A Transformer Approach for Cognitive Impairment Classification and Prediction.

INTRODUCTION: Early classification and prediction of Alzheimer disease (AD) and amnestic mild cognitive impairment (aMCI) with noninvasive approaches is a long-standing challenge. This challenge is further exacerbated by the sparsity of data needed for modeling. Deep learning methods offer a novel method to help address these challenging multiclass classification and prediction problems. METHODS: We analyzed 3 target feature-sets from the National Alzheimer Coordinating Center (NACC) dataset: (1) neuropsychological (cognitive) data; (2) patient health history data; and (3) the combination of both sets. We used a masked Transformer-encoder without further feature selection to classify the samples on cognitive status (no cognitive impairment, aMCI, AD)-dynamically ignoring unavailable features. We then fine-tuned the model to predict the participants' future diagnosis in 1 to 3 years. We analyzed the sensitivity of the model to input features via Feature Permutation Importance. RESULTS: We demonstrated (1) the masked Transformer-encoder was able to perform prediction with sparse input data; (2) high multiclass current cognitive status classification accuracy (87% control, 79% aMCI, 89% AD); (3) acceptable results for 1- to 3-year multiclass future cognitive status prediction (83% control, 77% aMCI, 91% AD). CONCLUSION: The flexibility of our methods in handling inconsistent data provides a new venue for the analysis of cognitive status data.

Neural Underpinnings of Learning in Dementia Populations: A Review of Motor Learning Studies Combined with Neuroimaging.

The intent of this review article is to serve as an overview of current research regarding the neural characteristics of motor learning in Alzheimer disease (AD) as well as prodromal phases of AD: at-risk populations, and mild cognitive impairment. This review seeks to provide a cognitive framework to compare various motor tasks. We will highlight the neural characteristics related to cognitive domains that, through imaging, display functional or structural changes because of AD progression. In turn, this motivates the use of motor learning paradigms as possible screening techniques for AD and will build upon our current understanding of learning abilities in AD populations.

Effects of heat strain on cognitive function among a sample of miners.

Heat stress is associated with workplace injuries, likely through a combination of fatigue, reduced cognitive function, and thermal discomfort. The purpose of this study was to evaluate four cognitive tasks for sensitivity to heat stress. Eight participants performed treadmill exercise followed by assessments of serial reaction time (RT), Stroop effect, verbal delayed memory, and continuous performance working memory in an environmental chamber. A control (21.1 °C) trial, and "Hot 1" and "Hot 2" (both 37.8 °C) trials were run sequentially on two separate days to evaluate the four cognitive tasks. Heat strain (comparing Hot 1 and Hot 2 with the control trial) resulted in impairments in the serial RT test response and Stroop accuracy. Delayed memory was impacted only in the Hot 2 trial compared with the control trial. Given the demonstrated impact of heat on cognitive processes relevant to workers' real-world functioning in the workplace, understanding how to assess and monitor vigilant attention in the workplace is essential.

The Measurement of Everyday Cognition (ECog): Revisions and Updates.

INTRODUCTION: The Everyday Cognition scale (ECog), a measure of everyday functioning developed in 2008, is sensitive to early detection and progression of neurodegenerative disease. The goal was to update ECog item content to ensure relevancy to contemporary older adults from diverse backgrounds. METHODS: Participants included 44 culturally diverse older adults (18 with normal cognition, 11 with mild cognitive impairment) and their study partners. Item understandability and relevance was evaluated using iterative interviewing methods that were analyzed using standard qualitative methods. On the basis of this information, items were modified, deleted, or developed as needed. RESULTS: Of the 39 original items, 19 were revised, 3 new items were added (primarily to cover contemporary activities such as the use of technology), and 1 was deleted. The revised version (ECog-II) includes 41 items. DISCUSSION: To ensure strong psychometric properties, and to facilitate harmonization of previously collected data, we preserved well over half of the items. Future work will validate the revised ECog by measuring associations with neuropsychological performance, external measures of disease, and other functional measures. Overall, the revised ECog will continue to be a useful tool for measuring cognitively relevant everyday abilities in clinical settings and intervention clinical trials.

The night out task and scoring application: an ill-structured, open-ended clinic-based test representing cognitive capacities used in everyday situations.

