The scale of artificial intelligence implementation in Russian healthcare has reached a fundamentally new level. The MosMedAI platform, initially launched as a Moscow experiment, is now operating on a federal scale. More than two thousand clinics from 75 regions of the country have connected to the system. The total volume of processed data has already exceeded 16 million radiological studies.
The project, launched in 2020 at the height of the pandemic, was originally developed as a tool for rapid COVID-19 diagnosis. The joint work of the Moscow government and specialists from the Center for Diagnostics and Telemedicine made it possible to create a robust platform. Free access for regional clinics became possible in 2024 following a corresponding directive from the president.
Experts tested about 70 different AI services, but only the best passed the strict selection process. As of today, doctors have access to 18 proven algorithms. These digital assistants do not just analyze images—they help identify hidden threats that may not be obvious to the human eye.
What can neural networks do?
The range of tasks is impressive. The algorithms work with mammography, detecting changes in breast tissue; analyze fluorography and X-rays, finding lung pathologies and signs of scoliosis; process CT data to search for focal formations and organ lesions.
The system not only highlights suspicious areas but also provides the radiologist with detailed additional information. In Moscow clinics, the neural network can recognize up to 14 different diseases in a single image. Recently, the list was expanded to include scoliosis diagnosis via spinal X-rays.
My analysis: The implementation of AI in diagnostics is not just a trend but an objective necessity. Processing 16 million images is a colossal amount of data that a human is physically incapable of analyzing with the same speed and accuracy. However, it is important to remember: AI is a tool for the doctor, not a replacement. The key challenge now is not so much technological as organizational: how to integrate these algorithms into daily clinical practice without increasing the bureaucratic burden on medical professionals.