R.V. Meshcheryakov, E.F. Schipunov, O.Ya. Vasiltseva

Investigation of medical parameters and to determine their relationship to disease is one of the urgent problems in medical informatics. One such area is the heart surgery. Medical parameters are selected in accordance with the opinion of experts in the field of cardiac surgery, having a serious theoretical foundation and rich practical experience in the subject area. These indicators include: group of laboratory values, the results of instrumental methods of treatment, as well as some information about the hospitalization of the patient (eg, number of days spent in hospital). To assess the relationship of these parameters to select the best criterion. Because of the parameters are presented in different scales, their processing must be carried out by various methods. To determine the dependencies you need to classify options, combining them into different groups - clusters. After classifying the data can be applied to clusters evaluation criteria interdependence parameters within and between clusters. In all cases it is important to take into account the variability and specificity of each estimated parameter. For example, if in death, diagnosis of \"pulmonary embolism\" can be verified as the main and collateral. Approximately 50% of cases of pulmonary embolism is not diagnosed at all, even after a post-mortem examination of the patient. To perform exploratory analysis we used a sample of 14 laboratory tests 41 patients diagnosed with pulmonary embolism as the subject of death, and without it. The total number of parameters was 722 value. The proposed approach in the article seeks to address uncertainty in the subsequent analysis of cardiovascular parameters. Further research is aimed at the classification of the parameters of the cardiovascular system of person, called to confirm these assumptions.