Investigator:
Honghuang Lin, UMass Chan Medical School.

MassAITC Cohort: Year 2 (AD/ADRD)

Project Accomplishments: This pilot project explored the use of wearable accelerometers to passively monitor walking cadence as a potential early indicator of cognitive decline in older adults. Using data from 1,212 participants in the Framingham Heart Study, the team analyzed 31 physical activity metrics and found that over half were significantly associated with neuropsychological test performance. Notably, higher peak walking cadence and time spent in moderate to vigorous physical activity were linked to better cognitive outcomes and a reduced risk of developing cognitive impairment over a 9-year follow-up. Machine learning models incorporating these physical activity measures alongside traditional risk factors achieved strong predictive performance, with an AUC of 0.80 for 2-year and 0.68 for 8-year risk predictions. 

These findings underscore the value of objective, wearable-based monitoring for assessing cognitive health and identifying individuals at risk for Alzheimer’s and related dementias. The study supports integrating physical activity tracking into routine health assessments for older adults and highlights the potential of digital phenotyping in preventive care. Preliminary results were presented at a national symposium, and a manuscript is in preparation for publication. The team is also pursuing further funding to expand this promising research into larger-scale applications.

Initial Proposal Abstract: Despite enormous efforts, therapeutic clinical trials for Alzheimer’s disease (AD) have largely failed, possibly because interventions are initiated too late into the disease course when neurodegeneration has begun and is irreversible. Early detection of people at high risk of AD is thus particularly important to help in the effort to develop prevention and treatment options. We hypothesize that subtle changes in motor function may serve as a potential biomarker of early signs of cognitive impairment. Tracking changes in cadence will enable the assessment of potential parallel trajectories of motor/cognitive decline and may prove to be an indicator of declining cognition.

This study will leverage a large collection of physical activity and cognitive assessment data, and derive novel digital phenotypes from commercially available wearable devices. We will then assess the correlation of these novel measures with declining cognitive function, and develop advanced machine learning models to identify people at high risk to develop cognitive impairment. Given the increasing popularity of wearable devices, the knowledge and tools developed from the current project could be readily applicable to the general population for the large-scale screening of cognitive health.