Automatic adjustment of pressure support by acomputer-driven knowledge-based system during noninvasive ventilation: afeasibility study

Objective: To evaluate the feasibility of using aknowledge-based system designed to automatically titrate pressure support (PS) to maintain the patient in a"respiratory comfort zone” during noninvasive ventilation (NIV) in patients with acute respiratory failure. Design and setting: Prospective crossover interventional study in an intensive care unit of auniversity hospital. Patients: Twenty patients. Interventions: After initial NIV setting and startup in conventional PS by the chest physiotherapist NIV was continued for 45 min with the automated PS activated. Measurements and results: During automated PS minute-volume was maintained constant while respiratory rate decreased significantly from its pre-NIV value (20 ± 3 vs. 25 ± 3 bpm). There was atrend towards aprogressive lowering of dyspnea. In hypercapnic patients PaCO2 decreased significantly from 61 ± 9 to 51 ± 2 mmHg, and pH increased significantly from 7.31 ± 0.05 to 7.35 ± 0.03. Automated PS was well tolerated. Two system malfunctions occurred prompting physiotherapist intervention. Conclusions: The results of this feasibility study suggest that the system can be used during NIV in patients with acute respiratory failure. Further studies should now determine whether it can improve patient-ventilator interaction and reduce caregiver workloa