Retrospective evaluation of a decision support system for controlled mechanical ventilation

Area of Science:

• Critical Care Medicine
• Biomedical Engineering
• Medical Informatics

Background:

• Mechanical ventilation management in intensive care is complex due to competing clinical objectives.
• Decision support systems (DSS) can aid clinicians in balancing these goals.
• Existing systems may not fully address the nuances of optimizing ventilation parameters.

Purpose of the Study:

• To evaluate a computerized, model-based DSS for its recommendations on inspired oxygen fraction, tidal volume, and respiratory frequency.
• To assess the DSS's performance against established mechanical ventilation best practices.
• To determine the robustness of the DSS's advice concerning variations in cardiac output.

Main Methods:

• Retrospective evaluation of the DSS in 16 intensive care patient cases.
• Physiological models were fitted to retrospective data to simulate patient responses to therapy changes.
• Sensitivity analysis was performed to evaluate the DSS's advice in response to cardiac output variations.

Main Results:

• The DSS recommended lower tidal volumes and inspired oxygen fraction, with a higher respiratory frequency compared to baseline settings.
• Simulated outcomes aligned well with the Acute Respiratory Distress Syndrome Network (ARDSNet) 2000 goals.
• Advice variations due to approximately 20% cardiac output changes were negligible, except in high oxygen consumption scenarios.

Conclusions:

• The computerized DSS provides clinically relevant and rational mechanical ventilation advice.
• The system's recommendations are consistent with current 'best practice' in critical care.
• The DSS advice demonstrates robustness against variations in cardiac output, enhancing its clinical utility.