The hidden risks of temporal resampling in clinical reinforcement learning

ArXi:2602.06603v3 Announce Type: replace 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 inherently irregular clinical records, researchers often resample the data into uniform time intervals before