Fig. 5From: Towards more efficient and robust evaluation of sepsis treatment with deep reinforcement learningThe performance of different OPE estimatorsBack to article page