Abstract Or105: <i>“If it Can Help Someone, Then They Want to Do It”:</i> Demonstrating Feasibility of End-of-Life Studies to Support Development and Validation of Cardiac Arrest Detection Technologies with Consumer Smartwatches
Bibliographic record
Abstract
INTRO: Most out-of-hospital cardiac arrest (OHCA) cases are unwitnessed, leading to poor survival. Wearable devices have been proposed to notify first responders that an OHCA has occurred. However, there are currently no systems that have been validated using real cardiac arrest data. To address this gap, we conducted an observational clinical trial in end-of-life settings to collect data on natural and induced cardiac arrest using wearable technologies. Methods: We recruited individuals across Canada undergoing Medical Assistance In Dying (MAID) procedures, as well as patients in hospice settings. Participants received a consumer grade study watch that collected raw photoplethysmography (PPG) data corresponding to blood volume changes. We completed an interim analysis to describe the cohort, reasons for participation, and assess the feasibility of using this data for algorithm development. We plotted raw PPG values for each participant and used clinical event timelines (time of MAID medication administration, time of respiratory arrest, time of clinician confirmed death) to classify recordings as pulsatile, transitory, or pulseless. Results: From May 1, 2024, to May 1, 2025, 64 individuals were enrolled. 53 met the primary outcome of death while wearing the study watch (82.8%, Table 1). In the MAID arm, all but one case was identified by the same clinician who approached all their community patients during the study period with 34/36 (94.4%) consenting. In the hospice arm, 30/186 (16.1%) patients across three study sites consented. 6 (20%) hospice patients withdrew due to watch discomfort and 5 (16.7%) patients did not wear the watch at time of death. Participants reported an interest in giving back and helping science as motivations for participation (Table 2). Data was successfully obtained from all participants who wore the watch at the time of death. A loss of pulsatile activity was observed for participants that corresponded to clinical event timelines (Figure 1). Conclusion: Wearable-based research in end-of-life settings was feasible and yielded useful data. A clinician advocate was essential for high patient opt-in in the MAID arm. Watch discomfort was a barrier to study continuation in the hospice arm, as the duration of device wear was often several weeks. Ongoing work to train models for cardiac arrest detection is supported by the diversity of this large dataset of pulsatile and pulseless recordings from natural and induced cardiac arrest.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.008 | 0.031 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.003 | 0.001 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.003 | 0.002 |
| Insufficient payload (model declined to judge) | 0.243 | 0.068 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".