In this paper, the risk assessment of venous thrombus embolization under phlebothrombosis and factors that lead to pulmonary embolism (PE) has been considered. The statistical model has been constructed using the results of non-specific clinical trials. The Neumann’s method of confidence intervals is utilized for selection of statistically significant differences among four data samples of patients with different diagnoses. Revealed statistically significant differences are used in further studies as a priori information for the development of PE risk profiles. Further studies are carried out using discriminant analysis. A complex of factors is revealed that has a maximum discriminating efficiency in PE determination for different categories of patients. The reliability of the results is evaluated by solving the identified systems of classification linear equations. Parameters commonly used in medical statistics, such as sensitivity and specificity, being the analogues of the first and second kind of errors are also evaluated.