(M-035) Predicting Improved Weaning Strategies for Mechanical Ventilation with a Physiologically-Based Mechanistic Model

Stefanos Papadopoulos – Chemical and Petroleum Engineering – University of Pittsburgh; Gilles Clermont – Critical Care Medicine – University of Pittsburgh; Robert Parker – Chemical and Petroleum Engineering – University of Pittsburgh

Objectives: Mechanical ventilator supports lung function through the control of breathing parameters such as fraction inspired oxygen (FIO2) and positive end expiatory pressure (PEEP). Prolonged mechanical ventilation with high PEEP can cause ventilator-induced lung injury (VILI) and is associated with a 14% increase in mortality [1]. A physiologically-based model, integrating organ level functions with a game theoretic approach for cellular adaptation, is used to compare predicted ventilation duration for two weaning protocols with varying acute respiratory distress syndrome (ARDS) severity.

Methods: Our previous multi-compartment model for oxygen, waste, and inflammation trafficking between the lung, liver and kidney is integrated with a control algorithm for implementation of FIO2 and PEEP [2]. A pathogen insult within the lung compartment leads to the development of ARDS. The pathogen acts as a disturbance triggering a systemic inflammatory response and decreasing lung function, which leads to controller activation. In a forthcoming publication we compared two clinically validated ventilation protocols in their respective improvement in virtual patient outcome, as measured by recovery, survival, or mortality. The controller maintained oxygen saturation within desired bounds (92-98%), prolonging survival and allowing recovery. Here, we compare two weaning protocols: a traditional static weaning approach, where FIO2 and PEEP are decreased gradually when oxygen saturation reaches 98% versus dynamic weaning, where PEEP is reduced to 0 cm H2O when the change in the oxygen partial pressure and FIO2 (P/F) ratio is greater than 20 mmHg over the previous 24 hours and above 175 mmHg. If oxygen saturation falls below 90%, PEEP is set to the previous setting and weaning cannot be reattempted for 12 hours.

Results: Dynamic simulations showed the use of dynamic weaning criteria decreased ventilation duration compared to static criteria. Across a two-dimensional inflammatory parameter space (pro- vs anti-inflammatory generation), the use of dynamic weaning reduced average ventilation time, but did not alter the size of the recovery envelope (organ failure caused by ARDS occurs outside the envelope). The weaning criteria only plays a role once simulated patient recovery has begun, so neither strategy altered the predicted outcome (recovery or failure). Across the simulated patient population, dynamic weaning decreased the average ventilation time by 23 hours. Simulated subjects needing multiple wean attempts still had a shorter ventilation time than those treated using static criteria. These subjects are near the boundary of recovery and failure, so recovery is limited by the time it takes inflammation and tissue damage to resolve.

Conclusions: Dynamic weaning criteria may reduce ventilation duration, reducing the risk of VILI as well as mortality. Integration of mechanical ventilation into the previously developed organ failure model can characterize the impact of organ failure on the progression and treatment of ARDS in silico.

Citations: [1] Lucile Laporte, Coralie Hermetet, Youenn Jouan, et al. Ten-year trends in intensive care admissions for respiratory infections in the elderly. Ann Intensive Care, 8(1):84, 2018.
[2] Stefanos Papadopoulos, Gilles Clermont, and Robert S. Parker. Modeling organ failure under hypoxic stress. IFAC-PapersOnLine, 55(23):33–34, 2022.

Keywords: nonlinear dynamics, mechanical ventilation, feedback control