Search

Background: Rural residents often experience multiple factors that increase their risk of cognitive impairment, e.g. less access to formal education and specialized healthcare (Livingston et al., 2020). Early, accurate diagnosis in the mild cognitive impairment stage could save up to 7.9 trillion dollars in health and long‐term care costs by 2050 (Alzheimer's Association, 2021). First year results from this federally‐funded new investigator mentoring program between computer science researchers and a nurse investigator are reported. The purpose was to test the feasibility of a novel approach to detecting early cognitive decline using smartwatch technology among a 92% racially and ethnically diverse rural community challenged by low digital access (51%) and digital literacy (48%). Method: We tested smartwatches programmed with the novel "n‐back" shape test for cognitive assessment, ecological momentary assessment, and movement tracking. Paper‐and pencil measures to assess cognitive risk were the PROMIS ® Applied Cognition, Mini‐Cog (Borson, 2003), and Mini‐MoCA (Nasreddine, 2019). Another novel aspect of the study was hiring the local high school students to mentor older adults in use of smartwatches during home visits. Student engagement with their mentees was monitored by their high school advisor, with oversight by well‐known faith‐based research assistants. Six additional pre‐post intervention surveys included "Technology Activities of Daily Living (T‐ADL) (Muñoz‐Neira, 2012) and "Relating to Older People" (Cherry and Palmore, 2008). Result: Mean years lived rural of the n = 28 enrolled was 60.3 (SD = 18.2), with an average age of 71.6 (SD = 8.1). Results regarding cognitive risk using the n‐back Shape test included correlation with Ecological Momentary Assessment Question 7 ("Right now, my environment is distracting"; r = ‐.35, p =.046) and approached significance with Question 1 ("In the past 2 hours, do you think your mind has been as sharp as usual?"; r =.33, p =.058). Pre‐post correlations in T‐ADL (r =.87) and Applied Cognition (r =.38) were significant (p <.001, p <.04). Conclusion: Testing this novel two‐pronged approach of detecting cognitive risk through smartwatch technology, and hiring high school students to mentor older adults in the home setting yielded implications for ADRD‐related research among larger samples of underrepresented groups. [ABSTRACT FROM AUTHOR]

Background: Electronic memory aids can facilitate positive health outcomes in older individuals experiencing memory difficulties. Despite this, long‐term maintenance of device use is often limited. Delivery of automated boosters has the potential to enhance use of these aids and retention of compensation training goals, but questions remain regarding optimal timing of boosters. For example, if delivered too frequently recipients will habituate. Here we describe our process for developing an automated schedule that will detect when use of an electronic memory aid has started to decline so that an automated booster can be delivered. We start by unobtrusively observing use patterns and generating initial delivery rules based on clinical judgement, and then apply these rules to develop an automated algorithm. Methods: We used three months of post‐training data from 28 older adults with mild cognitive impairment who learned to use an electronic memory and management aid (EMMA) application. Two clinicians examined graphs of each users' daily interactions with EMMA to identify specific days when device interactions had declined and iteratively developed a set of rules that identified the best time points to deliver boosters. The algorithm was then applied to data from an entirely different cohort of 50 older adults engaged in an intervention utilizing the EMMA app. We then compared when the clinician would have boosted vs when the algorithm would deliver a boost. Results: Following 4 iterations of clinicians examining graphs with possible boost points from the first set of users, an algorithm was developed. When applied to the second data set, the algorithm identified 371 boost points, with a false positive rate of 11% and false negative rate of 3%. It was also deemed that the timing of approximately 28% of the correctly identified boost points by the algorithm were off by +/‐ 2 days. Conclusion: Next steps will be to further modify the algorithm and then formally test the impact of boosts at different time points to empirically evaluate their impact on usage. Our ultimate goal is to deliver boosters in real time and as early as possible when usage wanes to improve intervention outcomes. [ABSTRACT FROM AUTHOR]

