Physical, cognitive, and social activity engagement are modifiable factors that may reduce dementia risk. Yet their assessment typically relies on self-report measures prone to recall bias. The COMPADRE-CART study uses passive home monitoring technology to objectively capture real-world activity patterns in Hispanic/Latinx older adults, a population at elevated risk for Alzheimer's disease and related dementias (ADRD), yet underrepresented in research. Objectives: Determine whether objectively measured room transitions (inhome mobility) and daily step counts (physical activity) are associated with cognitive performance across domains of global cognition, episodic memory, and semantic fluency. Population: Hispanic/Latinx older adults aged 62 and older at elevated risk for Alzheimer's disease in South Texas. Methods: Passive Monitorig Wireless motion sensors (NYCE) objectively track room transitions and time out of home. Wearable technology captures real-world physical activity through daily step counts. Cognitive Assessment • Global Cognition: Montreal Cognitive Assessment (MoCA) • Episodic Memory: Craft Story Immediate and Delayed Recall • Semantic Fluency: Animal Naming Key Findings: In 26 participants, frequent room transitions correlate with better global cognition ((β=0.75, p=0.010) and episodic memory performance (immediate recall: β=0.43, p=0.004; delayed recall: β=0.37, p=0.016. Higher daily steps was associated with better performance in semantic fluency (β=0.21, p=0.028; n=92). Time outside was associated with better semantic fluency (β=0.39, p=0.033) but worse episodic memory (immediate β=−0.41, p=0.025; delayed recall: β=−0.48, p=0.006; n=26).
Limitations include the cross-sectional design, current small sample for in-home mobility analyses (n=26), and potential self-selection bias inherent to single-person household restriction. Different activity dimensions map onto different cognitive domains, mobility within the home onto memory, steps onto fluency, pointing toward passive monitoring as a tool for domain-specific cognitive surveillance. Longitudinal analyses will assess whether these patterns predict conversion to MCI or dementia in this cohort.
Acknowledgments We thank all research volunteers, research team, and our funders who made this effort possible. P30AG066546, R01AG062531, AG059421, HL105756, N01-HC25195,HHSN268201500001I,75N92019D00031.