Automated mechanical ventilator design and analysis using neural network

Mechanical ventilation is the process through which breathing support is provided to patients who face inconvenience during respiration. During the pandemic, many people were suffering from lung disorders, which elevated the demand for mechanical ventilators. The handling of mechanical ventilators is to be done under the assistance of trained professionals and demands the selection of ideal parameters. In this work, a computer-aided simulation of ventilator design is performed for clinical complications like pneumonia and Chronic Obstructive Pulmonary Disease (COPD) and is validated against normal ventilatory parameters. The parameters such as tidal volume, respiratory rate, and inspiration to expiration ratio (I: E) are considered as control values to check the stability of the mechanical ventilator for stern performance. The check valves 1 and 2 governed by the control parameters provide optimal volume that must be sent inside the tracheal region. The hyperparameters are tuned using a low intricate feed-forward neural network (FFNN). The trained features serve as input to the sensors present in the mimicked lung model. The performance metrics of FFNN during the training and testing phases substantiate the optimal performance of the ventilator. The simulation and validation results indicate that the designed ventilator system is stable and effective for clinical use, providing optimal respiratory support for patients with pneumonia and COPD.