Abstract Research demonstrated that aging services technologies (ASTs) can reduce caregiver burden. However, ASTs are underutilized by caregivers due to poor awareness. This study evaluated the effectiveness of a video-based educational program to increase caregiver AST knowledge among 43 caregivers. Paired sample t-tests showed positive change in AST knowledge, stigma, and intention to engage in AST activities post-program. Caregivers endorsed highly positive AST views pre- and post-program without statistically significant change. Group by time repeated measures ANOVAs showed younger caregivers (<65 years old) had better knowledge and were more open to ASTs (improved attitude and stigma scores) than older caregivers (>65 years old). A significant interaction revealed caregivers of individuals who had fewer, but not more, domains of functional limitation reported a positive change in AST attitude post-program. The video program appeared to be beneficial to all caregivers, who provided vastly positive program feedback. [ABSTRACT FROM AUTHOR]

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. [ABSTRACT FROM AUTHOR]

Sensor-based time series data can be utilized to monitor changes in human behavior as a person makes a significant lifestyle change, such as progress toward a fitness goal. Recently, wearable sensors have increased in popularity as people aspire to be more conscientious of their physical health. Automatically detecting and tracking behavior changes from wearable sensor-collected physical activity data can provide a valuable monitoring and motivating tool. In this paper, we formalize the problem of unsupervised physical activity change detection and address the problem with our Physical Activity Change Detection (PACD) approach. PACD is a framework that detects changes between time periods, determines significance of the detected changes, and analyzes the nature of the changes. We compare the abilities of three change detection algorithms from the literature and one proposed algorithm to capture different types of changes as part of PACD. We illustrate and evaluate PACD on synthetic data and using Fitbit data collected from older adults who participated in a health intervention study. Results indicate PACD detects several changes in both datasets. The proposed change algorithms and analysis methods are useful data mining techniques for unsupervised, window-based change detection with potential to track users' physical activity and motivate progress toward their health goals. [ABSTRACT FROM AUTHOR]

Background: Difficulties in performing "Instrumental Activities of Daily Living" (IADL) such as managing finances and medications can lead to loss of independence among older adults. Performance‐based assessments allow for an objective and systematic quantification of IADLs utilizing real life paradigms. In an ongoing NIA‐funded study, we evaluated the association of a recently developed performance‐based assessment of functional ability (Night Out Task, NOT, Schmitter‐Edgecombe, 2021) with a performance‐based measure of financial capacity and an informant‐based measure of functional and decision‐making capacity in non‐demented older adults with varying degrees of cognitive impairment. Method: Participants (n = 22; mean age = 76) were classified as healthy control (HC, n = 4), subjective cognitive decline (SCD, n = 9), or mild cognitive impairment (MCI, n = 9) based on clinical criteria. The NOT was used to evaluate the ability to efficiently and accurately plan, multitask and complete eight activities in preparation for a night out. The primary outcome variable for the NOT was completion time. Additionally, participants completed the Financial Capacity Instrument (FCI‐SF‐short form), which assesses financial knowledge and problem‐solving skills (Marson, 2015). Meanwhile, informants completed the Assessment of Functional Capacity Interview (AFCI; Pare et al., 2020) which captures decision making related to home safety, financial, medical, and social competence. Result: NOT scores were normally distributed, as assessed by Shapiro‐Wilk's test (p > 0.05). We observed a statistical trend for group on NOT completion time such that greater clinical impairment was associated with greater completion time: MCI (M = 916.38s, SD = 218.54) > SCD (M = 741.83s, SD = 212) > HC (M = 728.05s, SD = 167.45). Similar trends were observed for other NOT variables. Correct sequencing and overall efficiency of NOT tasks were negatively correlated with AFCI Financial domain, r(21) = ‐0.606 and r(21) = ‐0.487, both P <0.05, respectively. Overall NOT task accuracy was negatively correlated with FCI‐SF Total Score, r(21) = ‐0.528, P <0.05. Conclusion: The NOT showed impressive associations with informant reports of functional and decision‐making capacity and objectively assessed financial capacity. The NOT can potentially distinguish among healthy controls and preclinical dementia groups, offering promise for early detection and characterization of cognitive changes. [ABSTRACT FROM AUTHOR]

