PURPOSE: Appropriate assessments of respiratory muscle loads, reflected by power of breathing (work of breathing/min), and tolerance of these loads, reflected by spontaneous breathing frequency (f) and tidal volume (VT), are essential for patient management. Pressure support ventilation (PSV) should be applied so that muscle loads are not too high or low, predisposing to fatigue and disuse atrophy, respectively. Under some conditions, clinicians to manage patients with respiratory failure may not always be available to make timely, proper assessments for setting PSV. For these conditions, we propose a respiratory monitor employing a fuzzy logic inference system using an artificial neural network for the noninvasive, real time calculation of power of breathing (POBn), combined with f and VT to make recommendations for setting PSV. To validate these recommendations, we performed a multi-center study.
METHODS: Adults (133) from three centers (age 59 ± 16 yrs, wt. 80 ± 24 kg) with respiratory failure receiving PSV were enrolled in an IRB approved study. Data from a combined pressure/flow sensor, positioned between the endotracheal tube and Y-piece of the ventilator breathing circuit, were directed to a monitor (NICO, Respironics) and computer (Convergent Engineering) for measurements of POBn, f and VT. Recommendations from the monitor for setting PSV were examined prospectively for validity (accuracy) by expert attending intensivists from all three centers. Data analysis: Fisher Sign Test; alpha was set at 0.05 for significance.
RESULTS: PSV range for all subjects was 5-25 cm H2O. There was a mean total of 92.3% agreement (p<0.05) (UF: 149 agree/163 recommendations = 91.5%; Duke: 68 agree/72 recommendations = 94.5%; Manitoba: 254 agree/279 recommendations = 91%) between all recommendations from the monitor to increase or decrease PSV and those of the panel of experts.
CONCLUSION: Valid recommendations for setting PSV to appropriately unload the respiratory muscles were provided by the monitor.
CLINICAL IMPLICATIONS: A respiratory monitor employing a complimentary Load (POBn) and Tolerance (f and VT) strategy for assessing patients with respiratory failure, may provide clinically valid recommendations for setting PSV.
DISCLOSURE: Andrea Gabrielli, University grant monies Michael Banner receives grant money from Convergent Engineering; Shareholder Neil Euliano is the President of Convergent Engineering; Product/procedure/technique that is considered research and is NOT yet approved for any purpose, artificial neural network.