Highlights
– In silico “digital twins” of 98 ARDS patients compared airway pressure release ventilation (APRV) with documented pressure-controlled ventilation (PCV).
– APRV settings (Phigh 25 cm H2O, Plow 0 cm H2O, Tinsp 5 s, Tlow ≈ time to 75% peak expiratory flow, mean 0.5 s) reduced mean mechanical power by 32% and mean tidal alveolar recruitment/de-recruitment by 34% versus observed PCV.
– Driving pressure, tidal volume, and static stress/strain were similar between modes; APRV produced moderate hypercapnia (mean PaCO2 58.5 mm Hg, pHa 7.32).
– Computational optimization (over 4.8 million setting variations) suggested these APRV parameters were near-optimal for minimizing mechanical power and tidal recruitment while preserving gas exchange.
Background
Ventilator-induced lung injury (VILI) remains a major determinant of outcome in patients with acute respiratory distress syndrome (ARDS). Strategies that reduce cyclic lung stress, overdistension, and repetitive alveolar opening/closing can potentially reduce VILI. Conventional protective ventilation emphasizes low tidal volumes and attention to plateau and driving pressures; more recently, integrative metrics such as mechanical power have been proposed to capture the combined injurious effects of pressure, volume, flow and respiratory rate. Airway pressure release ventilation (APRV) is a time-cycled pressure mode that maintains a high continuous distending pressure with brief releases to facilitate CO2 removal. APRV may reduce cyclic recruitment–derecruitment and limit mechanical power, but randomized outcome data are sparse and physiological effects can be heterogeneous.
Study design
This modeling study by Joy et al. (Crit Care Med. 2025) used high-fidelity cardiopulmonary simulation models to build patient-specific digital twins for 98 patients with ARDS. The dataset comprised pairs of clinician-set ventilator parameters and arterial blood gas measurements recorded while patients were receiving pressure-controlled ventilation (PCV). Each digital twin was tuned to the patient data, then used to calculate multiple VILI-relevant indices for the documented PCV settings and for APRV settings simulated in two modes: fixed and time-controlled adaptive.
Investigators used global optimization algorithms to evaluate over 4.8 million combinations of ventilator settings to identify parameter sets that minimized indices of VILI while maintaining acceptable gas exchange. Key outcomes calculated in the digital twins included mechanical power (MP), tidal alveolar recruitment/de-recruitment, driving pressure, tidal volume, and lung stress/strain. APRV parameterization of interest was Phigh 25 cm H2O, Plow 0 cm H2O, Tinsp 5 s, with Tlow adjusted to terminate at 75% of peak expiratory flow (mean Tlow ≈ 0.5 s).
Key findings
Reduction in mechanical power and tidal recruitment. In the tuned digital twins, the APRV configuration described above reduced mean mechanical power by 32% and mean tidal alveolar recruitment–derecruitment by 34% compared with the actual PCV settings documented in the 98 patients. These reductions were consistent across the cohort and were reproduced by global optimization procedures that explored a very large parameter space.
Gas exchange and acid–base effects. APRV achieved lower mechanical power and tidal recruitment while preserving oxygenation targets used in the simulations. However, APRV produced moderate hypercapnia: mean PaCO2 increased to 58.5 mm Hg (pHa ~7.32) versus mean PaCO2 45.6 mm Hg (pHa ~7.37) with PCV. The authors framed this hypercapnia as a trade-off that may be clinically acceptable in many ARDS patients (permissive hypercapnia) but requiring consideration of hemodynamic and neurologic tolerance.
Driving pressure, tidal volume, and static stress/strain. Mean driving pressure, tidal volume, and calculated lung stress/strain were similar between APRV and the recorded PCV in the cohort—suggesting that the observed reductions in mechanical power and tidal recruitment were not driven by reductions in plateau pressure or tidal volume alone but by the specific time-pressure pattern of APRV.
Optimization results. The computational optimization (>4.8 million simulations) identified APRV parameter regions that minimized mechanical power and tidal recruitment while respecting gas-exchange constraints. The empiric APRV setting (Phigh 25, Plow 0, Tinsp 5 s, Tlow to 75% peak expiratory flow) used in the main analysis was close to the Pareto-optimal solution for the outcomes evaluated.
Population and modeling scope. The study used digital twins fit to real patient data, which permits personalized, physiology-based comparisons between modes without exposing patients to experimental protocols. Nonetheless, the underlying results remain model-based and not a substitute for randomized clinical trials.