Background: American Indians present with increased risk factors for Alzheimer's disease and related dementias (ADRDs). Intraindividual cognitive variability (IICV), which measures the variability in performance on neuropsychological tests within a person at a single time, has been proposed as a noninvasive marker of neurodegeneration and early dementia. In this study, we investigated the hypothesis that higher IICV was related to more features of neurodegeneration on magnetic resonance imaging (MRI) measurements among older American Indians. Method: American Indian participants over 64 years of age who underwent MRI and completed a battery of neuropsychological tests were included in the study (N = 750). Data from three neuropsychological tests were used to compute a multidomain IICV measure of cognition along with IICV measures representing the cognitive domains of memory, executive function and processing speed. Regression models estimated IICV associations with total brain volume, hippocampal volume, and graded white matter disease, adjusting for age, sex, education, BMI, and intracranial volume. Result: Memory, executive function and processing speed, and multidomain IICV measures were associated with total brain volume before correction for multiple comparisons. After correction with Bonferroni, only higher IICV of memory was associated with lower total brain volume (Beta = −0.20; 95%CI −0.332, −0.068; p = 0.009). Furthermore, additional adjustment of mean scores from the tests used for the variability measure resulted in loss of significance for all IICV measures, with only mean scores remaining significant. Measures of IICV were not associated with hippocampal volume or white matter disease. Conclusion: These findings suggest that IICV of memory may be associated with imaging measures related to neurodegenerative injury in older American Indians. However, these associations representing variability in cognition are smaller than those representing mean cognition. More research is needed to understand the clinical and etiologic associations for IICV measures. [ABSTRACT FROM AUTHOR]

Pervasive computing offers an unprecedented opportunity to unobtrusively monitor behavior and use the large amount of collected data to perform analysis of activity-based behavioral patterns. In this paper, we introduce the notion of an activity curve , which represents an abstraction of an individual’s normal daily routine based on automatically-recognized activities. We propose methods to detect changes in behavioral routines by comparing activity curves and use these changes to analyze the possibility of changes in cognitive or physical health. We demonstrate our model and evaluate our change detection approach using a longitudinal smart home sensor dataset collected from 18 smart homes with older adult residents. Finally, we demonstrate how big data-based pervasive analytics such as activity curve-based change detection can be used to perform functional health assessment. Our evaluation indicates that correlations do exist between behavior and health changes and that these changes can be automatically detected using smart homes, machine learning, and big data-based pervasive analytics. [ABSTRACT FROM AUTHOR]

Background: Advances in technology have fueled new opportunities to improve external compensatory devices. The Digital Memory Notebook (DMN) is an iOS tablet application (app) developed in collaboration with older individuals with mild cognitive impairment (MCI) to support functional independence. To understand users for whom DMN training was beneficial, we examined data from a pilot clinical trial for factors that influenced sustained use of this compensatory tool. Method: Twenty‐five participants (M age = 72; M education = 16; 48% female) with MCI learned to use the DMN app in 5‐6 (two‐hour) training sessions delivered within one‐month. Participant use of the DMN app was then monitored through real‐time collection of data usage metrics for an additional three months. Based on app use during the third‐month post‐training, participants were divided into DMN user (N = 13; M daily distinct uses = 4.5) and non‐user (N = 12) groups. Demographics, clinician ratings, baseline clinical measures and DMN usage metrics were all examined. Result: No group differences emerged in demographics, pre‐morbid ability estimate (High Average), or clinician ratings of motivation, comprehension, insight and competency during training sessions. DMN non‐users exhibited lower scores on standardized measures of global cognitive status (Low Average vs. Average) and memory (Borderline vs. Low Average), and self‐reported poorer everyday memory. DMN usage metrics captured during training revealed no group differences in week one (sessions 1‐2). By week two (session 3), DMN non‐users showed fewer daily distinct uses of the DMN (4 vs. 8). During weeks 3 and 4, DMN non‐users continued to show fewer daily distinct uses as well as fewer interactions with all sections of the app (i.e., calendar, daily event scheduling, notes and journal). Non‐users also demonstrated poorer use of advanced app functions during training (e.g., high priority scheduling, reminder alarms, picture use, prospective memory use). Conclusion: Individuals with MCI who were actively using the DMN three‐months post‐intervention were less cognitively impaired, demonstrated greater DMN use as early as week two of training, and made use of advanced functions. Next steps include increasing DMN use through early adaptive learning and continued automated booster training. [ABSTRACT FROM AUTHOR]

