Several EMG-based approaches to muscle fatigue assessment have recently been proposed in the literature. In this work, two multivariate fatigue indices developed by the authors: a generalized mapping index (GMI) and the first component of principal component analysis (PCA) were compared to three univariate indices: Dimitrov`s normalized spectral moments (NSM), Gonzalez-Izal`s waveletbased indices (WI), and Talebinejad`s fractal-based Hurst Exponent (HE). Nine healthy participants completed two repetitions of fatigue tests during isometric, cyclic and random fatiguing contractions of the biceps brachii. The fatigue assessments were evaluated in terms of a modified sensitivity to variability ratio yielding the following scores (mean ± std.dev.): PCA: (12.6 ± 5.6), GMI: (11.5 ± 5.4), NSM: (10.3 ± 5.4), WI: (8.9 ± 4.6), HE: (8.0 ± 3.3). It was shown that PCA statistically outperformed WI and HE (p < 0.01) and that GMI outperformed HE (p < 0.02). There was no statistical difference among NSM, WI and HE (p > 0.2). It was found that taking the natural logarithm of NSM and WI, although reducing the parameters` sensitivity to fatigue, increased SVR scores by reducing variability.
© Copyright 2013 Journal of Electromyography and Kinesiology. Elsevier. Alle Rechte vorbehalten.