Nocturnal sensor signals may predict next-day agitation in dementia cohorts

Photo by KJ Gonzales / Unsplash

medRxiv Published April 18, 2026 Medically reviewed April 25, 2026 Study authors: Liu, Z.; Bono, M.; Flisar, A.; Decloedt, R.; Van Den Bossche, M.; De Vos, M. DOI ↗ By Dr. Sofia Müller, MD · Lifespan & Whole-Person Care

Key Takeaway

Consider nocturnal sensor data as a potential tool for short-term agitation risk stratification in dementia, noting limited evidence.

This was a cohort study in long-term care cohorts (N=55) and an external home-monitoring cohort (N=18) with dementia. Researchers used unobtrusive under-mattress sensors to collect nocturnal physiological signals and assessed their association with next-day daytime agitation as the primary outcome. Secondary outcomes included agitation occurrence and severity.

The main results showed that lower nocturnal respiratory rate and greater activity instability independently predicted higher odds of agitation occurrence. Associations were stronger for motor than verbal agitation. No nocturnal features significantly predicted agitation severity. Effect sizes, absolute numbers, and p-values or confidence intervals were not reported.

Safety and tolerability were not reported, with no data on adverse events, serious adverse events, or discontinuations. Key limitations include the small sample sizes and observational design, which preclude causal inference. The study supports the potential of under-mattress sensing for short-term risk stratification and more proactive dementia care, but findings are preliminary and require validation.

Study Details

Study typeCohort

Sample sizen = 55

EvidenceLevel 3

PublishedApr 2026

View Original Abstract ↓

INTRODUCTIONAgitation is a common and burdensome neuropsychiatric symptom in dementia that fluctuates from day to day, but objective tools for short-term risk stratification are limited. We examined whether nocturnal physiological signals from unobtrusive under-mattress sensors predict next-day daytime agitation and whether associations differ for agitation occurrence versus severity. METHODSWe extracted cardiorespiratory, movement, and sleep-proxy features from two long-term care cohorts (N=55; 333 nights) and one external home-monitoring cohort (N=18; 803 nights). A two-part mixed-effects framework was used to model next-day agitation episodes. RESULTSLower nocturnal respiratory rate and greater activity instability independently predicted higher odds of next-day agitation occurrence. Associations were stronger for motor than verbal agitation. Respiration-related predictors were validated externally. Conversely, no nocturnal features significantly predicted agitation severity. DISCUSSIONPassive sleep monitoring identified reproducible, physiologically interpretable markers of next-day agitation occurrence, supporting the potential of under-mattress sensing for short-term risk stratification and more proactive dementia care.