The majority of intensive care unit patients are administered intravenous sedatives with the goal of maintaining physiological stability and comfort during mechanical ventilation. Efforts to reduce the duration of ventilation via weaning and sedation protocols can greatly improve clinical outcomes and reduce costs but opinion on the best protocols vary, due to a lack of robust evidence.

At present, I’m working towards the application of reinforcement learning to inform the control of sedation levels given vital readings, and predict as soon as possible when a patient will be ready to undergo and pass a spontaneous breathing trial (SBT) in preparation for extubation.