Разработчиков медицинского ИИ обяжут автоматически передавать данные о сбоях в Росздравнадзор

According to experts, the idea is similar to the current drug safety control system

Developers of medical technologies based on artificial intelligence will be required to automatically transmit data on the frequency of use of their solutions, failures, and deviations in operation to Roszdravnadzor. According to the agency, this concerns decision support systems for doctors — from analyzing medical images to processing electronic patient records.

Currently, 48 such medical devices are registered in Russia, 43 of which are of domestic design. Market leaders include Botkin.AI, Webiomed, and "Третье мнение". Currently, problem reports are sent to Roszdravnadzor manually and once a year, but the new initiative involves continuous collection of technical telemetry from devices.

According to experts, the idea is similar to the current pharmacovigilance, but with an adjustment for the specifics of the technologies: the system will have to independently record errors and transmit information to the regulator. This does not concern personal patient data, but information about failures and software configurations is considered sensitive: it can reveal weaknesses in the product and be of interest to competitors or attackers.

To reduce risks, it is proposed to transmit data through secure government channels (СМЭВ) using cryptographic encryption and strict access audits.

The initiative is currently under interdepartmental discussion. If the project is approved, Roszdravnadzor will have to create a new platform for automated supervision of medical AI solutions.

Read also on the topic:

The government plans to support AI for medicine with rubles

AI is being integrated into medical services in Russia

Artificial intelligence is being introduced into Russian medicine at an incredible speed — Murashko

Now on home