OBJECTIVE: The night out task (NOT) was developed as a naturalistic, open-ended, multitasking measure that requires individuals to complete eight subtasks comparable to those encountered during real-world functioning (e.g., pack travel bag, prepare tea). We examined psychometric properties and administration feasibility of this direct observation measure within a clinic-like setting using a tablet-based coding application. METHOD: A sample of 148 community-dwelling older adults (82% cognitively healthy; 18% mild cognitive impairment) and 57 younger adults completed the NOT along with other neurocognitive tests and questionnaires. RESULTS: Inter-rater reliability across NOT primary (i.e., time, accuracy, efficiency) and process-related (e.g., error-types, self-corrections) variables was mostly excellent. NOT primary measures showed expected patterns of convergent and discriminant validity with measures of cognition, demographics, and well-being. External validity was established by the NOT ability to distinguish between age and diagnostic (cognitively healthy vs. mild cognitive impairment) groups. Demonstrating incremental validity, the NOT primary variables (execution time in particular) were predictive of self-reported functional abilities and completion quality of in-home everyday tasks over and earlier variables such as demographics, cognition, and mobility. CONCLUSIONS: These findings suggest that the NOT and its app interface, which allows for continuous logging of observations, are a feasible in-clinic measure to assess cognitive capacities important for real-world functioning. With further validation, the NOT may allow for earlier detection of functional difficulties. Understanding errors and strategies used during NOT performance could also have implications for individualized interventions.

The Everyday Compensation (EComp) Questionnaire: Construct Validity and Associations with Diagnosis and Longitudinal Change in Cognition and Everyday Function in Older Adults.

OBJECTIVE: The Everyday Compensation scale (EComp) is an informant-rated questionnaire designed to measure cognitively based compensatory strategies that support both everyday memory and executive function in the context of completing instrumental activities of daily living (IADLs). Although previous findings provided early support for the usefulness of the initial version of EComp, the current paper further describes the development, refinement, and validation of EComp as a new assessment tool of compensation for IADLs. METHOD: Confirmatory factor analysis (CFA) was used to examine its factor structure. Convergent and predictive validity was evaluated by examining the relationship between EComp and markers of disease, including diagnosis, cognitive change, and trajectories of functional abilities. RESULTS: CFA supported a general compensation factor after accounting for variance attributable to IADL domain-specific engagement. The clinical groups differed in compensatory strategy use, with those with dementia using significantly fewer compensatory strategies as compared to individuals with normal cognition or mild cognitive impairment. Greater levels of compensation were related to better cognitive functions (memory and executive function) and functional abilities, as well as slower rates of cognitive and functional decline over time. Importantly, higher levels of compensation were associated with less functional difficulties and subsequently slower rate of functional decline independent of the level of cognitive impairment. CONCLUSIONS: Engagement in compensatory strategies among older adults has important implications for prolonging functional independence, even in those with declining cognitive functioning. Results suggest that the revised EComp is likely to be useful in measuring cognitively based compensation in older adults.

Compensatory strategy use improves real-world functional performance in community dwelling older adults.

OBJECTIVE: Compensatory strategies such as assistive technology, external reminders, and environmental cues may help support instrumental activities of daily living (IADLs) and independence. However, functional ability is most often evaluated in clinical settings where everyday compensation cannot be readily observed. The present study used a novel, real-world evaluation of everyday activities (prospective memory, household chores, complex IADLs, and planning tasks) to examine the impact of compensation. METHOD: Fifty community dwelling older adults were recruited with cognitive status ranging from healthy to mildly impaired. Participants completed a battery of validated performance-based and cognitive tasks, an IADL questionnaire (also completed by an informant), and real-world activities carried out in their own homes. RESULTS: The real-world evaluation demonstrated adequate interrater reliability (intraclass coefficient = 0.92) and construct validity (r = 0.55). Compensation improved real-world task outcome on prospective memory and household chores tasks. Further, cognition emerged as a moderator between compensation and prospective memory task outcome. Participants' ratings on an IADL questionnaire were related to real-world and performance-based functioning whereas informants' ratings were related to performance-based tasks and cognition. CONCLUSIONS: Our results suggest that proxy measures of functional ability (i.e., performance-based, cognitive, and IADL questionnaire measures) do not fully capture the complexity of real-world performance for nondemented community dwelling older adults. Compensation appeared to improve prospective memory performance, particularly for individuals with average to low average cognitive abilities. A difference in IADL questionnaire ratings suggest that participants may be better able to judge their real-world performance than informants. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Effectiveness of a video-based aging services technology education program for health care professionals.

