The aim of the article is to study Data mining methods to predict the cardiovascular disease risks' level, based on data from medical information systems. We propose a new approach to the development of information and analytical support for biomedical research. Based on the proposed approach, we have developed the special methods, technologies and services to extract and anonymize valid problem-oriented information from unstructured electronic medical records. Created data store contains more than 70,000 records of electronic medical records and provides the researcher anonymized "smart" information in accordance with the possible scenario of its use. The developed services for visualizing the values of objective indicators allow us to determine the optimal data structure for the diagnosis of a specific group of diseases. This is shown by the example of risk testers of cardiovascular diseases. Based on the selected indicators, models for personified prediction of the cardiovascular disease risks' level were developed and tested)/