Correlates of Person-Specific Rates of Change in Sensor-Derived Physical Activity Metrics of Daily Living in the Rush Memory and Aging Project

Proof-of-concept for integrating multimodal digital health assessments into lifestyle interventions for older adults with dementia risk factors

Multimodal digital health assessments overcome the limitations of patient-reported outcomes by allowing for continuous and passive monitoring but remain underutilized in older adult lifestyle interventions for brain health. Therefore, we aim to (1) report ecological momentary assessment (EMA) and ActiGraph adherence among older adults during a lifestyle intervention; and (2) use dynamic data collected via EMA and ActiGraph to examine person-specific patterns of mindfulness, steps, and sleep throughout the intervention. We analyzed EMA and ActiGraph data from a pilot study of the 8-week My Healthy Brain program (N = 10) lifestyle group for older adults (60+) with subjective cognitive decline. EMA adherence metrics included proportion of EMA completed and the proportion of days with at least 10 mindfulness minutes. ActiGraph GT9X adherence metrics included the number of valid wear days (≥ 7 h) and the number of days participants achieved their step goal. We used linear mixed-effects models to examine person-specific patterns of step count, sleep efficiency, and mindfulness practice. On average, participants completed 39 of the 49 possible EMAs (80%) during the program. ActiGraph adherence was slightly higher than EMA (M = 61.40 days, 87.71%). Participants achieved the daily mindfulness goal (10 min/day) and step goal on 46.32% and 55.10% of days, respectively. Dynamic data revealed that on average, participant step counts increased by approximately 16.5 steps per day (b = 16.495, p = 0.002). However, some participants exhibited no changes while improvements made by other participants returned to baseline levels of activity. There was substantial heterogeneity in trajectories of mindfulness practice and sleep efficiency. EMA and ActiGraph are feasible for older adults enrolled in dementia risk reduction lifestyle interventions. Future studies are needed to better understand how mechanisms of lifestyle behaviors captured by EMA and ActiGraph are related to cognitive outcomes in older adults.

A Mindfulness-Based Lifestyle Intervention for Dementia Risk Reduction: Protocol for the My Healthy Brain Feasibility Randomized Controlled Trial

BACKGROUND

Lifestyle behavior change and mindfulness have direct and synergistic effects on cognitive functioning and may prevent Alzheimer disease and Alzheimer disease-related dementias (AD/ADRD). We are iteratively developing and testing My Healthy Brain (MHB), the first mindfulness-based lifestyle group program targeting AD/ADRD risk factors in older adults with subjective cognitive decline. Our pilot studies (National Institutes of Health [NIH] stage 1A) have shown that MHB is feasible, acceptable, and associated with improvement in lifestyle behavior and cognitive outcomes.

OBJECTIVE

We will compare the feasibility of MHB versus an education control (health enhancement program [HEP]) in 50 older adults (aged ≥60 y) with subjective cognitive decline and AD/ADRD risk factors. In an NIH stage 1B randomized controlled trial (RCT), we will evaluate feasibility benchmarks, improvements in cognitive and lifestyle outcomes, and engagement of hypothesized mechanisms.

METHODS

We are recruiting through clinics, flyers, web-based research platforms, and community partnerships. Participants are randomized to MHB or the HEP, both delivered in telehealth groups over 8 weeks. MHB participants learn behavior modification and mindfulness skills to achieve individualized lifestyle goals. HEP participants receive lifestyle education and group support. Assessments are repeated after the intervention and at a 6-month follow-up. Our primary outcomes are feasibility, acceptability, appropriateness, credibility, satisfaction, and fidelity benchmarks. The secondary outcomes are cognitive function and lifestyle (physical activity, sleep, nutrition, alcohol and tobacco use, and mental and social activity) behaviors. Data analyses will include the proportion of MHB and HEP participants who meet each benchmark (primary outcome) and paired samples 2-tailed t tests, Cohen d effect sizes, and the minimal clinically important difference for each measure (secondary outcomes).

RESULTS

Recruitment began in January 2024. We received 225 inquiries. Of these 225 individuals, 40 (17.8%) were eligible. Of the 40 eligible participants, 21 (52.5%) were enrolled in 2 group cohorts, 17 (42.5%) were on hold for future group cohorts, and 2 (5%) withdrew before enrollment. All participants have completed before the intervention assessments. All cohort 1 participants (9/21, 43%) have completed either MHB or the HEP (≥6 of 8 sessions) and after the intervention assessments. The intervention for cohort 2 (12/21, 57%) is ongoing. Adherence rates for the Garmin Vivosmart 5 (128/147, 87.1% weeks) and daily surveys (105/122, 86.1% weeks) are high. No enrolled participants have dropped out. Enrollment is projected to be completed by December 2024.

CONCLUSIONS

The RCT will inform the development of a larger efficacy RCT (NIH stage 2) of MHB versus the HEP in a more diverse sample of older adults, testing mechanisms of improvements through theoretically driven mediators and moderators. The integration of mindfulness with lifestyle behavior change in MHB has the potential to be an effective and sustainable approach for increasing the uptake of AD/ADRD risk reduction strategies among older adults.

TRIAL REGISTRATION

ClinicalTrials.gov NCT05934136; https://www.clinicaltrials.gov/study/NCT05934136.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID)

DERR1-10.2196/64149.

