Wearable sensors will track the dying process in dementia with continuous physiologic data
People with dementia often die after a prolonged decline in institutional care, where clinical observation can be fragmented and the final trajectory is difficult to reconstruct. This BMJ Open protocol sets out to examine death and dying through digital thanotyping, a systems-based approach that uses continuous sensor data to map the physiologic changes that accompany the terminal phase of illness.
The study is designed around a simple but clinically important problem: bedside impressions alone rarely capture the timing, sequence, or interaction of the symptoms and body-system changes that precede death in frail older adults. In people with dementia, that challenge is amplified by cognitive impairment, limited verbal communication, and the frequent need for surrogate decision-making. A monitoring strategy built on wearable and contact-free devices offers a way to observe those changes with more resolution than routine chart review or intermittent vital-sign checks.
Continuous monitoring in a frail population
The protocol uses wrist-worn and wall-mounted sensors to collect longitudinal data while minimizing burden for participants. Non-invasive monitoring matters in this population because aggressive testing is often unrealistic near end of life, and repeated bedside assessments can be distressing or impractical. By relying on devices that can operate continuously without direct contact, the study aims to capture objective measurements across the dying process rather than isolated snapshots.
Those measurements are intended to be high-resolution enough to examine temporal relationships among multiple physiologic and clinical indicators. The framework is not limited to one endpoint or one organ system. Instead, it is built to assess how different signals evolve together as death approaches, which may help clarify patterns that are otherwise obscured when clinicians document only the final days in broad strokes.
The approach also reflects the realities of institutional end-of-life care. Older adults with dementia are frequently cared for in settings where staffing constraints, fluctuating clinical status, and competing priorities can make detailed manual observation difficult. Continuous sensor data can fill some of those gaps, although the protocol acknowledges that data completeness will still be challenged by technical failures, logistical barriers, and the unavoidable complexity of end-of-life care.
Why proxy consent and data loss matter here
Recruitment is expected to depend heavily on proxy consent because many participants will not have decision-making capacity. In dementia research, that requirement is more than an administrative detail; it shapes who can be enrolled, when consent can be obtained, and how representative the final cohort will be. It also means the study must be designed with special attention to ethics, communication, and the burden placed on families or legal surrogates.
Missing or incomplete data are another central limitation built into the protocol itself. Continuous monitoring near death can be interrupted by device displacement, patient discomfort, transitions in care, or clinical events that prevent uninterrupted recording. Those gaps may be unavoidable, but they also influence how well the final analysis can characterize the terminal phase. For a study built on temporal relationships, interruptions in the time series are not a minor nuisance; they directly affect the fidelity of the biologic narrative the investigators hope to reconstruct.
One sentence captures the central challenge. End-of-life data in dementia will be hard to collect, and even harder to keep complete.
Digital thanotyping and the bedside meaning of physiologic trajectories
Digital thanotyping is an effort to define dying through measurable signals rather than retrospective inference. In practice, that means using sensors to detect patterns that may include changes in activity, motion, rest, and other physiologic indicators as the body moves toward death. The study’s systems-based framework is designed to interpret those signals as interconnected events rather than isolated abnormalities.
That matters because dying in dementia rarely follows a clean, linear path. Patients may experience gradual decline over months, with superimposed fluctuations caused by intercurrent illness, reduced mobility, nutritional compromise, or worsening neurologic function. A continuous monitoring model may help distinguish short-lived variation from the broader terminal trajectory, which could eventually support better prognostication and more informed bedside decisions.
The protocol also addresses a gap in how end-of-life care is often studied. Traditional research methods depend heavily on clinician assessment, caregiver recollection, or sparse chart documentation, all of which can miss subtle physiologic shifts. Continuous sensor-based observation does not solve every problem, but it offers a more objective record of what happens as death approaches in this vulnerable population.
The biological rationale is straightforward. As dementia advances and the body enters its terminal phase, multiple systems change together, and those linked changes are likely to appear in activity patterns, rest-activity rhythms, and other measurable signals before they are obvious in routine clinical documentation. A high-frequency monitoring platform gives investigators a way to see those trajectories unfold in real time.
What this protocol can and cannot establish
Because this is a study protocol, no clinical outcome data are presented yet. The value of the paper lies in its design: a non-invasive, longitudinal framework for studying the dying process in people with dementia under real-world conditions. Its strengths are the use of continuous monitoring, the attempt to reduce observer dependence, and the potential to collect data without adding much burden to very frail participants.
The same features create its main vulnerabilities. Technical and logistical problems may limit completeness, and proxy consent may restrict recruitment. Those constraints will shape the dataset as much as the sensors themselves. Any future analysis will need to account for the fact that end-of-life research is conducted in a setting where missingness is often inseparable from the clinical reality being studied.
Even with those limitations, the protocol marks a clear step toward a more precise description of death in dementia. It moves the field away from isolated bedside impressions and toward a continuous record of physiologic change, which is exactly where a meaningful understanding of dying has been missing.