Tomashevskyi R. Improving the reliability of monitoring and diagnostics of biological objects under the conditions of nonstationarity of measuring signals

The thesis is devoted to the solution of an important scientific and applied problem of creating and improving methodological and information technology support for the conditions of a biomedical measurement experiment (BMME), which made it possible to increase the reliability of monitoring and diagnosing the state of a biological object with limitations on the amount of measurement information and statistical heterogeneity of non-stationary biophysical signals. The paper proposes a fundamental basis for solving this problem based on the methods of factor analysis taking into account the probabilistic properties of biological objects, the dynamic characteristics of biophysical signals and the actual conditions of a BMME, such as limited research time, real-time work and the influence of external factors that were not included into the experiment plan. Mathematical models were developed for calculating the expected amount of diagnostic information, taking into account the type of plan of a biomedical measuring experiment and the restriction on the amount of measurements, which allowed a quantitative assessment of the contribution of each of the parameters of the experiment plan to the total amount of information received. The information theory of diagnostic and diagnostic solutions for planning a biomedical measurement experiment was further developed. For the first time, mathematical models have been developed for calculating the expected number of diagnostic and control information, taking into account the type of plan of the biomedical measurement experiment and the restriction on the amount of measurements. A method for the covariance analysis of two-dimensional medical signals with multiple measurements, correlation-spectral methods for detecting violations of stationarity of biophysical signals under dynamically continuous factorial loads, a method and technical means for monitoring dynamically non-stationary bioelectric signals have been developed. The work proves a stochastic relationship between the levels of quantization of induced signals in a biological object and the functional states of this object by using the developed covariance and correlation spectral methods of data analysis, which allowed for the classification and express control of biological objects and states. Based on this, a hardware information system has been developed that allows for the rapid diagnosis of liquid biological samples, including oncological diseases (colorectal cancer). The results of experimental studies of the system showed its compliance with medical and technical requirements and the possibility of obtaining diagnostic information for normalized mechanical effects on biological samples.