Models and methods statistical data processing for analyzing the elements of dynamic systems

UDC 681.518.2:519.816-616.65-006

Models and methods statistical data processing for
analyzing the elements of dynamic systems
КЛ. Goryachaya, postgraduate, dep. 304
National aerospace university named after N. Ye. Zhukovsky “KhAI"

The development of defects in engineering or biomedical systems is a complex dynamic process. Subject area experts can not always predict how quickly they will be developed. It is not always possible to reach a consensus at what stage of development are located the defects and, as a consequence of which methods of eliminating defects needs to apply. Monitoring and forecasting dynamic process of the system work, help to experts and/or patients (if biomedical systems are considered) to make decisions that lead to better values criteria quality of the work system or the survival and quality of life for patients.

As an example of a complex system, we consider medico-biological system, which consists of the following elements: doctor, patients and subsystem diagnosis of patients. The beginning and the end stages of the life cycle of the treatment patients we defined as the set of final states of the patients. The number of states have taken to address is set by the expert in the subject area based on the results of the cluster analysis. It has accepted as a working hypothesis, the hypothesis of local equilibrium, according to which the patient is uniquely determined, with the fundamental system his state variables.
We assume that the critical on the set of final states is a resistant state in which, due to the progressive development of defects in the functional parts, the patient in the treatment process becomes unmanageable. The objective problem is that there is no structured allowing rule transition for the patient in question in a resistant state.
The purpose of this study is to analyze the statistical relationship between the variables that determine the state of the elements biomedical systems (patients); Identification affiliations patient to a particular class based on its conditions sensed variables. During the research process were solving the following scientific objectives: the construction of the diagnostic model; reduction of dimension vector input data [1]; aggregation of state variables into complexes, which pairwise would have been poorly correlated variables (orthogonalization); classification state of patients.
The method for solving the classification condition of elements complex systems has been improved. The system model of the diagnosing medical process and biological elements of the system (patients) has been proposed. Based on the information content analysis and aggregating state variables made the reduction of the dimension variables of the status patients. Computer decision support system for biomedical diagnose condition of patients has been developed.