Background: Advances in technology have fueled new opportunities to improve external compensatory devices. The Digital Memory Notebook (DMN) is an iOS tablet application (app) developed in collaboration with older individuals with mild cognitive impairment (MCI) to support functional independence. To understand users for whom DMN training was beneficial, we examined data from a pilot clinical trial for factors that influenced sustained use of this compensatory tool. Method: Twenty‐five participants (M age = 72; M education = 16; 48% female) with MCI learned to use the DMN app in 5‐6 (two‐hour) training sessions delivered within one‐month. Participant use of the DMN app was then monitored through real‐time collection of data usage metrics for an additional three months. Based on app use during the third‐month post‐training, participants were divided into DMN user (N = 13; M daily distinct uses = 4.5) and non‐user (N = 12) groups. Demographics, clinician ratings, baseline clinical measures and DMN usage metrics were all examined. Result: No group differences emerged in demographics, pre‐morbid ability estimate (High Average), or clinician ratings of motivation, comprehension, insight and competency during training sessions. DMN non‐users exhibited lower scores on standardized measures of global cognitive status (Low Average vs. Average) and memory (Borderline vs. Low Average), and self‐reported poorer everyday memory. DMN usage metrics captured during training revealed no group differences in week one (sessions 1‐2). By week two (session 3), DMN non‐users showed fewer daily distinct uses of the DMN (4 vs. 8). During weeks 3 and 4, DMN non‐users continued to show fewer daily distinct uses as well as fewer interactions with all sections of the app (i.e., calendar, daily event scheduling, notes and journal). Non‐users also demonstrated poorer use of advanced app functions during training (e.g., high priority scheduling, reminder alarms, picture use, prospective memory use). Conclusion: Individuals with MCI who were actively using the DMN three‐months post‐intervention were less cognitively impaired, demonstrated greater DMN use as early as week two of training, and made use of advanced functions. Next steps include increasing DMN use through early adaptive learning and continued automated booster training. [ABSTRACT FROM AUTHOR]

The article presents a study which examines the effect and pattern of verbal memory encoding and consolidation deficits to patients undergoing traumatic brain injury (TBI) recovery. The study utilizes Item Specific Deficit Approach and Rey Auditory Verbal Learning Test indices of consolidation, encoding, and retrieval deficits among 23 patients. The results show that the patients expressed verbal memory impairment characterized by consolidation deficits and encoding deficits at the same time.

Background: The protocol for 'Brain Boosters', a novel combined compensatory training and lifestyle modification intervention aimed at Alzheimer's disease (AD) risk reduction is presented. Training in compensation will focus on a calendar and tracking system, organizational strategies, and goal setting, all aimed at supporting independent function. These same compensatory tools will be employed to facilitate adoption of lifestyle changes that support brain health (e.g., physical exercise, cognitive exercise, wellbeing exercise) through behavioral monitoring and feedback. A comprehensive suite of digital tools encapsulated in the Electronic Memory and Management Aid (EMMA), an easy to use, interactive application, will be used to facilitate behavioral change and enhance participant motivation. EMMA also allows collection of real‐time data to track intervention adherence and factors that influence adherence. Methods: To capitalize on a critical window of opportunity to intervene, we will target cognitively normal older adults with subjective cognitive concerns (SCC), an established risk for AD and cognitive impairment. The trial will enroll 200 older adults with SCC who will be randomized to our digital application‐supported compensation training and lifestyle modification intervention or to an education only control group that will not use the EMMA app or be taught how to implement the educational material into their daily lives. Both intervention arms will be delivered in a group setting over 6 months, followed by 12 months of unsupervised follow‐up. Results: Specific aims of the project include to: 1) evaluate intervention efficacy on primary outcomes (global cognition and everyday function); secondary outcomes focus on indicators of well‐being, cognitive domains (memory and executive function), physical function, compensation, and health behaviors; 2) evaluate characteristics of treatment responders; 3) evaluate adherence and identify the effective components of the target intervention using a mixed‐method approach; and 4) design machine learning algorithms that use patterns of change in real‐time data metrics to identify incipient declines in treatment adherence and changes in health status. Conclusions: The project is expected to expand understanding of factors that may impact adherence to and outcomes of a preventative intervention leading to optimization of a scalable intervention to reduce dementia risk applicable to diverse populations. [ABSTRACT FROM AUTHOR]