Self-supervised learning of wrist-worn daily living accelerometer data improves the automated detection of gait in older adults

Progressive gait impairment is common among aging adults. Remote phenotyping of gait during daily living has the potential to quantify gait alterations and evaluate the effects of interventions that may prevent disability in the aging population. Here, we developed ElderNet, a self-supervised learning model for gait detection from wrist-worn accelerometer data. Validation involved two diverse cohorts, including over 1000 participants without gait labels, as well as 83 participants with labeled data: older adults with Parkinson's disease, proximal femoral fracture, chronic obstructive pulmonary disease, congestive heart failure, and healthy adults. ElderNet presented high accuracy (96.43 ± 2.27), specificity (98.87 ± 2.15), recall (82.32 ± 11.37), precision (86.69 ± 17.61), and F1 score (82.92 ± 13.39). The suggested method yielded superior performance compared to two state-of-the-art gait detection algorithms, with improved accuracy and F1 score (p < 0.05). In an initial evaluation of construct validity, ElderNet identified differences in estimated daily walking durations across cohorts with different clinical characteristics, such as mobility disability (p < 0.001) and parkinsonism (p < 0.001). The proposed self-supervised method has the potential to serve as a valuable tool for remote phenotyping of gait function during daily living in aging adults, even among those with gait impairments.

Motor and Nonmotor Measures and Declining Daily Physical Activity in Older Adults

Importance

Difficulties in identifying modifiable risk factors associated with daily physical activity may impede public health efforts to mitigate the adverse health outcomes of a sedentary lifestyle in an aging population.

Objective

To test the hypothesis that adding baseline sensor-derived mobility metrics to diverse baseline motor and nonmotor variables accounts for the unexplained variance of declining daily physical activity among older adults.

Design, Setting, and Participants

This cohort study analyzed data from participants of the Rush Memory and Aging Project (MAP), an ongoing longitudinal clinical pathological study that began to enroll older adults (age range, 59.4-104.9 years) in 1997. Wrist- and waist-worn sensors were added to MAP in 2005 and 2012, respectively, to record participants' physical activity and mobility performances. Included participants were examined at baseline and annually followed up for a mean (SD) duration of 4.2 (1.6) years.

Exposure

Twelve blocks of variables, including 3 blocks of mobility metrics derived from recordings of a belt-worn sensor to quantify a 32-foot walk, a Timed Up and Go (TUG) test, and a standing balance task, and 9 other blocks with 41 additional variables.

Main Outcomes and Measures

A linear mixed-effects model was used to estimate the person-specific rate of change (slope) of total daily physical activity obtained from a wrist-worn sensor. Twelve linear regression models were used to estimate the adjusted R2 to quantify the associations of the variables with the slope.

Results

A total of 650 older adults (500 females [76.9%]; mean [SD] age at baseline, 81.4 [7.5] years; 31 Black individuals [4.8%], 17 Latino individuals [2.6%], and 602 White individuals [92.6%]) were included. During follow-up, all but 1 participant showed declining daily physical activity, which was equivalent to approximately 16.8% decrease in activity level per year. In separate models, waist sensor-derived mobility metrics (32-foot walk: adjusted R2, 23.4% [95% CI, 17.3%-30.6%]; TUG test: adjusted R2, 22.8% [95% CI, 17.7%-30.1%]) and conventional motor variables (adjusted R2, 24.1% [95% CI, 17.7%-31.4%]) had the largest percentages of variance of declining daily physical activity compared with nonmotor variables. When the significant variables from all 12 blocks were included together in a single model, only turning speed (estimate [SE], 0.018 [0.006]; P = .005) and hand dexterity (estimate [SE], 0.091 [0.034]; P = .008) showed associations with declining daily physical activity.

Conclusions and Relevance

Findings of this study suggest that sensor-derived mobility metrics and conventional motor variables compared with nonmotor measures explained most of the variance of declining daily physical activity. Further studies are needed to ascertain whether improving specific motor abilities, such as turning speed and hand dexterity, is effective in slowing the decline of daily physical activity in older adults.

Automated Gait Detection in Older Adults during Daily-Living using Self-Supervised Learning of Wrist-Worn Accelerometer Data: Development and Validation of ElderNet

Progressive gait impairment is common in aging adults. Remote phenotyping of gait during daily living has the potential to quantify gait alterations and evaluate the effects of interventions that may prevent disability in the aging population. Here, we developed ElderNet, a self-supervised learning model for gait detection from wrist-worn accelerometer data. Validation involved two diverse cohorts, including over 1,000 participants without gait labels, as well as 83 participants with labeled data: older adults with Parkinson's disease, proximal femoral fracture, chronic obstructive pulmonary disease, congestive heart failure, and healthy adults. ElderNet presented high accuracy (96.43 ± 2.27), specificity (98.87 ± 2.15), recall (82.32 ± 11.37), precision (86.69 ± 17.61), and F1 score (82.92 ± 13.39). The suggested method yielded superior performance compared to two state-of-the-art gait detection algorithms, with improved accuracy and F1 score (p < 0.05). In an initial evaluation of construct validity, ElderNet identified differences in estimated daily walking durations across cohorts with different clinical characteristics, such as mobility disability (p < 0.001) and parkinsonism (p < 0.001). The proposed self-supervised gait detection method has the potential to serve as a valuable tool for remote phenotyping of gait function during daily living in aging adults.

Identifying Subtle Motor Deficits Before Parkinson's Disease is Diagnosed: What to Look for?

Motor deficits typical of Parkinson's disease (PD), such as gait and balance disturbances, tremor, reduced arm swing and finger movement, and voice and breathing changes, are believed to manifest several years prior to clinical diagnosis. Here we describe the evidence for the presence and progression of motor deficits in this pre-diagnostic phase in order to provide suggestions for the design of future observational studies for an effective, quantitatively oriented investigation. On the one hand, these future studies must detect these motor deficits in as large (potentially, population-based) cohorts as possible with high sensitivity and specificity. On the other hand, they must describe the progression of these motor deficits in the pre-diagnostic phase as accurately as possible, to support the testing of the effect of pharmacological and non-pharmacological interventions. Digital technologies and artificial intelligence can substantially accelerate this process.