Experimental animal data suggest that increasing breathing pattern variability in mechanical ventilation could be beneficial. Variable ventilation can be induced through the following modes: Neurally Adjust Ventilatory Assist (NAVA), Proportional Assist Ventilation (PAV) and Variable-Pressure Support Ventilation (V-PSV). These modes have not yet been compared to each other. Pilot observations in our department suggest a feasibility in patients.

The objectives of the study are to compare the impact of PSV, NAVA, PAV and V-PSV on the variability of the breathing pattern, patient-ventilator asynchrony, risk of lung overdistension, gas exchange, and repartition of ventilation.

Cross-over, prospective, randomized controlled monocentric trial . Patients on pressure support ventilation will be ventilated with PSV, NAVA, PAV, and V-PSV (in a randomized sequence).

In each mode the level of assistance will be set to achieve a similar tidal volume of 6-8 ml/kg.

The following data will be measured and recorded: airway flow and pressure, electrical activity of the diaphragm (EAdi), blood gases, electrical impedance tomography, end tidal partial pressure in carbon dioxide (PEtCO2).

Will be calculated: the coefficient of variation (CV, standard deviation (SD)/mean) of the peak pressure (Ppeak), EAdi and of the main descriptors of the breathing pattern, the prevalence of the main patient-ventilator asynchronies, the prevalence of tidal volume (VT) >10ml/kg, ventrodorsal repartition of ventilation and inhomogeneity index, dead space

Inclusion Criteria:

• patient on mechanical ventilation for >48 h
• estimated duration of mechanical ventilation >24h
• hemodynamic stability

Exclusion Criteria:

• pregnancy
• impossibility to insert an EAdi catheter
• neuromuscular disease, phrenic nerve lesions