Analysis of vocal disorders in a feature space.

This paper provides a way to classify vocal disorders for clinical applications. This goal is achieved by means of geometric signal separation in a feature space. Typical quantities from chaos theory (like entropy, correlation dimension and first lyapunov exponent) and some conventional ones (like autocorrelation and spectral factor) are analysed and evaluated, in order to provide entries for the feature vectors. A way of quantifying the amount of disorder is proposed by means of a healthy index that measures the distance of a voice sample from the centre of mass of both healthy and sick clusters in the feature space. A successful application of the geometrical signal separation is reported, concerning distinction between normal and disordered phonation.

Denoising human speech signals using chaoslike features.

A local projective noise reduction scheme, originally developed for low-dimensional stationary deterministic chaotic signals, is successfully applied to human speech. This is possible by exploiting properties of the speech signal which resemble structure exhibited by deterministic dynamical systems. In high-dimensional embedding spaces, the strong inherent nonstationarity is resolved as a sequence of many different dynamical regimes of moderate complexity.

Local estimates for entropy densities in coupled map lattices

We present a method to derive an upper bound for the entropy density of coupled map lattices with local interactions from local observations. To do this, we use an embedding technique that is a combination of time delay and spatial embedding. This embedding allows us to identify the local character of the equations of motion. Based on this method we present an approximate estimate of the entropy density by the correlation integral.

Convective lyapunov exponents and propagation of correlations

We conjecture that in one-dimensional spatially extended systems the propagation velocity of correlations coincides with a zero of the convective Lyapunov spectrum. This conjecture is successfully tested in three different contexts: (i) a Hamiltonian system (a Fermi-Pasta-Ulam chain of oscillators); (ii) a general model for spatiotemporal chaos (the complex Ginzburg-Landau equation); (iii) experimental data taken from a CO2 laser with delayed feedback. In the last case, the convective Lyapunov exponent is determined directly from the experimental data.

Coping with nonstationarity by overembedding

We discuss how nonstationarity in observed time series data due to pronounced fluctuations of system parameters can be resolved by making use of embedding techniques for scalar data. If a D-dimensional deterministic system is driven by P slowly time dependent parameters, a (D+P)-dimensional manifold has to be reconstructed from the scalar time series, which is done by an m>2(D+P)-dimensional time delay embedding. We show that in this space essential aspects of determinism are restored. We demonstrate the validity of the idea heuristically, for numerical examples and for human speech data.

Estimating the Lyapunov spectrum of time delay feedback systems from scalar time series.

On the basis of a recently developed method for modeling time delay systems, we propose a procedure to estimate the spectrum of Lyapunov exponents from a scalar time series. It turns out that the spectrum is approximated very well and allows for good estimates of the Lyapunov dimension even if the sampling rate of the time series is so low that the infinite dimensional tangent space is spanned quite sparsely.

Improved false nearest neighbor method to detect determinism in time series data.

The false nearest neighbor method introduced by Kennel et al. [Phys. Rev. A 45, 3403 (1992)] is revisited and modified in order to ensure a correct distinction between low-dimensional chaotic data and noise. Still, correlated noise processes can yield vanishing percentages of false nearest neighbors for rather low embedding dimensions and can be mistaken for deterministic signals. Therefore, the false nearest neighbors method has always to be combined with a surrogate data test.

The short DNA sequences in the cytoplasm of Ehrlich ascites tumor cells are tightly associated with proteins.

Mouse Ehrlich tumor cells harbor extrachromosomal DNA elements in a non-mitochondrial fraction of the cytoplasm (Abken et al., Proc. Natl. Acad. Sci USA 90: 6518-6522, 1993). The cytoplasmic DNA sequences constitute a distinct group of extrachromosomal genetic elements with common properties: (i) the DNA molecules are 50-500 bp in length and of linear configuration, (ii) most of the DNA sequences exhibit the potential to modulate the activity of a transcriptional promoter, and (iii) the DNA elements are preferentially found in tumor cells, not in cells with normal phenotype. Unexpectedly, the extrachromosomal DNA sequences are found to be tightly associated with at least three proteins (52 kD, 62 kD, 64 kD) forming DNA-protein (DNP) complexes. The DNA-protein interaction is stable during extraction with phenol and chloroform and during incubation with proteinase K and pronase P, in the presence of detergents (SDS, NP40), guanidine-HCl, high salt concentrations, and alkali. Hydrolysis of the DNA by DNAse I makes the proteins of the DNP complex accessible to proteolytic degradation. Western blot analyses imply that the proteins associated with the cytoplasmic DNA sequences are antigenically related to nucleomatrix proteins that are covalently bound to certain regions of chromosomal DNA. Covalent bonds between the cytoplasmic DNA and the polypeptides would explain the unexpected high stability of the cytoplasmic DNA-protein complexes in tumor cells.

Short DNA sequences from the cytoplasm of mouse tumor cells induce immortalization of human lymphocytes in vitro.

Cytoplasts of mouse L929 and Ehrlich ascites tumor cells harbor DNA sequences that induce unlimited proliferation ("immortalization") of human lymphocytes after transfection in vitro. By equilibrium centrifugation of cytoplasmic lysates in a neutral CsCl gradient, the immortalizing activity was recovered together with extramitochondrial fractions at high salt densities (1.85-1.87 g/cm3). Unexpectedly, these fractions contain linear DNA molecules of 50-500 bp in length. In contrast, cytoplasts of primary, senescent cells (mouse embryo fibroblasts, human lymphocytes) do not harbor DNA in the corresponding fractions. Cytoplasmic DNA isolated from high-density fractions of mouse tumor cells was cloned in subset libraries, and of 45 DNA sequences we identified 2 clones--one from L929 cytoplasts (203 bp) and another one from the cytoplasm of Ehrlich ascites cells (372 bp)--that induce unlimited proliferation of human lymphocytes in vitro. Immortalized lymphoid cells harbor 1-5 copies of transfected DNA integrated into chromosomal DNA, whereas about 100 copies were found as episomal DNA in the cytoplasmic fraction. No immortalization could be induced by transfection of nuclear DNA randomly fragmented to 200-500 bp. Although the cloned DNA sedimented at 1.70 g/cm3, after transient transfection into lymphocytes, these DNA sequences form salt-stable complexes that sediment in fractions at the same high density (1.82-1.88 g/cm3) from which they were originally cloned. The high-density banding of these cytoplasmic DNA sequences may be due to association with RNA and/or with (metallo-) proteins in vivo. Since both cloned DNA sequences with immortalizing activity have stop codons for protein translation in all possible reading frames, immortalization may be induced by insertional inactivation or functional suppression of genes that are needed to be expressed during cellular senescence or programmed cell death.