Abstract-We investigated the time-varying behavior of the autoregressive (AR) parameters in a myoelectric (ME) signal detected during a linear force increasing contraction. The AR pa rameters of interest were the reflection coefficients, the AR model spectrum, and the prediction errors. We used well-conditioned ME signals for which the complete time record of the motor units firings was available. In addition, the influence of the recruitment of a new motor unit, the conduction velocity of action potentials, and additive broad-band noise were investigated using simulated ME signals. The simulated ME signals were constructed from a selected group of the available motor unit action potential trains. The results revealed that, as the contraction progressed, the AR parameters displayed a time-varying behavior which coincided with the recruitment of newly recruited motor units whose spectrum of the waveform differed from that of the rest of the ME signal. This property of the AR parameters was obscured by the presence of broad-band noise and low-amplitude motor unit action potentials, both of which are more pronounced during low-level force contractions.
I. INlRODUCfION
IT HAS BECOME common practice to use the frequency spectrum of the surface myoelectric (ME) signal as a fatigue index for sustained muscle contractions (see reviews by De Luca [1] and Merletti et al. [2], among others). For such analysis, it is important that the ME signal be stationary. This is an important concern because motor units (MU's) may be recruited or derecruited during a contraction due to fluctuations in the force output of the muscle: Such fluctuation may occur even in attempted constant-force isometric contractions.Previous approaches for analyzing the time-varying aspects of the ME signals have used a linear prediction model. Among them, the autoregressive (AR) model has been used to deal with time-varying ME signals because it emphasizes spectral peaks for time records having a small number of samples [3]. This approach was introduced by Graupe and Cline [4] who attempted to use the surface ME signal for controlling prostheses. Subsequently, Sherif et al. [5] studied the behavior of autoregressive integrated moving average (ARIMA) coef ficients of the ME signals from the deltoid muscle during dynamic contractions. Recently, Capponi et al. [6] represented ME signals, detected from the biceps and triceps muscles, with the time courses of AR coefficients during rapid isometric contractions. The benefit of the AR model in ME signal analysis has been confirmed for applications in prosthesis control [4], [7], functional electrical stimulation [8], and clinical diagnosis [9]. These applications notwithstanding, the problems of applying AR model to time-varying ME signals and the time-varying behavior of AR parameters have not been studied in detail. The use of AR modeling for physiological interpretation of the behavior of the ME signal has been limited. In an early report, Inbar and Noujaim [9] described the influence of the statistics of M...