A virtual patient model for mechanical ventilation.

Area of Science:

• Respiratory physiology
• Computational modeling

Background:

• Mechanical ventilation (MV) is crucial for acute respiratory failure but carries risks of lung injury from suboptimal settings.
• Heterogeneity in lung tissue necessitates precise ventilator adjustments, particularly positive-end-expiratory-pressure (PEEP).

Purpose of the Study:

• To develop and validate a physiologically relevant respiratory model for predicting lung mechanics under varying PEEP levels.
• To enhance clinician confidence in adjusting MV settings for lung protection.

Main Methods:

• A single-compartment lung model incorporating elastance and resistance using basis functions was developed.
• The model is personalized using low PEEP data to predict lung mechanics at higher PEEP levels.
• Validation was performed using data from four patients and eight recruitment maneuver arms.

Main Results:

• The model demonstrated low prediction error for peak inspiratory pressure across a clinically relevant pressure range.
• Predictions showed less than 10% error for over 90% of PEEP changes up to 12 cmH2O.
• The model accurately captures lung elastance and resistance.

Conclusions:

• The developed in-silico model provides a method for predicting patient responses to MV therapy changes.
• This forms the basis for a virtual patient simulator for mechanical ventilation.
• The model supports safer and more effective ventilator management.