In Silico Trial for Preclinical Evaluation of an Automated Weaning Protocol

Abstract

In silico clinical trials (ISCTs) may reduce risks of automated ventilation by evaluating performance across physiologic variability with traceable model credibility. We present a case study using a patient–device model (PDM) to assess an automated weaning function in ICU patients. The workflow defines the question of interest, context of use, cohort/scenario generation, sampling, execution, and analysis. A virtual ICU cohort captures variability in demographics and pathophysiology (ARDS, COPD, postoperative, neuromuscular, cardiopulmonary) using literature-based parameters. Nine scenarios test robustness, including elastance/resistance steps, drive shifts, shunt/dead-space increases, and sedation taper. Primary endpoints are time in target for tidal volume (V_T), respiratory rate (RR), and end-tidal CO₂ (etCO?), excursion durations, and overshoot. Respiratory rate and etCO? remained ?80% in target, while V_T only remained in target range about 50 % due to narrow protective bounds. This compact ISCT provides a credibility-aware framework for evaluation of automated protocols, shown here as an example for weaning.