CausalBioRL: Reinforcement Learning with Causal World Models for Autonomous Drug Discovery

Model-based RL framework integrating causal discovery, structural causal models, and do-calculus planning for drug discovery. Features a DrugDiscovery-v0 Gymnasium environment (244D obs, 130D action), hierarchical planner (UCB1 + CEM), adaptive reward learner, and surrogate docking. Benchmarked against PPO, SAC, and random baselines across 3 environments.