Development and validation of a reinforcement learning model for ventilation control during emergence from general anesthesia

Abstract

Ventilation should be assisted without asynchrony or cardiorespiratory instability during anesthesia emergence until sufficient spontaneous ventilation is recovered. In this multicenter cohort study, we develop and validate a reinforcement learning-based Artificial Intelligence model for Ventilation control during Emergence (AIVE) from general anesthesia. Ventilatory and hemodynamic parameters from 14,306 surgical cases at an academic hospital between 2016 and 2019 are used for training and internal testing of the model. The model’s performance is also evaluated on the external validation cohort, which includes 406 cases from another academic hospital in 2022. The estimated reward of the model’s policy is higher than that of the clinicians’ policy in the internal (0.185, the 95% lower bound for best AIVE policy vs. −0.406, the 95% upper bound for clinicians’ policy) and external validation (0.506, the 95% lower bound for best AIVE policy vs. 0.154, the 95% upper bound for clinicians’ policy). Cardiorespiratory instability is minimized as the clinicians’ ventilation matches the model’s ventilation. Regarding feature importance, airway pressure is the most critical factor for ventilation control. In conclusion, the AIVE model achieves higher estimated rewards with fewer complications than clinicians’ ventilation control policy during anesthesia emergence.

Opportunities of AI-powered applications in anesthesiology to enhance patient safety

“…For example, a system using reinforcement learning was developed to control ventilation during the emergence from general anesthesia while preventing hemodynamic and ventilatory complications. 15 In this case, optimal cardiorespiratory status served as a reward, while values outside of prespecified ranges served as a checkpoint for the model to further self-tune. Eventually, the model learned the optimal strategy for controlling ventilation.…”

“…This method is utilized when the model needs to learn a series of actions and receives feedback on each action based on how well the action was measured. For example, a system using reinforcement learning was developed to control ventilation during the emergence from general anesthesia while preventing hemodynamic and ventilatory complications 15 . In this case, optimal cardiorespiratory status served as a reward, while values outside of prespecified ranges served as a checkpoint for the model to further self-tune.…”