Clinically relevant monitoring of long-term night-time behaviour and physiology from the homes of people living with dementia.

Abstract

Disturbances of sleep and night-time behaviours are amongst the most disabling symptoms of dementia. They often increase carers' burden and the risk of institutionalization. The causes are complex and are difficult to investigate because of a lack of acceptable methods for monitoring behaviours in the home. Here we show that a passive under-mattress can be used to track changes in night-time behaviour and physiology, and that a range of digital biomarkers produced are informative in understanding the effects of medication changes, disease progression and intercurrent illness in patients living with dementia (PLWD). We used contactless Withings Sleep Mattress (WSM) to monitor bed-occupancy in 4 PLWD (age 74-93, 3males) enrolled into the CR&T MINDER cohort study. Each participant was tracked over 1000 nights between 2019 and 2021. Minute-to-minute timeseries were extracted from WSM to calculate bed occupancy metrics and nocturnal physiology measures (heart and breathing rates (HR/BR)). Raw measures were standardized within subjects by comparing each time point to the mean of the time points that preceded it. We then investigated the relationship between these metrics and clinical events such as infections and medication. The 4 case studies illustrate the potential of this technology for passive health monitoring in PLWD. High levels of intraindividual variability in behavioural and physiological metrics were observed. Progressive changes in bed occupancy were observed in two patients with frontotemporal dementia and Alzheimer's disease (Cases 1&2). Intercurrent illness and medications changes influenced the measures. For example, Patient 1 showed progressive night-time wandering with increasing time spent out of bed, which improved following the initiation of risperidone. Case 2 showed recurrent episodes of heart failure accompanied by increased nocturnal HR. Cases 3 and 4 showed urinary tract infections, which were accompanied by transient changes in bed occupancy measures. Passive monitoring of night-time behaviour and physiology is a simple and cost-effective way to monitor the effects of medication change and medical events in PLWD. Digital biomarkers derived from these simple measures hold great promise as measures of dementia progression, to enhance dementia care by identifying at risk individuals and as novel clinical outcome measures in clinical trials.