Health care professionals (HCPs) are a critical source of recommendations for older adults. Aging services technologies (ASTs), which include devices to support the health-care needs of older adults, are underutilized despite evidence for improving functional outcomes and safety and reducing caregiver burden and health costs. This study evaluated a video-based educational program aimed at improving HCP awareness of ASTs. Sixty-five HCPs viewed AST videos related to medication management, daily living, and memory. Following the program, participants' objective and perceived AST knowledge improved, as did self-efficacy and anticipated AST engagement. About 95% of participants stated they were more likely to recommend ASTs postprogram. Participants benefitted equally regardless of years of experience or previous AST familiarity. Furthermore, change in self-efficacy and perceived knowledge were significant predictors of engagement change. Overall, the educational program was effective in improving HCPs' awareness of ASTs and appeared to benefit all participants regardless of experience and prior knowledge.

Naturalistic assessment of task interruption in individuals with mild cognitive impairment.

OBJECTIVE: Interruptions are ubiquitous in everyday life, and recovering from interruptions requires several cognitive processes working in tandem. In this study, we assessed the effects of an interruption on the performance of older individuals with and without mild cognitive impairment (MCI) completing everyday tasks in a naturalistic apartment setting. METHOD: Thirty-two persons with MCI and 64 cognitively healthy older adults (HOA) completed two different sets of everyday activities, of which one received an interruption. Participants also completed tests assessing cognitive constructs thought to be important in interruption recovery including retrospective memory, prospective memory, planning, working memory, and executive function. RESULTS: As a consequence of an interruption, participants with MCI took longer to complete primary task demands and made more substitution errors, but did not make more omission errors. In contrast, an interruption led HOAs to make more omission errors, but their time on task was not affected. Results from a hierarchical linear regression suggest that the ecologically valid interruption task time was more predictive of everyday functional status than the traditional neuropsychological measures. CONCLUSIONS: Results suggest that a brief task interruption taxed cognitive resources of both MCI and HOA groups, but was more detrimental to MCI in terms of time on task and total errors committed. Participant groups appeared to use a speed-accuracy trade-off to mitigate negative effects, where HOAs emphasized speed and MCI participants focused on accuracy. Amount of cognitive engagement/disengagement was also theorized to have played a role, where MCI may have maintained information online throughout the interruption, and HOAs disengaged and reengaged resulting in worse reactivation of goals. Although MCI held onto task goals, their execution of details was imperfect over the interruption delay resulting in substitution errors likely due to further taxed executive abilities. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Smart Home-Based Prediction of Multidomain Symptoms Related to Alzheimer's Disease.

As members of an increasingly aging society, one of our major priorities is to develop tools to detect the earliest stage of age-related disorders such as Alzheimer's Disease (AD). The goal of this paper is to evaluate the possibility of using unobtrusively collected activity-aware smart home behavior data to detect the multimodal symptoms that are often found to be impaired in AD. After gathering longitudinal smart home data for 29 older adults over an average duration of 2 years, we automatically labeled the data with corresponding activity classes and extracted time-series statistics containing ten behavioral features. Mobility, cognition, and mood were evaluated every six months. Using these data, we created regression models to predict symptoms as measured by the tests and a feature selection analysis was performed. Classification models were built to detect reliable absolute changes in the scores predicting symptoms and SmoteBOOST and wRACOG algorithms were used to overcome class imbalance where needed. Results show that all mobility, cognition, and depression symptoms can be predicted from activity-aware smart home data. Similarly, these data can be effectively used to predict reliable changes in mobility and memory skills. Results also suggest that not all behavioral features contribute equally to the prediction of every symptom. Future work therefore can improve model sensitivity by including additional longitudinal data and by further improving strategies to extract relevant features and address class imbalance. The results presented herein contribute toward the development of an early change detection system based on smart home technology.

