On the Use of the Statistical Methods for Biomedical Signals and Data Processing

This article presents a review of the use of modern algorithms for statistical biomedical signals and data processing. We considered recent studies, in which real practical problems of analyzing a big quantitative data were solved. In the structure of the article, the most frequently and widely used statistical methods are considered in stages. At first the use of a posteriori clustering is shown to reveal efficiency of the developed algorithm of noise-resistant speech signal processing. We present the potential of using machine learning based on the neural network clustering algorithm. This algorithm uses ANN to classify biomedical data. Results of statistical analysis of the cardiovascular system activity by modalities (PCG, pulse wave curve etc.) are presented as a promising field of modern biomedicine. Also, we consider the degree of distortion of the obtained filtered PCG signal in a study made by applying a new filtering algorithm. To prove its efficiency, a scatter diagram (scattergram) is shown. To study dynamics of the variability of a pulse wave curve, the use of heart rate variability on the basis of statistics was considered. The use of a quantile scattergram of regression analysis results is also considered there as a way to analyze the substances which are toxic for growth of sugar beet calluses (Betavulgaris l.). At last, this article focuses on the issue of statistical significance, and reliability of obtained results. The use of a quantile scatterogram of regression analysis results with normal data distributions was considered.