EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

The impact of temporal resampling on the safety of clinical reinforcement learning: A simulated trial

Thomas Frost iD, Hrisheekesh Vaidya iD, Steve Harris iD

PLOS Digital Health · 06.11.2026 · 5 (11) · e0001801

DOI: 10.1371/journal.pdig.0001801 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS Digital Health, online erschienen am 06.11.2026. Verfasst von Thomas Frost und 2 weiteren Autor:innen. Beteiligte Einrichtung: University College London, London, United Kingdom. Die Arbeit nennt 94 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Engineering and Physical Sciences Research Council, University College London Hospitals Biomedical Research Centre, National Institute for Health and Care Research.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Reinforcement learning (RL) is a type of artificial intelligence for making optimal choices. In healthcare, researchers generally use offline RL (ORL), where models are trained and evaluated from retrospective observational data. To accommodate naturally irregular clinical records, researchers often resample the data into uniform time intervals before training (known as binning). However, binning may create counterfactual trajectories that do not represent clinical practice. We first quantified these distortions in patients receiving continuous insulin infusions in the MIMIC-IV intensive care database. Binning reversed the apparent causal order for over 70% of paired glucose measurements and insulin adjustments within the same interval, and frequently masked significant dysglycaemic events and merged distinct treatment decisions together. We then conducted an in silico clinical trial using 30 virtual patients with type 1 diabetes from the FDA-approved UVA/Padova simulator. The simulator was modified to include stochastic intervals between decisions and used to generate a training dataset for offline RL. We trained three ORL algorithms on both the unprocessed dataset and equivalent datasets resampled at 10-minute, 2-hour, and 4-hour intervals. When deployed back into the simulated environment, temporal resampling was found to reduce model performance by up to 60% relative to unprocessed data, with 4-hour binning causing all agents to perform worse than the dataset’s baseline. Retrospective evaluation on resampled data actively obscured this effect, predicting 1.5–3x better returns than agents achieved in practice. We recommend that future research in this area prioritises datasets with natural clinical timings between decisions, which may be a necessary step before these models can be safely deployed into patient care.

Autor:innen und Institutionen

  • University College London, London, United Kingdom.
  • University College London Hospitals NHS Foundation Trust, London, United Kingdom.

ORCID-Profile: Thomas Frost, Hrisheekesh Vaidya, Steve Harris

Förderung

  • Engineering and Physical Sciences Research Council
  • University College London Hospitals Biomedical Research Centre
  • National Institute for Health and Care Research

Zitieren

APA

Frost, T., Vaidya, H., & Harris, S. (2026). The impact of temporal resampling on the safety of clinical reinforcement learning: A simulated trial. PLOS Digital Health, 5(11), e0001801. https://doi.org/10.1371/journal.pdig.0001801

Vancouver

Frost T, Vaidya H, Harris S. The impact of temporal resampling on the safety of clinical reinforcement learning: A simulated trial. PLOS Digital Health. 2026;5(11):e0001801. doi:10.1371/journal.pdig.0001801

BibTeX
@article{Frost2026,
  author = {Frost, Thomas and Vaidya, Hrisheekesh and Harris, Steve},
  title = {The impact of temporal resampling on the safety of clinical reinforcement learning: A simulated trial},
  journal = {PLOS Digital Health},
  year = {2026},
  volume = {5},
  number = {11},
  pages = {e0001801},
  doi = {10.1371/journal.pdig.0001801}
}
Kostenlos starten →