Computer scientists urge a fundamental shift in how problems are formulated in reinforcement learning for healthcare ...