We consider the problem of controlling an invasive mechanical ventilator for pressure-controlled ventilation: a controller must let air in and out of a sedated patient's lungs according to a trajectory of airway pressures specified by a clinician. Hand-tuned PID controllers and similar variants have comprised the industry standard for decades, yet can behave poorly by over- or under-shooting their target or oscillating rapidly. We consider a data-driven machine learning approach: First, we train a simulator based on data we collect from an artificial lung. Then, we train deep neural network controllers on these simulators.We show that our controllers are able to track target pressure waveforms significantly better than PID controllers. We further show that a learned controller generalizes across lungs with varying characteristics much more readily than PID controllers do.
@article{arxiv.2102.06779,
title = {Machine Learning for Mechanical Ventilation Control},
author = {Daniel Suo and Naman Agarwal and Wenhan Xia and Xinyi Chen and Udaya Ghai and Alexander Yu and Paula Gradu and Karan Singh and Cyril Zhang and Edgar Minasyan and Julienne LaChance and Tom Zajdel and Manuel Schottdorf and Daniel Cohen and Elad Hazan},
journal= {arXiv preprint arXiv:2102.06779},
year = {2022}
}