Background: Traditional paper‐based calendars and to‐do lists have limitations related to difficulty retrieving information and no automated way to prompt completion of activities. Such limitations may be overcome using a digital format through organized files, search functions, and alarms. This pilot project trained older adults at risk for cognitive decline to use the Digital Memory Notebook (DMN), a tablet‐based application, to support everyday memory. The primary goal was to obtain preliminary information regarding a 6‐week, group‐based DMN training intervention in terms of acceptability and satisfaction. Method: Ten participants with subject cognitive complaints (SCC), a risk factor for Alzheimer's disease, enrolled in the study. Participants met 2‐hours per week for 6‐weeks. A manualized curriculum covered specific DMN functions and weekly homework/goal‐setting targets. Following the intervention, participants continued to use the DMN for 4‐weeks and then provided feedback, which was audio recorded and transcribed. Results: Three participants left the study after the initial consenting session and one participant discontinued after the second session, all specifying that they were already efficiently using similar electronic systems. The remaining six participants had an average attendance rate of 89%, which was above our initial target rate of 80%. Overall, participants reported that they enjoyed the group intervention and believe the DMN is a useful everyday memory aid. Participants endorsed seeing benefit of having multiple functions (e.g., calendar, to‐do list, notes, journal) in a single, senior‐friendly application. Participants particularly enjoyed the journal and note taking functions, emphasizing the organization and search functions to be the most beneficial. Three participants chose to download the DMN onto their personal tablet device for continued use and one participant downloaded a notes/journal specific application to their phone. Conclusions: Initial feedback for the DMN group intervention is promising and suggests that individuals find value in using the DMN as a compensatory tool. That said, participants may lack initial buy‐in if they are already using a similar electronic system. Information gleaned from this pilot study will help improve the group intervention and digital health tool. [ABSTRACT FROM AUTHOR]

Background: Cognitive changes experienced by individuals with mild cognitive impairment (MCI) can impact everyday functioning and lower self‐confidence. The Electronic Memory and Management Aid (EMMA) is a senior‐friendly iOS tablet application designed to support everyday activities. To examine whether learning and sustained app use could be improved, we partnered EMMA with a smart home that delivered activity‐aware prompting. Method: In a pilot trial, participants with MCI were randomly assigned to learn to use EMMA with (N = 14; M age = 74) or without (N = 15; M age = 71) smart home prompting. Training consisted of 5‐6 (two‐hour) structured sessions delivered in a 4‐week timeframe by a clinician in the participant's home (half turned into smart homes). Real‐time data were collected on participant use of EMMA, with metrics averaged across these intervals: weeks 1‐2 (early‐training), 5‐6 (post‐training), 10‐11 (prompting turned off) and 15‐16 (follow‐up; prompting back on). Baseline, post‐training and follow‐up clinical data (coping, life satisfaction, everyday functioning) were also compared. Result: No group differences emerged in age, education, estimated pre‐morbid ability (High Average), memory (Low Average) or executive functioning (Average). Participants in the partnered condition engaged daily with EMMA more times post‐training (M = 5.0) compared to the control (not‐partnered) condition (M = 2.0). Participants in the partnered condition also maintained a high level of engagement with EMMA even after prompting stopped (M = 5.0, M = 3.8 at follow‐up). This contrasted with significantly lowered use (M = 1.2 and M = 0.6, respectively) in the control condition. Similar patterns were found for usage rates of app components (event scheduling, calendar use) and advanced features (searches, alarms, pictures). Clinical data showed self‐reported positive changes in coping and life satisfaction that did not differ by condition; no significant changes were self‐reported in everyday functioning. Conclusion: Partnering EMMA with a smart home that cued app use at opportune times and after periods of non‐use improved habit formation and earlier mastery of the tool. Once a habit was established, turning off the prompts did not impact use of the EMMA app, demonstrating the importance of habit formation and early mastery to sustain tool use. [ABSTRACT FROM AUTHOR]

