Real-Time Algorithm-Driven Ventilation Feedback to Improve Lung-Protective Ventilation in Critically Ill Patients

Overview

The REALVENT trial is designed to evaluate whether a real-time, algorithm-driven ventilation feedback strategy can improve lung-protective ventilation (LPV) achievement rates in critically ill patients receiving invasive mechanical ventilation. This multicentre randomised controlled trial will compare real-time respiratory waveform monitoring with automated feedback against standard ICU care. The primary endpoint is the LPV achievement rate over the first 72 hours.

Mechanical ventilation is essential in modern intensive care but may cause ventilator-induced lung injury (VILI) when delivered with excessive tidal volume, airway pressure, or mechanical power, or in the presence of unrecognised patient-ventilator asynchrony. Despite guideline recommendations to limit tidal volume, plateau pressure, and driving pressure, real-world adherence to lung-protective ventilation (LPV) remains suboptimal, and clinicians often rely on intermittent, manual review of ventilator settings and waveforms. The REALVENT trial tests a cloud-based respiratory dynamics monitoring and feedback system that continuously acquires high-frequency ventilator waveforms (pressure, flow, volume) and automatically computes key LPV metrics, including tidal volume indexed to predicted body weight, driving pressure, plateau pressure, mechanical power, and patient-ventilator asynchrony events. For patients in the intervention arm, the platform provides three layers of feedback over the first 72 hours after randomisation: (1) real-time alerts when LPV thresholds are exceeded; (2) 4-hour window indicator checks to capture sustained deviations; and (3) standardised 24-hour summary reports with recommendations for ventilator adjustment. These reports are reviewed by bedside clinicians and a central monitoring team, but all treatment decisions remain at the discretion of the local ICU team. The control group receives usual care with standard bedside ventilator monitoring but without structured feedback from the platform. All other aspects of care, including fluid management, sedation, prone positioning, neuromuscular blockade, and adjunct respiratory monitoring (e.g., esophageal manometry or EIT), are left to clinician judgement and recorded. The primary hypothesis is that algorithm-driven feedback will increase the proportion of time during the first 72 hours that all four LPV targets are simultaneously achieved compared with standard care. Secondary hypotheses are that improved LPV adherence will translate into more ventilator-free days, fewer ventilator-associated complications, lower inflammatory biomarker levels, and acceptable clinician workload and usability ratings.