PaperarXiv:2509.11789
costream: A Streaming Evaluation Protocol for Wearable Fall Detection
arXiv preprint (cs.LG), v2 2026-11-12
TL;DR
Protocol for scoring wearable fall detectors on continuous, unsegmented streams instead of pre-cut windows. On real-world FARSEEING falls the two protocols picked different best methods in 6 of 20 configurations, and only streaming exposes false alarms per hour and detection delay. Code public, no licence file.
Abstract
Wearable fall detectors are commonly evaluated on presegmented windows, where a fall window is extracted at an annotated impact and non-fall windows are sampled from fall-free signal. However, a deployed detector processes a continuous sensor stream in which the location of a fall is unknown. We specify a streaming evaluation protocol in which a detector traverses an unsegmented recording, aggregates window-level scores into a temporal confidence map, and emits timestamped events that are evaluated against the annotated fall. Holding the dataset, participant-wise splits, window construction and fitted models fixed, we compare presegmented and streaming evaluation on clinically verified real-world falls. The two protocols identify the same best-performing method overall, but select different methods in 6 of 20 model-window configurations, with the divergence associated with differences in the temporal clustering of false alarms. Streaming evaluation also exposes deployment-relevant quantities, including false alarm rate over elapsed monitoring time, detection delay, and the effect of the decision threshold, that are not captured by per-window evaluation. These results show that evaluation protocol can materially affect conclusions about wearable fall detectors and should be considered when assessing their suitability for continuous monitoring.
Verbatim from arXiv (opens in a new tab) · arXiv metadata (abstracts are released under CC0)
Results on ChatPPG leaderboards
All leaderboardsNo results from this paper on ChatPPG leaderboards yet. Browse the leaderboards.
Tasks & signals
Signals
Keywords
How it works
Cite this paper
Aderinola, T. B., D'Ascanio, I., Palmerini, L., Chiari, L., Klenk, J., Becker, C., Caulfield, B., & Ifrim, G. (2025). costream: A Streaming Evaluation Protocol for Wearable Fall Detection. arXiv preprint , v2. https://arxiv.org/abs/2509.11789
Aderinola, Timilehin B., Ilaria D'Ascanio, Luca Palmerini, Lorenzo Chiari, Jochen Klenk, Clemens Becker, Brian Caulfield, and Georgiana Ifrim. 2025. "costream: A Streaming Evaluation Protocol for Wearable Fall Detection." arXiv preprint , v2. https://arxiv.org/abs/2509.11789.
Aderinola, T. B., D'Ascanio, I., Palmerini, L., Chiari, L., Klenk, J., Becker, C., Caulfield, B. and Ifrim, G. (2025) 'costream: A Streaming Evaluation Protocol for Wearable Fall Detection', arXiv preprint , v2. Available at: https://arxiv.org/abs/2509.11789
Aderinola, Timilehin B., et al. "costream: A Streaming Evaluation Protocol for Wearable Fall Detection." arXiv preprint , v2, 2025. https://arxiv.org/abs/2509.11789.
T. B. Aderinola et al., "costream: A Streaming Evaluation Protocol for Wearable Fall Detection," arXiv preprint , v2, 2025, [Online]. Available: https://arxiv.org/abs/2509.11789.
@misc{aderinola2025costream,
title = {{costream: A Streaming Evaluation Protocol for Wearable Fall Detection}},
author = {Aderinola, Timilehin B. and D'Ascanio, Ilaria and Palmerini, Luca and Chiari, Lorenzo and Klenk, Jochen and Becker, Clemens and Caulfield, Brian and Ifrim, Georgiana},
year = {2025},
howpublished = {arXiv preprint , v2},
eprint = {2509.11789},
archivePrefix = {arXiv}
}Added to the ChatPPG research hub