| Singular value decomposition is a well-known statistical tool that allows the factorization of rectangular matrixes. Here, we will use it to estimate the complexity of short, irregular neurological time series by estimating the distribution of the strength of its othogonal osciallatory modes. We will show how the proposed index closely follows the estimated largest Lyapunov Exponent of model systems, and how it can be used to assess differences in complexity as a function of the neurological state (within-group) or the group of subjects analyzed (between-group differences) even from short, possibly non-linear, non-stationary neurophysiological data. |
|