulus range via web-based assessments. 1554 participants (18-77 years) performed an online mnemonic discrimination task, tested on a pool of 2708 stimuli (Berron et al., 2018). Using corrected hit-rate (Pr) as a measure of performance, we show age-related decline in mnemonic discrimination in both domains, notably with a stronger decline in object memory, driven by a linear increase in the false recognition rate with advancing age. These data are the first to identify a linear age-related decline in mnemonic discrimination and a stronger, linear trajectory of decline in the object domain. Our data can inform basic and clinical memory research on the effects of aging on memory and help advancing the implementation of digital cognitive research tools.

Conventional metaphors in longer passages evoke affective brain response.

Conventional metaphorical sentences such as She's asweetchild have been found to elicit greater amygdala activation than matched literal sentences (e.g., She's akindchild). In the present fMRI study, this finding is strengthened and extended with naturalistic stimuli involving longer passages and a range of conventional metaphors. In particular, a greater number of activation peaks (four) were found in the bilateral amygdala when passages containing conventional metaphors were read than when their matched literal versions were read (a single peak); while the direct contrast between metaphorical and literal passages did not show significant amygdala activation, parametric analysis revealed that BOLD signal changes in the left amygdala correlated with an increase in metaphoricity ratings across all stories. Moreover, while a measure of complexity was positively correlated with an increase in activation of a broad bilateral network mainly involving the temporal lobes, complexity was not predictive of amygdala activity. Thus, the results suggest that amygdala activation is not simply a result of stronger overall activity related to language comprehension, but is more specific to the processing of metaphorical language. SIGNIFICANCE STATEMENT: This work is the first to show that conventional metaphorical language in naturalistic longer passages that includes a range of metaphors elicits more activation in the amygdala-an area recognized to be involved in emotional processing-than carefully matched literal control passages. We probe this finding with parametric analyses using a measure of syntactic complexity and subjective judgments of metaphoricity. While complexity correlates with more overall bilateral activation of the temporal lobes, it does not correlate with amygdala activation. Instead, amygdala activation correlates with metaphoricity, suggesting that the increase in emotional salience is specific to metaphoricity and is not simply a result of an overall increase in brain activity in regions associated with language comprehension.