Automatic assessment of functional health decline in older adults based on smart home data.

In the context of an aging population, tools to help elderly to live independently must be developed. The goal of this paper is to evaluate the possibility of using unobtrusively collected activity-aware smart home behavioral data to automatically detect one of the most common consequences of aging: functional health decline. After gathering the longitudinal smart home data of 29 older adults for an average of >2 years, we automatically labeled the data with corresponding activity classes and extracted time-series statistics containing 10 behavioral features. Using this data, we created regression models to predict absolute and standardized functional health scores, as well as classification models to detect reliable absolute change and positive and negative fluctuations in everyday functioning. Functional health was assessed every six months by means of the Instrumental Activities of Daily Living-Compensation (IADL-C) scale. Results show that total IADL-C score and subscores can be predicted by means of activity-aware smart home data, as well as a reliable change in these scores. Positive and negative fluctuations in everyday functioning are harder to detect using in-home behavioral data, yet changes in social skills have shown to be predictable. Future work must focus on improving the sensitivity of the presented models and performing an in-depth feature selection to improve overall accuracy.

Compensation Strategies in Older Adults: Association With Cognition and Everyday Function.

BACKGROUND/RATIONALE: Compensation strategies may contribute to greater resilience among older adults, even in the face of cognitive decline. This study sought to better understand how compensation strategy use among older adults with varying degrees of cognitive impairment impacts everyday functioning. METHODS: In all, 125 older adults (normal cognition, mild cognitive impairment, dementia) underwent neuropsychological testing, and their informants completed questionnaires regarding everyday compensation and cognitive and functional abilities. RESULTS: Cognitively normal and mild cognitive impairment older adults had greater levels of compensation use than those with dementia. Higher levels of neuropsychological functioning were associated with more frequent compensation use. Most importantly, greater frequency of compensation strategy use was associated with higher levels of independence in everyday function, even after accounting for cognition. CONCLUSION: Use of compensation strategies is associated with higher levels of functioning in daily life among older adults. Findings provide strong rational for development of interventions that directly target such strategies.

Subjective cognitive complaints and objective memory performance influence prompt preference for instrumental activities of daily living.

INTRODUCTION: Declines in memory and executive functioning often lead to difficulties completing instrumental activities of daily living (IADLs). Prompting technologies have the potential to help promote aging in place by providing support for the initiation and accurate completion of IADLs. In this study, we evaluate preferences of older adults for different levels of prompting support based on subjective and objective measures of cognitive functioning. METHOD: Participants were 170 community-dwelling older adults split into two cognitive complaint groups: cognitive complaints and few cognitive complaints. After completing six IADL tasks (e.g., organize a pillbox, cook), each participant was asked to make a specific error (e.g., leave stove on) on three of the tasks. They were then prompted to correct the error with one of three different prompt modes: verbal indirect, verbal direct, multimodal verbal direct and video. RESULTS: The cognitive complaints group reported greater preference for the multimodal prompt compared to the few cognitive complaints group. The indirect prompt was the least preferred by both groups. Furthermore, participants who recalled less on objective memory measures preferred more support in terms of prompt mode. Executive functioning did not appear to be related to prompt preference. CONCLUSION: Level of subjective cognitive complaints and objective memory performance may influence amount of support preferred in a prompt.

Neuropsychological test selection for cognitive impairment classification: A machine learning approach.

INTRODUCTION: Reducing the amount of testing required to accurately detect cognitive impairment is clinically relevant. The aim of this research was to determine the fewest number of clinical measures required to accurately classify participants as healthy older adult, mild cognitive impairment (MCI), or dementia using a suite of classification techniques. METHOD: Two variable selection machine learning models (i.e., naive Bayes, decision tree), a logistic regression, and two participant datasets (i.e., clinical diagnosis; Clinical Dementia Rating, CDR) were explored. Participants classified using clinical diagnosis criteria included 52 individuals with dementia, 97 with MCI, and 161 cognitively healthy older adults. Participants classified using CDR included 154 individuals with CDR = 0, 93 individuals with CDR = 0.5, and 25 individuals with CDR = 1.0+. A total of 27 demographic, psychological, and neuropsychological variables were available for variable selection. RESULTS: No significant difference was observed between naive Bayes, decision tree, and logistic regression models for classification of both clinical diagnosis and CDR datasets. Participant classification (70.0-99.1%), geometric mean (60.9-98.1%), sensitivity (44.2-100%), and specificity (52.7-100%) were generally satisfactory. Unsurprisingly, the MCI/CDR = 0.5 participant group was the most challenging to classify. Through variable selection only 2-9 variables were required for classification and varied between datasets in a clinically meaningful way. CONCLUSIONS: The current study results reveal that machine learning techniques can accurately classify cognitive impairment and reduce the number of measures required for diagnosis.

