MMOOZZGG.com

Fields

AI in medical education: train the model and train people

1 min read3 sources

AI in medical education has two sides: how to teach a model medicine, and how to teach doctors to work with the model. Below are a Russian case with an exam for a language model and a Kenyan case where the result depended on training doctors.

On this page · 3 sections

Train the model

The Almazov Center, together with a bank, trained a large language model on the data of a cardiology institution. More than 300 doctors, postgraduates and students took part, and the model passed the final exam in general medicine. Products for patients, doctors and students are being built on it (2024). The exam gave a quality bar that could be checked before launch.

Train people

In Kenya, the effect of an AI consultant depended on the rollout: the share of visits with an unresolved red alert fell from 35–40% to 20% after training and mentoring (2025). In the US, 73.5% of doctors who were offered AI for documentation started to use it, and 73% of those surveyed said they did less paperwork after their shift.

How we train

For a clinic’s agents, the exam is the clinic’s real conversations, labeled by the administrator and the chief physician. We train staff on their own cases: half an hour a day of a responsible employee during the pilot, plus an intensive course on working with AI from the training program on the site.

Questions and answers

Do administrators need training to work with an AI agent?

Yes, about two hours at the start: how to read the report, how to take over hard conversations and how to correct answers.

Sources and fact-checking

Checked by the editors on 29 September 2026.

  1. Digital Platforms, case no. 2799, 2024: Almazov National Medical Research Center (platforms.su)
  2. OpenAI, 22 July 2025: Penda Health clinic network, Kenya (openai.com)
  3. JAMIA Open, 21 February 2025: University of Kansas Medical Center, USA (academic.oup.com)

Found a mistake? Tell us

Publications

Mike Mozg

Agentic AI use cases at AT&T: nearly 500 processes now run on agents

AT&T no longer uses AI only to help staff work faster: it is rebuilding whole business processes around agents. SiliconANGLE reported on 14 November 2026 that nearly 500 agentic AI use cases at AT&T are already in production. Here is how the company rebuilds them, where a human still checks the agent, and what other large companies can take from it.

News4 min

Cases on this topic: 1

  • Market case · Russia

    Almazov National Medical Research Center more than 300 doctors and students trained a language model, and the model passed the General Medicine exam

    In 2024 the center and a bank trained a large language model on the medical data of a cardiology institution, with more than 300 staff and students taking part. The result: the model passed the final exam in General Medicine, an LLM medical exam with a clear standard.

    Education

    2024 · source: platforms.su

    Read the case
All cases: 62
Contact us