Slight parameter changes detection in biological models: a multiresolution approach

This paper deals with the automatic detection of slight parameter changes from the analysis of signals generated by nonlinear biological systems. The interest is focused here on investigating particular aspects of the relation between signal analysis and systems dynamics, involving the automatic detection of changes in parameters of nonlinear systems. The continuous multiresolution entropies combine advantages stemming from both entropy (Shannon and parametric q-entropy) and wavelet analysis, have been shown that they are sensitive to dynamical complexity changes. Classical statistical approaches offer a new tool that enables the automatic detection of such changes. In this paper, multiresolution and standard tools for the automatic detection of slight parameter changes in nonlinear dynamical systems from the analysis of the corresponding time series are introduced and compared. The relevance of the multiresolution approach, together with its robustness in the presence of moderate noise, and their comparison are discussed in numerical simulations and they are applied to biological signals.