Model-based decision support for pressure support mechanical ventilation - implementation of physiological and clinical preference models

The paper presents modifications of the INVENT decision support system for providing decision support in pressure support mechanical ventilation. Physiological models of interstitial fluid and tissue buffering and metabolism, cerebrospinal fluid acid-base status, and central and peripheral chemoreflex respiratory drive were integrated with existing models of the system and methods were implemented for learning effect of changes in pressure support on tidal volume, metabolism, anatomical dead space and muscle response. Preference models were implemented describing risk of muscle atrophy and stress in relation to pressure support level. The system was evaluated retrospectively in three patients responding differently to changes in pressure support. The system was able to describe patients' changes in tidal volume, respiratory rate and alveolar ventilation and adapt decision support appropriately when patient responses were different from that expected from measurements at baseline level.