Description of non-invasively determined

work of breathing per minute (WOBN)

Michael J. Banner PhD, Neil R. Euliano PhD, Andrea Gabrielli MD

University of Florida, College of Medicine, Department of Anesthesiology,

and Convergent Engineering, Gainesville, FL, USA

Work of breathing per minute or power of breathing can be calculated non-invasively with reasonable clinical accuracy for patients receiving pressure support ventilation (PSV) by using an Artificial Neural Network (ANN).1This method obviates the need for inserting an esophageal catheter, and, thus, greatly simplifies measurement of work of breathing. WOBN may be a clinically useful tool for consideration when setting PSV to unload the inspiratory muscles.

A laptop PC contains a pre-trained ANN using data from a previously approved IRB study of 200 adults with ARDS treated with PSV. Data presented in the aforementioned study 1were not used to train or develop the ANN. A multi-layer perceptron (MLP) ANN, which can be thought of as a nonlinear, multi-layer extension to a logistic regression system,2, 3, 4was trained with stable regions from a continuous recording of patient data. Stable two minute segments of data were extracted from a continuous recording and the average work for breathing and other parameters were calculated and used to train the ANN. For instance, the spontaneous minute ventilation parameter is the average minute ventilation over a selected two minute stable segment. No nursing interventions were performed during the data collection phase.

During the training period for the ANN, numerous possible predictor variables of WOBN were evaluated. For example, the spontaneous breathing frequency (f) to tidal volume (VT) ratio or rapid shallow breathing index (f / VT) was the first variable examined; no predictive relationship between it and invasively measured (esophageal catheter) work of breathing per minute was found. Similarly, no significant (p > 0.05) relationships were found when comparing invasively measured work of breathing per minute to the following: Ratio of inspiratory time to total breathing cycle time (TI / Ttot) (duty cycle), f, VT, lung carbon dioxide elimination rate, partial pressure end-tidal carbon dioxide (PetCO2), partial pressure mean exhaled carbon dioxide, PaCO2, PaO2, PaO2 / FIO2 ratio, respiratory system compliance, respiratory system resistance, inspiratory time constant, applied positive end expiratory pressure (PEEP), peak inflation pressure (PIP), and mean airway pressure.

Five predictor variables / input elements used in the ANN model, found to be highly correlated with invasively measured work of breathing per minute, were chosen because they produced the best (lowest mean squared error) possible predictive value. Incorporating inputs in addition to those five variables did not increase predictive performance, so this set was considered sufficient for predicting/calculating WOBN. (1) Minute ventilation, the spontaneous minute ventilation (not including Intermittent Mandatory Ventilation [IMV] breaths), was found to be closely tied to invasively measured work of breathing per minute. (2) Intrinsic positive end expiratory pressure (PEEPi) estimate, this parameter is calculated from the exhalation portion of a flow-volume loop. In patients in whom PEEPi was suspected, we observed that exhaled flow did not return to zero prior to the onset of inhalation for the next breath. Under this condition, flow does not reset / return to zero but some discrete value at end-exhalation on a flow-volume loop. By linearly extrapolating this value to the zero flow axis, an estimate of air-trapping or excess volume remaining in the lungs at end-exhalation is made. By knowing this volume and respiratory system compliance, a pressure is calculated (pressure = volume / compliance) that is a reflection of PEEPi.5, 6Respiratory system compliance (measured during the controlled IMV breaths) is determined by dividing VT by the static elastic recoil pressure of the respiratory system or plateau pressure obtained from the end inspiratory pause minus the level of applied PEEP (Compliance = VT / plateau pressure - PEEP). (This method assumes linear changes in compliance and may be a source of error.) (3) Trigger pressure depth, this parameter is the measured decrease in pressure below baseline airway pressure at the Y-piece of the breathing circuit just before the ventilator triggers "ON." Large inspiratory efforts tend to decrease this pressure more rapidly causing a large decrease in pressure or trigger pressure depth. (4) Inspiratory flow rise time, this parameter assesses how rapidly the inspiratory flow waveform rises during a PSV breath. An actively breathing patient with a strong inspiratory effort and demanding a high flow rate from the ventilator tends to display a rounded or sinusoidal shaped inspiratory flow profile, peak flow occurs during the mid to latter portion of the breath. It takes a longer time for flow to reach maximum during inhalation. In comparison, a patient who inhales passively while receiving PSV typically displays a rapid rise in flow very early in the breath. Flow reaches maximum at nearly the onset of the breath in a very brief time and then decelerates for the remainder of inhalation. A coded value for inspiratory flow rise time ranges from zero to one. The higher the value, the more rounded or sinusoidal the flow profile. Conversely, the lower the value, the higher the flow rate at the onset of inhalation and is associated with a decelerating inspiratory flow profile. (5) Respiratory muscle pressure, the sum of elastic and resistive pressures, was determined using the equation of motion, [i. e., Pressure = (VT / Respiratory system compliance) + (Respiratory system resistance x Inspiratory flow rate) ]. 7This is a reflection of the pressure generated by the respiratory muscles during spontaneous inhalation. The larger the value, the larger the estimated effort to inhale and vice versa. Respiratory system resistance (measured during the controlled IMV breaths) was calculated using the following equation: Resistance = PIP - Plateau pressure / Inspiratory flow rate.

References

  1. Banner MJ, Euliano NR, Brennan V, Peters C, Layon AJ, Gabrielli A (2006) Power of breathing determined non-invasively with use of an artificial neural network in patients with respiratory failure. Crit Care Med 34:1052–9
  2. Rodvold DM, McLeod DG, Brandt JM, Snow PB, Murphy GP (2001) Introduction to artificial neural networks for physicians: Taking the lid off the black box. Prostate 46: 39 – 44
  3. Rumelhart DE, McClelland JL, PDP Research Group (1986) Learning representations by error propagation, in Parallel Distributed Processing: Explorations in the Microstructure of Cognition. MIT Press, Cambridge MA, Vol. 1, pp. 318-62
  4. Principe J C, Euliano NR, Lefebvre KL (2000) Neural Systems: Fundamentals through

Simulation, John Wiley and Sons, New York.

  1. Fontan JJP, Heldt GP, Targett RC, Willis MM, Gregory GA (1986) Dynamics of

expiration and gas trapping in rabbits during mechanical ventilation at rapid rates.

Crit Care Med 14: 39 – 42

  1. Vinegar A, Sinnet EE, Leith DE (1979) Dynamic mechanisms determining functional

residual capacity in mice. J Appl Physiol 46: 867 – 870

  1. Milic-Emili J (1991) Work of breathing, in Crystal RG, West JB (eds): The Lung. New York, Raven Press, pp. 1065 - 1075