Coordination of respiratory muscles assessed by means of nonlinear forecasting of demodulated myographic signals

Abstract

Pulmonary diseases such as obstructive sleep apnea syndrome (OSAS) affect function of respiratory muscles. Individuals with OSAS suffer intermittent collapse of the upper airways during sleep due to unbalanced forces generated by the contraction of the diaphragm and upper airway dilator muscles. Respiratory rhythm and pattern generation can be described via nonlinear or coupled oscillators; therefore, the resulting activation of different respiratory muscles may be related to complex nonlinear interactions. The aims of this work were: to evaluate locally linear models for fitting and prediction of demodulated myographic signals from respiratory muscles; and to analyze quantitatively the influence of a pulmonary disease on this nonlinear forecasting related to low and moderate levels of respiratory effort. Electromyographic and mechanomyographic signals from three respiratory muscles (genioglossus, sternomastoid and diaphragm) were recorded in OSAS patients and controls while awake during an increased respiratory effort. Variables related to auto and cross prediction between muscles were calculated from the r(2) coefficient and the estimation of residuals, as functions of prediction horizon. In general, prediction improved linearly with higher levels of effort. A better prediction between muscle activities was obtained in OSAS patients when using genioglossus as the predictor signal. The prediction was significant for more than two respiratory cycles in OSAS patients compared to only a half cycle in controls. It could be concluded that nonlinear forecasting applied to genioglossus coupling with other muscles provides a promising assessment to monitor pulmonary diseases.

Conflict of interest statement

Conflict of interest statement

The authors declare that there are no matters of conflict of interest arising in this study.

Copyright © 2011 Elsevier Ltd. All rights reserved.

Figures

Figure 1

Examples of surface EMG (upper traces) and MMG signals (lower traces) of the sternomastoid. Gray traces represent filtered myographic signals and black lines correspond to their demodulated counterparts.

Figure 2

Example of cross-prediction for two demodulated EMG signals along time. Different respiratory cycles are indicated by different background shadings. Black and gray traces correspond to two demodulated EMG signals from different muscles, and circles indicate the samples of one cycle that are used by locally linear models to predict another sample located a respiratory period ahead. DT and PH indicate delay time and prediction horizon respectively.

Figure 3

Examples of r2 coefficient and error estimation (as functions of prediction horizon) and their associated variables calculated on data corresponding to a single patient.

Figure 4

Time course of the r2 coefficient and the error estimation for a typical patient and a typical healthy subject at low level of effort.

Figure 5

Mean and standard deviation of PH07, RP, R0 and ErrorPT corresponding to cross-prediction between GEN_EMG and SMM_EMG, shown as a function of %MMP. Gray solid lines represent healthy subjects and black dotted lines refer to OSAS patients. Symbols on the upper left corner of each plot indicate statistically significant differences between patients and healthy subjects (* refers to the linear contrast of the effort·group interaction, and † refers to the average variable at 20–60% of MMP).

Figure 6

Mean and standard deviation of PH07, RP, R0 and ErrorPT corresponding to cross-prediction between GEN_EMG and DIA_EMG, shown as a function of %MMP. Gray solid lines represent healthy subjects and black dotted lines refer to OSAS patients. Symbols on the upper left corner of each plot indicate statistically significant differences between patients and healthy subjects (* refers to the linear contrast of the effort·group interaction, and † refers to the average variable at 20–60% of MMP).

Figure 7

Examples of delay embedding for a time series of length N = 20. Three different embeddings with ED = 3 are depicted: DT=1, DT=2 and DT=3. Some samples of the time series are gray-shaded to act as reference points when looking at the embedded matrix.

Figure 8

Graphical example of local linear prediction using ED = 2 and 2 neighbors. Five points (1 to 5) with known images in the future are shown, whereas image of point 6 after a certain prediction horizon is unknown. Using images of the 2 nearest neighbors (points 2 and 3), forecast for point 6 is obtained.