Neuromed AI Neural Network for Physicians

The main articles are:

• Speech Recognition (Technology, Market)
• Speech technology: On the path from recognition to understanding
• Chatbots in medicine
• Medical Decision Support Systems (CDS)

2025: Grading Exam for Residents

Chat boat Neuromed AI on the basis, generative AI developed by scientists Sechenov University and specialists of the technological startup "," Neurohoney passed the state final certification, completing the process of training residents in the areas of cardiology and. oncology

This result is a meaningful step not only for our university, but also for everything medical sciences in. Russia Artificial intelligence It goes to a new level, demonstrating the ability to solve complex clinical problems, analyze data and support doctors in decision-making. We see in this tool a powerful assistant who will allow doctors to delegate neuronets routine tasks and improve the quality of diagnosis and treatment, "said Philip Kopylov, director of the Institute of Personalized Cardiology.

The exams that Neuromed AI took are mandatory for cardiologists and oncologists at the end of their residency. They include testing, solving situational problems and testing practical skills. Of course, the neural network is not yet able to examine the patient, but she coped with the rest of the stages of the exam at a level sufficient to assign qualifications to resident doctors in the field of cardiology and oncology. Neuromed AI was able to analyze patient data, answer questions about diagnosis and treatment tactics, interpret the results of analyzes, instrumental studies and previous treatment.

When we began cooperation with Sechenov University, our neural network did not yet have sufficient accuracy in the field of cardiology and oncology. However, thanks to joint work on data validation and systematic further training, she was able to reach a new level. Now our neural network has demonstrated that it is not inferior to qualified resident doctors. This confirms the quality of the work carried out by our team together with the university experts, - said Alexander Sotnik, General Director of Neuromede.

The neural network is trained on a datacet with a volume of over 50 gigabytes, which includes both Russian and international clinical recommendations, scientific articles, reference books and open data sets. Previously, she was already able to prepare standard medical documents, provide access to professional information about drugs, help assess the risks associated with the patient's health and make recommendations for diagnosis and treatment based on current clinical guidelines. Now the "qualification" of the AI assistant allows it to be used in even more complex clinical scenarios.

The chat bot Neuromed AI is included in the register of Russian software, in beta testing it is already used by more than 500 oncologists and cardiologists in different regions of Russia. A number of pilot projects are also being implemented to centrally test the neural network as an assistant to cardiologists and oncologists on the basis of the Clinical Center of Sechenov University, in the First City Hospital, Lipetsk Regional Oncological Dispensary and the Republican Specialized Scientific and Medical Center of Cardiology of the Ministry of Health of the Republic of Uzbekistan. The developers continue to improve the system, taking into account the feedback of doctors. The plans include the release of a number of applied tools for doctors based on the Neuromed AI neural network and their integration into the Unified State Health System of Russia, as well as the implementation of projects in other countries.