Auto-Mode Shifting: A Machine Learning-Based Automatic Ventilation Mode Selection Technique Personalized to Physiological Characteristics of Intubated Patients

Respiratory assistance is the external Positive Pressure Ventilation (PPV) support that is given to a critically ill patient suffering from an improper respiratory drive with the aim to reduce the Work of Breathing (WOB). The modern 4th Generation ventilation systems have a number of settings to be optimized based on patient-specific physiological characteristics. Selection of the various ventilation modes and the parameters to be set in these modes which ultimately decide the oxygenation and ventilation process are predominantly determined by the Arterial Blood Gas (ABG) analysis. Artificial intelligence has been found akin to human experts in detecting Patient Ventilator Asynchronies (PVA) using ventilation-associated parametric analysis and it can play a pivotal role in the automation of ventilator modes. In the current paper, a K- Nearest Neighbours (K-NN) classifier- based auto-mode shifting technique is presented with the help of which machine will give the indication about the need for mode change and also assist in mode selection decision-making. This may help in reducing Ventilator Triggered Lung and Diaphragm Injuries (VTLDI) and also reduce the need for prolonged ventilation to a significant extent.