Methodology for Interpretable Reinforcement Learning for Optimizing Mechanical Ventilation | Synapse