Noninvasive Estimation of Lung Parameters via Constrained Optimization: Toward Improved Ventilation

This paper presents a noninvasive method to estimate lung parameters (resistance and elastance) in mechanically ventilated patients. Currently, estimation of lung parameters requires invasive measures (e.g. esophageal pressure) or, more frequently, a ventilator maneuver that involves a short pause in air delivery (e.g. airway occlusion). This disrupts the desired ventilation operation and causes pain or discomfort when the pause is long or the patient is not fully sedated. The proposed method (interior point method, IPM) estimates these parameters noninvasively, requiring only routinely available airway pressure, flow and volume, and by so doing, IPM, hence has the potential to usher a new and safer ventilation method. The widely accepted single-compartment model of the respiratory system is used to formulate a modified objective function for the constrained optimization problem. Physiological constraints on the lung parameters (e.g. positivity) have been included in the cost function through barrier functions to restrict the space of feasible solutions. IPM, using steepest descent with a Newton-Raphson step and backtracking line search, was employed to solve the optimization problem. The method was validated through experiments on artificial lungs (IngMar QuickLung®) while ventilated using the Viasys Avea® ventilator in pressure-regulated volume control (PRVC) mode and compared to standard estimation methods like least-squares estimation (LSE), clinical formulae, and the ventilator's own reported values.Clinical Relevance- This allows clinicians to obtain reliable and noninvasive lung health estimates in PRVC, without requiring maneuvers or invasive measures, thus preventing patient pain/discomfort and aiding in diagnosis/treatment through better ventilator management.