Analysis of verbal fluency ability in Alzheimer's disease: the role of clustering, switching and semantic proximities.

The underlying nature of verbal fluency deficits in Alzheimer's disease (AD) was investigated in this study. Participants were 48 individuals with AD and 48 cognitively healthy older adults. Fluency performance on letter and category tasks was analyzed across two 30-s intervals for total words produced, mean cluster size, and total switches. Compared with the control group, AD participants produced fewer words and switches on both fluency tasks and had a reduced category cluster size. The AD group was differentially impaired on category compared with letter fluency and produced more repetitive responses but fewer category exemplars than controls on the category task. A multidimensional scaling approach revealed that AD participants' semantic maps were similar to controls. Overall, the data suggest that executive abilities involving search and retrieval processes and a reduced availability of semantically related words contributed to the AD group's poorer performance despite similar temporal recall and organizational patterns.

Analysis of verbal fluency ability in amnestic and non-amnestic mild cognitive impairment.

The purpose of this study was to investigate the pattern of performance on letter and category fluency tests of individuals with mild cognitive impairment (MCI). Previous research has suggested that organization strategies, including "clustering" (i.e., groups of related words) and "switching" (i.e., shift from one cluster to another), are important for efficient verbal fluency performance. Participants were 25 individuals with single-domain amnestic MCI (aMCI), 49 with multidomain aMCI, 16 with non-amnestic MCI (naMCI), and 90 cognitively healthy older adults. Fluency performances were analyzed across two 30-s intervals for total words produced, cluster size, and switching. Analyses of variance (ANOVAs) with follow-up tests revealed that the single-domain aMCI group performed comparably with healthy controls on each dependent measure across both fluency tasks. In contrast, the multidomain aMCI group showed performance decrements in total words and switching production compared with healthy controls on both fluency tasks, whereas the naMCI group produced fewer words and switches on letter fluency. Each group generated more words and switches during the first 30-s on both fluency tasks, with the exception of the naMCI group, whose switching on letter fluency did not decrease as the task progressed. As indicated by the single-domain aMCI group's unimpaired performance, our findings demonstrate that verbal fluency performance decreases as domains beyond memory become impaired in MCI. Reduced switching ability, which has been linked to prefrontal executive functioning, contributed the most to the poorer performance of individuals with multidomain MCI and naMCI.

Naturalistic assessment of everyday functioning in individuals with mild cognitive impairment: the day-out task.

OBJECTIVE: The day-out task (DOT), a naturalistic task that requires multitasking in a real-world setting, was used to examine everyday functioning in individuals with mild cognitive impairment (MCI). METHOD: Thirty-eight participants with MCI and 38 cognitively healthy older adult controls prioritized, organized, initiated, and completed a number of subtasks in a campus apartment to prepare for a day out (e.g., determine and gather change for bus, bring a magazine). Participants also completed tests assessing cognitive constructs important in multitasking (i.e., retrospective memory, prospective memory, planning). RESULTS: As compared with controls, the MCI group required more time to complete the DOT and demonstrated poorer task accuracy, performing more subtasks incompletely and inaccurately. Despite poorer DOT task accuracy, the MCI and control groups approached completion of the DOT in a similar manner. For the MCI group, retrospective memory was a unique predictor of the number of subtasks left incomplete and inaccurate, while prospective memory was a unique predictor of DOT sequencing. The DOT measures, but not the cognitive tests, were predictive of knowledgeable informant report of everyday functioning. CONCLUSIONS: THESE findings suggest that difficulty remembering and keeping track of multiple goals and subgoals may contribute to the poorer performance of individuals with MCI in complex everyday situations.