Can new smart vents beat intensivists in the MICU? (ACTIVE trial)
A large RCT tests closed loop adaptive support ventilation in critical illness
New “intelligent” ventilator modes purport to wean patients faster (and safely), with minimal hands-on adjustment by clinicians. These go by a variety of names and have under-the-hood differences, but conceptually speaking, they dynamically and algorithmically hybridize volume control (in which a set volume is delivered, potentially producing high pressures) and pressure control (in which a set pressure is delivered, potentially producing low volumes due to low flow). Because volume and pressure are effectively interchangeable, smart vents blur the distinctions between vent modes and even start to dissolve the concept itself.
We dived into some of the technical basics here:
Perhaps the most sophisticated of the new class of vent modes is closed loop adaptive support ventilation.
ASV is a complex, proprietary, mostly autonomous mode in which the desired “percent minute ventilation” is entered by the clinician, and the vent dynamically adjusts pressure (and therefore volume) and respiratory rate to achieve that minute ventilation target on a rolling basis, indexed to the patient’s ideal body weight. Closed loop ASV systems can even dynamically adjust oxygen delivery (FiO2) and PEEP, and ventilation in response to changes in end-tidal CO2.
Vents with close loop smart modes installed are marketed as supporting weaning by automatically reducing support as the patient’s spontaneous effort increases.
“Smart” modes’ autonomy permits them to deviate from volumes and pressures, potentially substantially, but no significant adverse safety signal has emerged from published studies.
In randomized trials, many funded by vendors, ASV often reduced the time to extubation without complications or increased clinician involvement.
A majority of these trials were conducted in cardiac surgery and other postoperative populations.
The data on ASV in critically ill patients generally is of lower quality and no firm conclusions should be drawn from it. A review of ten randomized trials testing one vendor’s system concluded that it was safe but not clearly superior to conventional ventilation in terms of patient-centered outcomes.
A new large randomized trial more thoroughly compared ASV with conventional ventilation in critical illness.
The ACTIVE Trial
At seven ICUs in the Netherlands and Switzerland, 1,514 critically ill adults were randomized to receive INTELLiVENT® adaptive support ventilation (within an hour) or protocolized conventional mechanical ventilation. About 300 patients withdrew or did not confirm their (initially deferred) consent, and 1,201 were analyzed.
Optional presets for acute respiratory distress syndrome, chronic hypercapnia, or brain injury could be selected on the ASV menu. About 86% of patients had nonsurgical indications for mechanical ventilation—almost all either primary respiratory failure, cardiac arrest, or neurologic injury.
Automatic SBTs
The vents (from Hamilton Medical, outfitted with its INTELLiVENT adaptive support ventilation software) included the option for automatic spontaneous breathing trials.
In auto-SBTs, patients who the vent identifies as spontaneously breathing with stable oxygenation and without frequent alarms are transitioned to 5 cmH2O, 5 cm of PEEP, for 30 minutes by default.
Those who fail the SBT are returned to previous settings; those who pass (without sustained hypoxemia, tachypnea, etc) are maintained on minimal support, with the display indicating a successful SBT.
Results
At 28 days, median ventilator-free days were virtually identical (16.7 vs 16.3), the primary outcome.
There were no differences in 28-day mortality (~27% in each) or complications between groups.
Extubation failure (reintubation within 24 hours) occurred virtually equally in both groups (7.7% vs 7.3%).
Those receiving ASV had numerically fewer episodes of severe hypoxemia and hypercapnia. These were statistically significant, but not after adjustment for multiple variables. (This adjustment is performed to safeguard against finding spurious differences by chance when multiple secondary outcomes are tested.)
ASV patients had improved “ventilation quality” in the first 6 hours, which was an interesting parameter, representing the time patients spent in predefined optimal zones of respiratory physiology:
The concept of “ventilation quality” was not visible to clinicians and was not part of bedside decision-making. It was computed from recorded ventilator data to evaluate adherence to lung-protective and physiologic targets. However, clinicians were aware that ventilation quality was one of the trial outcomes, and the predefined zones reflected the same lung-protective ventilation targets described and reinforced during training.
Among patients with full data available, those receiving ASV were 1.5 times more likely to be in a more optimal zone of ventilation than conventionally vented patients.
