B-spline modelling of inspiratory drive in NAVA-ventilated patients

Abstract Model-based approaches are often used to estimate mechanical properties of lungs, such as elastance (E) and airway resistance (R), during invasive and non-invasive mechanical ventilation (MV). Current models are less effective when spontaneous breathing is present. This analysis utilises b-spline functions within a single compartment model framework to identify patient-specific inspiratory driving pressure. A series of 2nd-order, constrained b-spline basis functions are used to identify inspiratory driving pressure breath to breath alongside single E and R using inspiration and expiration data from n=20 breaths for 10 patients ventilated using NAVA. Median [IQR] per patient RMS error for n = 20 breaths was 0.75 [0.6 – 0.9] cmH2O, with elastance ranging from 2.1 – 29.8 cmH20/L, and per-patient median peak driving pressure ranging from -1.9 to -7.9 cmH2O. Inspiratory driving pressure profiles matched esophageal pressures from literature and its value at peak nervous signal to the diaphragm (Eadi) was correlated with peak Eadi (R2=0.25-0.86). Average trans-pulmonary pressure remained consistent between breaths for each patient, despite differences in peak Eadi and peak airway pressure. Overall, the model-based approach resulted in physiologically reasonable inspiratory driving pressures, with trends with electrical activity and matched literature data showing neuro-muscular decoupling as a function of pressure and/or volume.