The task of processing medical information is currently being solved in our country and abroad by means of heterogeneous medical information systems, mainly at the local and regional levels. The ever-increasing volume and complexity of the accumulated information, along with the need to ensure transparency and continuity in the processing of medical data (in particular, for bronchopulmonary diseases) in various organizations, requires the development of a new approach to integrating their heterogeneous sources. At the same time, an important requirement for solving the problem is the possibility of web-oriented implementation, which will make the corresponding applications available to a wide range of users without high requirements for their hardware and software capabilities. The paper considers an approach to the integration of heterogeneous sources of medical information, which is based on the principles of building microservice web architectures. Each data processing module can be used independently of other program modules, providing a universal entry point and the resulting data set in accordance with the accepted data schema. Sequential execution of processing steps implies the transfer of control to the corresponding program modules in the background according to the Cron principle. The schema declares two types of data schemas - local (from medical information systems) and global (for a single storage system), between which the corresponding display parameters are provided according to the principle of constructing XSLT tables. An important distinguishing feature of the proposed approach is the modernization of the medical information storage system, which consists in creating mirror copies of the main server with periodic replication of the relevant information. At the same time, the interaction between clients and data storage servers is carried out according to the type of content delivery systems with the creation of a connection session between end points based on the principle of the nearest distance between them, calculated using the haversine formula. The computational experiments carried out on test data on bronchopulmonary diseases showed the effectiveness of the proposed approach both for loading data and for obtaining them by individual users and software systems. Overall, the reactivity score of the corresponding web-based applications was improved by 40% on a stable connection.