Expert commentary and mechanistic insights
What these findings mean mechanistically. Mechanical power quantifies the energy delivered to the respiratory system per unit time and integrates multiple determinants of VILI (pressure amplitude, volume, flow, respiratory rate). APRV maintains a high continuous distending pressure for prolonged durations, minimizing the number of large cyclical pressure–volume excursions and limiting repetitive alveolar opening and closing. Short, controlled release times (Tlow) that terminate at a fraction of peak expiratory flow are intended to prevent full deflation and thus limit tidal recruitment–derecruitment. Together, these features can reduce the energy transmitted to the parenchyma with each breath and the number of injurious cyclical events—consistent with the reductions seen in mechanical power and tidal recruitment in the digital twins.
Clinical plausibility and alignment with prior evidence. The ARDS Network low-tidal-volume strategy remains foundational to lung-protective ventilation. Observational and physiologic work (e.g., Amato et al., NEJM 2015) has emphasized the importance of driving pressure as a key mediator of outcome. Mechanical power is a newer concept aiming to integrate injurious determinants. The digital twin results are consistent with the idea that time-domain modification of the pressure waveform (as in APRV) can alter mechanical power and micro-atelectasis dynamics even when tidal volume and driving pressure are similar.
Limitations and caveats
Model-based inference. Digital twins are powerful hypothesis-generating tools but depend on model structure and the accuracy of parameter fitting. The simulation assumes that the tuned cardiopulmonary model captures relevant patient physiology and the complex interactions among recruitment, compliance heterogeneity, hemodynamics, and spontaneous breathing—and that patient responses to APRV can be translated from the model to the bedside.
Dataset constraints. All real patient inputs were recorded during PCV; the APRV scenarios were simulated rather than observed clinically in these patients. The cohort size (n=98) is reasonable for modeling but may not represent all ARDS phenotypes (e.g., focal vs. diffuse, body habitus, comorbidities) or the spectrum of hemodynamic responses.
Spontaneous breathing and patient–ventilator interaction. APRV is often used to promote spontaneous breathing, which can be protective in some contexts but harmful if strong inspiratory efforts amplify transpulmonary pressures (patient self-inflicted lung injury). The study primarily considered passive mechanics as recorded during PCV; how spontaneous efforts would modify the conclusions is uncertain.
Hypercapnia and clinical tolerability. The simulated APRV strategy produced moderate hypercapnia. Permissive hypercapnia may be acceptable in many ARDS cases but has limits in unstable hemodynamics, raised intracranial pressure, or severe acidosis. Any ventilatory approach must be integrated with the patient’s overall physiology and goals.
Implications for practice and research
These digital twin results suggest APRV configured with a high continuous distending pressure and short expiratory releases tailored to flow termination can materially reduce mechanical power and tidal recruitment compared with clinician-selected PCV settings, without worsening oxygenation targets but with increased PaCO2. This provides a physiologically coherent rationale to evaluate APRV in prospective clinical trials focused on patient-centered outcomes, as well as mechanistic intermediate endpoints (biomarkers of lung injury, imaging-based recruitment, and physiologic measures of regional strain).
Recommended next steps include carefully controlled physiological studies in patients (or pilot randomized trials) that rigorously monitor lung mechanics, regional ventilation (e.g., electrical impedance tomography, CT when feasible), hemodynamics, and markers of lung injury, and that define protocols for titration of Tlow and sedation/spontaneous breathing. Trials should prespecify strategies to manage hypercapnia, assess feasibility and safety, and identify subgroups most likely to benefit.
Conclusion
Using patient-specific digital twins, Joy et al. provide model-based evidence that APRV—when configured with prolonged high-pressure periods and very short, flow-terminated releases—can reduce mechanical power and tidal recruitment versus recorded PCV in ARDS patients, albeit with predictable increases in PaCO2. These findings are hypothesis-generating and support the rationale for prospective physiologic and randomized clinical studies to determine whether the modeled benefits translate into reduced VILI and improved clinical outcomes.
Funding and clinicaltrials.gov
The primary article reports funding and acknowledgements; readers should consult the original publication for specific funding sources. No clinical trial registration applies to this modeling study; future clinical investigations of APRV should be registered prospectively on clinicaltrials.gov or equivalent registries.
References
1. Joy W, Albanese B, Oakley D, et al. Digital Twins to Evaluate the Risk of Ventilator-Induced Lung Injury During Airway Pressure Release Ventilation Compared With Pressure-Controlled Ventilation. Crit Care Med. 2025 Dec 1;53(12):e2573-e2582. doi:10.1097/CCM.0000000000006885 .
2. Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med. 2000 May 4;342(18):1301-8.
3. Amato MBP, Meade MO, Slutsky AS, et al. Driving pressure and survival in the acute respiratory distress syndrome. N Engl J Med. 2015 Mar 26;372(8):747-55.