At the NTI Competence Center based on “Technologies for Storing and Analyzing Big Data” of the Moscow State University has been developed a software project that makes it possible to carry out the automated analysis of the data of X-ray images, mammograms, and cardiograms based on artificial intelligence with an accuracy of 93%.
Special methods for processing large data arrays generated during the medical images analysis allow the attending physician to turn to the cloud solution, to develop a personalized treatment plan for a particular patient.
The software is based on an algorithm for automated diagnosis of chest diseases (e.g., tuberculosis, oncological and other diseases) using neural networks, which has been machine-trained on more than 270,000 x-ray images (fluorograms) marked by medical specialists of the Ministries of Healthcare of the Republic of Tatarstan and the Central Federal District. Medical data of patients are stored on the database of the data center of an industrial partner, which provides high standards of data security. The diagnostic accuracy of the algorithm in terms of the analysis of chest radiographs was evaluated by the specialists of the Center for Diagnostics and Telemedicine of the Moscow Department of Healthcare.
A distinctive feature of the project is the telemedicine function and the presence of a cloud data center where the corresponding software is located. Any X-ray departments of stationary medical institutions of any levels and mobile telemedicine complexes (MTK) designed for mass screening of the population in rural areas, remote and hard-to-reach areas can be connected to the developed software.
Thanks to this, medical diagnostic images are analyzed almost instantly, in real-time, and the likelihood of errors in data processing is also reduced.
To verify and receive feedback from radiologists, the project was organized to access the project through an AI-based cloud service for fluorography rooms of clinics in 59 regions of Russia. As a result of the first stages of testing, the diagnostic accuracy of 93% was achieved (the best level in Russia among similar systems). In the future, it is planned to increase the accuracy of diagnostics up to 98%.
Another result of patient data's simultaneous processing is the answers to the requests of healthcare organizers at all levels (federal, regional, municipal) with recommendations for optimizing the functioning of the healthcare system, as well as for controlling epidemics of infectious and noncommunicable diseases.
The project of the NTI Competence Center for Big Data Technologies in Healthcare is being implemented jointly with the industrial partner of the Scientific Research Institute of Vector Radio Company, JSC. Potential customers of the project are public and private medical institutions at all levels, which have fluorographic devices used to conduct a medical examination of the population and one-time examinations of patients, research medical institutions and the Ministry of Health of the Russian Federation, which creates favorable prospects for the commercialization of the project.