Only 153 patients (a tenth of the total enrolled) had granular ventilator data available to be considered in the “ventilation quality” analysis.
The Dutch government funded the trial.
But, But, But
By the standards of modern ICUs, this was an “easy” cohort to work with from a standpoint of respiratory mechanics. Only one-third were intubated for respiratory failure, and the overall cohort had high oxygenation and lung compliance as a group. Most were intubated for neurologic reasons (post-cardiac arrest, stroke, etc), and most were either weaned off ventilation (~40%) or had died (~20%) within three days of enrollment. Only about 25% of patients in each group were intubated for nine days or longer (and were numerically fewer in the conventional group).
Based on the patient mix, many of those left probably remained intubated for encephalopathy. No ventilator mode is smart enough to analyze that, yet. This would skew the results toward neutrality.
Patients at the study centers received a very high intensity of care, being screened three times daily for readiness to extubate. This would have an unpredictable effect on the outcome, but limits its extrapolability to most centers following the care standard of daily assessments.
The characteristics of the 20% who withdrew their consent are unknown, or how their inclusion would have influenced the outcomes.
Discussion
In the ACTIVE trial, automated adaptive support ventilation led to similar durations of ventilation among a heterogeneous group of critically ill patients, who were primarily not intubated for primary respiratory failure, and who were usually extubated (or had died) within 3 days.
Although one could argue this population was not truly representative, “smart” ventilator modes are also being tested against conventional ventilation in more complex and difficult-to-wean patients. These trials have generally been small and single-center, but several have shown equivalent or faster weaning times with ASV [1, 2]. A Cochrane review concluded, “Automated closed loop systems may result in reduced duration of weaning, ventilation, and ICU stay. Reductions are more likely to occur in mixed or medical ICU populations.” That was eleven years ago, analyzing trials conducted with relatively ancient iterations of the software.
Although they’re not ready for wholesale deployment, smart vent modes’ performance appears to be good and will keep getting better. It makes sense to assume that deidentified patient data is being shared in agreements between health systems and ventilator manufacturers to improve the models.
Of course, large multicenter trials are necessary to draw any firm conclusions, and these have yet to be performed.
But there’s a saying that if you and a companion are running from a bear, you don’t have to be faster than the bear—you only need to be faster than the other guy.
Smart vent modes may never surpass an experienced intensivist on the most complex patients, but they don’t have to. From a health services administration standpoint, near-parity in performance and safety by the bot will be a no-brainer hiring decision over the far more expensive human “provider.” This could be an important accelerant of the trend toward midlevel-run ICUs with minimalist physician staffing.
There may be a positive story here, too, which is that off-loading ordinary ventilator weaning and knob-twiddling to algorithms should free up much-needed time and energy for the higher-level cognitive, emotional, and relational work that can so often be crowded out by the multitude of momentary perturbations in critically ill patients’ physiology that repeatedly divert our attention from larger matters.
References
Sinnige JS, Horn J, Paulus F, et al. Effect of Automated Closed-Loop Ventilation vs Protocolized Conventional Ventilation on Ventilator-Free Days in Critically Ill Adults. JAMA. Published online December 8, 2025. doi:https://doi.org/10.1001/jama.2025.24384
Intraoperative Driving Pressure–Guided High PEEP vs Standard Low PEEP for Postoperative Pulmonary Complications. JAMA. Published online December 3, 2025:e2523540. doi:https://doi.org/10.1001/jama.2025.23540
Tsai YC, Jhou HJ, Huang CW, Lee CH, Chen PH, Hsu SD. Effectiveness of adaptive support ventilation in facilitating weaning from mechanical ventilation in postoperative patients. Journal of Cardiothoracic and Vascular Anesthesia. Published online April 24, 2024. doi:https://doi.org/10.1053/j.jvca.2024.04.030
Rose L, Schultz MJ, Cardwell CR, Jouvet P, McAuley DF, Blackwood B. Automated versus non-automated weaning for reducing the duration of mechanical ventilation for critically ill adults and children. Cochrane Database of Systematic Reviews. Published online June 10, 2014. doi:https://doi.org/10.1002/14651858.cd009235.pub3
Trials testing
The bots will NOT take our jobs as Intensivists after all...