Technology and Dementia Preconference
Alzheimers Dement. 2025 Dec;21 Suppl 9:e110453. doi: 10.1002/alz70863_110453.
ABSTRACT
BACKGROUND: Passive actigraphy monitoring via commercial wearable devices offers a scalable opportunity to capture real-world changes in sleep-wake cycles affected in Alzheimer's disease and related dementias (ADRD). We characterized continuous rest-activity patterns derived from Fitbit™ data, and examined differences between healthy adults and ADRD cohorts.
METHOD: Tri-axial actigraphy data (step counts per minute), clinicodemographic information, cognitive and functional performance, and mood scores were completed at baseline in functionally intact adults (n = 118), individuals with single/multi-domain mild cognitive impairment (all-cause MCI; n = 37), Alzheimer's disease dementia (AD; n = 18); or frontotemporal lobar degeneration (FTLD, n = 30; see Table 1). Activity patterns were quantified by calculating rest-activity aggregates (e.g. averaged step count variability) and features derived from minute-level step count data (see Table 2). Principal component analysis (PCA) was performed for data reduction of nine features using singular value decomposition (SVD). Five components (accounting for >85% of the total variance in rest-activity patterns) were characterized according to patterns of strong variable contributions (+/- 0.4; see Table 2). Principal components (PCs) 1-5 were subsequently examined for associations with measures of cognitive and functional decline. The first of these components was further analyzed for group differences with the objective of characterizing syndrome-specific 24-hour activity rhythms.
RESULT: Controlling for participant age and biological sex, PC1 [activity variability and amplitude] was negatively associated with both CDR®+FTLD-NACC sum of boxes score [β=-0.30, s.e.=0.15 p = 0.010] and CDR®+FTLD-NACC global score [β=-0.06, s.e.=0.03, p = 0.010]). Dynamics relating to rest-activity start timing [PC3] was also negatively associated with CDR®+FTLD-NACC global score [β=-0.06, s.e.=0.04, p = 0.040]). PC1 also revealed diagnostic group differences (ηp2=0.06, p = 0.004). Post hoc Tukey HSD analysis indicated that this omnibus group difference was driven by a reduction in FTLD-associated group PC1 scores relative to both functionally intact (mean difference=-1.05, p = 0.004) and all-cause MCI cohorts (mean difference= -1.09, p = 0.019; see Figure 1).
CONCLUSION: Activity variability and amplitude derived from Fitbit™ data revealed distinct activity profile behaviors in individuals diagnosed with FTLD-associated syndromes, which differed from individuals living with mild cognitive impairment or individuals who are functionally intact.
PMID:41433334 | DOI:10.1002/alz70863_110453
