Comparative Virtual Trials: Pressure-Controlled versus Volume-Controlled Stochastic Integrated Model-Based Mechanical Ventilation Protocols

Pressure-controlled (PC) ventilation is a pressure-targeted mode of mechanical ventilation (MV) to prevent barotrauma by using pre-set driving pressures. However, the lack of clinician consensus on approaches towards pressure-targeted MV management can lead to suboptimal MV treatment and adverse patient outcomes. This study presents an in-silico trial for the validation of a pressure-based, stochastic model-integrated SiVENT protocol to guide this care. The PC SiVENT protocol uses a respiratory elastance (Ers) stochastic model and model-based methods to recommend a safe combination of MV input settings. The protocol was implemented on PC virtual patients (VP) (N=29; 2,630 hours MV data) in a virtual trial to derive protocol-suggested MV settings and resulting VP responses to these MV settings. Virtual trial outcomes were compared with the retrospective clinical data and similar volume-control (VC) SiVENT virtual trials, by assessing the compliance rate of measured VP responses to established clinical safety thresholds. Out of 29 PC VPs, 3 retrospective VPs (∼10%) and 23 SiVENT VPs (∼79%) achieve an all-compliance rate of > 80%. The results of the virtual trials demonstrate the efficacy of the SiVENT PC in regulating pressure-based MV outcomes, as well as the potential to safely streamline the development, optimisation and validation process of new MV decision support systems. The protocols presented would make measurable advances in individualised care with significant potential following improved patient outcomes in MV. The overall results justify prospective clinical trials.