Background: Accurate error monitoring is important for successful completion of everyday tasks and use of compensatory strategies. This work examined error monitoring in individuals with amnestic MCI (aMCI) as compared to healthy older adults (HOA) using a computerized task. We also examined cognitive domains associated with error monitoring and the contribution of error monitoring to objective and self‐reported measurement of everyday functioning. Methods: Twenty‐four individuals with aMCI and 24 HOAs were administered a battery of neuropsychological tests (domains assessed: attention, working memory, executive functioning, memory, language, visuospatial abilities); a sustained attention task that included an error‐monitoring score (Sustained Attention to Response Task; SART); a naturalistic, performance‐based measure of everyday functioning (Day Out Task; DOT); and self‐ and informant‐report measures that captured everyday dysexecutive difficulties (DEX). Results: T‐test comparisons revealed the aMCI group (M = 0.48; SD = 0.22) exhibited poorer error monitoring (measured by the number of accurately self‐corrected errors on the SART divided by total errors) compared to the HOA group (M = 0.71; SD = 0.23), d = 1.02. Partial correlation analyses with age as a covariate revealed significant correlations between the error monitoring score and the cognitive domains of working memory and executive functioning for both the aMCI group and the HOA group. Hierarchical regression analyses revealed that after accounting for age and global cognitive status, the SART error monitoring score significantly predicted DOT total time, but not accuracy, as well as both self‐ and informant DEX scores. Conclusion: This study revealed that the ability to recognize errors was significantly poorer in individuals with aMCI compared to HOAs and most associated with the cognitive domains of executive functioning and working memory. After controlling for age and general cognitive status, error monitoring predicted performance in everyday functioning as assessed by a naturalistic, performance‐based test and questionnaires assessing everyday dysexecutive difficulties. This suggests that if individuals with aMCI do not recognize performance inaccuracies, they will not be able to correct errors, leading to mistakes in everyday task completion and reduction in implementation of appropriate compensatory strategies. These results implicate error monitoring as a potential target for early interventions with individuals with aMCI. [ABSTRACT FROM AUTHOR]

Within a theoretical framework for rehabilitation that distinguishes between automatic and controlled processes, this article reviews basic science research and addresses the following issues: What cognitive abilities remain relatively intact after severe traumatic brain injury, and how can we incorporate intact skills into rehabilitative techniques? The reviewed research indicates that individuals with severe traumatic brain injury (a) generally perform similar to controls on cognitive tasks (or task components) that require automatic and implicit processes as opposed to consciously controlled processes; (b) can learn through implicit learning mechanisms; and (c) can acquire and use automatic processes in complex, cognitive task performance. [ABSTRACT FROM AUTHOR]

Evaluates the effectiveness of a nine-week memory notebook treatment for closed-head-injured participants with documented memory deficits. Administration of memory outcome indicators; Observed everyday memory failures.

Background: Poorly managed pain can lead to substance use disorders, depression, suicide, worsening health, and increased use of health services. Most pain assessments occur in clinical settings away from patients' natural environments. Advances in smart home technology may allow observation of pain in the home setting. Smart homes recognizing human behaviors may be useful for quantifying functional pain interference, thereby creating new ways of assessing pain and supporting people living with pain.Objective: This study aimed to determine if a smart home can detect pain-related behaviors to perform automated assessment and support intervention for persons with chronic pain.Methods: A multiple methods, secondary data analysis was conducted using historic ambient sensor data and weekly nursing assessment data from 11 independent older adults reporting pain across 1-2 years of smart home monitoring. A qualitative approach was used to interpret sensor-based data of 27 unique pain events to support clinician-guided training of a machine learning model. A periodogram was used to calculate circadian rhythm strength, and a random forest containing 100 trees was employed to train a machine learning model to recognize pain-related behaviors. The model extracted 550 behavioral markers for each sensor-based data segment. These were treated as both a binary classification problem (event, control) and a regression problem.Results: We found 13 clinically relevant behaviors, revealing 6 pain-related behavioral qualitative themes. Quantitative results were classified using a clinician-guided random forest technique that yielded a classification accuracy of 0.70, sensitivity of 0.72, specificity of 0.69, area under the receiver operating characteristic curve of 0.756, and area under the precision-recall curve of 0.777 in comparison to using standard anomaly detection techniques without clinician guidance (0.16 accuracy achieved; P<.001). The regression formulation achieved moderate correlation, with r=0.42.Conclusions: Findings of this secondary data analysis reveal that a pain-assessing smart home may recognize pain-related behaviors. Utilizing clinicians' real-world knowledge when developing pain-assessing machine learning models improves the model's performance. A larger study focusing on pain-related behaviors is warranted to improve and test model performance. [ABSTRACT FROM AUTHOR]