Saved in:
| Main Authors: | , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2606.01028 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911736366366720 |
|---|---|
| author | Wang, Yuepeng Kawano, Ken Zhou, Yongqi Fujisawa, Yoshihiko Weiss, Richard Wachi, Akifumi Fujisawa, Katsuki Chen, Ying Sadria, Mehrshad Liu, Xin Kim, Kyoung-Sook Hu, Xiao Gros, Sebastien Shen, Xun |
| author_facet | Wang, Yuepeng Kawano, Ken Zhou, Yongqi Fujisawa, Yoshihiko Weiss, Richard Wachi, Akifumi Fujisawa, Katsuki Chen, Ying Sadria, Mehrshad Liu, Xin Kim, Kyoung-Sook Hu, Xiao Gros, Sebastien Shen, Xun |
| contents | Medical treatment recommendation poses several challenges to reinforcement learning (RL): patient physiology evolves in continuous time, measurements and interventions are performed at irregular intervals, and treatment effects vary substantially across individuals. Existing RL formulations and simulated environments, however, are based on discrete-time MDP or POMDP abstractions with fixed or pre-specified decision intervals. Thus, it remains difficult to evaluate whether RL methods can handle time-interval-dependent disease progression, personalized treatment response, and safety between consecutive measurement points. To address this gap, we introduce MedGym, a benchmark environment for dynamic treatment recommendation. MedGym models longitudinal patient evolution in a continuous-time framework and constructs a configurable medical RL benchmark from clinical data by using Physics-Informed Neural Networks. The resulting benchmark supports both offline and online RL, and enables direct comparison between discrete-time and continuous-time methods under irregular treatment timing and patient-specific dynamics. Besides, MedGym supports evaluation from clinically important perspectives, including personalization, trajectory-level safety, and the performance gap between model-based offline learning and online deployment. By providing a standardized and configurable benchmark for continuous-time dynamic treatment, MedGym aims to facilitate more realistic and informative evaluation of medical RL methods. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2606_01028 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | MedGym:A Unified Continuous-Time Benchmark for Dynamic Medical Treatment Reinforcement Learning Wang, Yuepeng Kawano, Ken Zhou, Yongqi Fujisawa, Yoshihiko Weiss, Richard Wachi, Akifumi Fujisawa, Katsuki Chen, Ying Sadria, Mehrshad Liu, Xin Kim, Kyoung-Sook Hu, Xiao Gros, Sebastien Shen, Xun Machine Learning Medical treatment recommendation poses several challenges to reinforcement learning (RL): patient physiology evolves in continuous time, measurements and interventions are performed at irregular intervals, and treatment effects vary substantially across individuals. Existing RL formulations and simulated environments, however, are based on discrete-time MDP or POMDP abstractions with fixed or pre-specified decision intervals. Thus, it remains difficult to evaluate whether RL methods can handle time-interval-dependent disease progression, personalized treatment response, and safety between consecutive measurement points. To address this gap, we introduce MedGym, a benchmark environment for dynamic treatment recommendation. MedGym models longitudinal patient evolution in a continuous-time framework and constructs a configurable medical RL benchmark from clinical data by using Physics-Informed Neural Networks. The resulting benchmark supports both offline and online RL, and enables direct comparison between discrete-time and continuous-time methods under irregular treatment timing and patient-specific dynamics. Besides, MedGym supports evaluation from clinically important perspectives, including personalization, trajectory-level safety, and the performance gap between model-based offline learning and online deployment. By providing a standardized and configurable benchmark for continuous-time dynamic treatment, MedGym aims to facilitate more realistic and informative evaluation of medical RL methods. |
| title | MedGym:A Unified Continuous-Time Benchmark for Dynamic Medical Treatment Reinforcement Learning |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2606.01028 |