Life support systems are playing a critical role on keeping a patient alive when admitted in ICU bed. One of the most popular life support system is Mechanical Ventilation which helps a patient to breath when breathing is inadequate to maintain life. Despite its important role during ICU admission, the technology for Mechanical Ventilation hasn't change a lot for several years. In this paper, we developed a model using artificial neural networks, in an attempt to make ventilators more intelligent and personalized to each patient's needs. We used artificial data to train a deep learning model that predicts the correct pressure to be applied on patient's lungs every timepoint within a breath cycle. Our model was evaluated using cross-validation and achieved a Mean Absolute Error of 0.19 and a Mean Absolute Percentage Error of 2%.
Keywords: Deep learning; LSTM; Life support; Mechanical ventilation.
© 2023. The Author(s), under exclusive license to Springer Nature Switzerland AG.