At the beginning of the year, the Eesti.ai initiative was launched and one of its goals is to make the healthcare sector more efficient. For example, family physicians currently spend a great deal of time filling out paperwork, but in the future, artificial intelligence could write summaries and discharge reports based on a patient's visit.

"There is a device that first helps recognize speech and then transcribes the entire conversation, and once all the interfaces are further developed, it could essentially go directly into the digital health record," said Eesti.ai team lead Kirke Maar.

Family physicians are cautious about the introduction of artificial intelligence.

"Because we work with a great deal of sensitive personal data, the biggest risk is still a data leak. In addition, I do not really have a clear understanding of where all of that information is or to what extent and in what way it will be accessible and connected in the future to the person about whom I am making an inquiry. There is also the question of responsibility — who is responsible?" said Triinu-Mari Ots, a board member of the Family Physicians Association.

The authorities also plan to make ambulance work easier using a similar approach where instead of filling out paperwork, artificial intelligence would listen to the patient and medical staff and draw conclusions and summaries from what they say. Questions remain unresolved there as well.

"From our perspective, the problem is that artificial intelligence often cannot clearly distinguish who is speaking at a given moment. It may also have difficulty determining whether the discussion concerns medical matters or something else. That is certainly one of the nuances that will have to be worked out during the course of this project," said Karl-Henrik Peterson, a member of the Health Insurance Fund's management board.

Kirke Maar said that in addition to the issues already mentioned, artificial intelligence models also struggle with medical terminology and the Estonian language. She repeatedly emphasized that setbacks are to be expected during testing.

"The health sector is extremely sensitive when it comes to where we store our data and how we store it. It is also emotionally very complicated — whether people are ready for the fact that, first, we are recording them. Not everyone is ready for that, nor do they have to be. Second, there is the issue of where we store the data. We certainly cannot send that data to some cloud server in the United States. We need to think carefully about which models we use and where we store the data, and we are working on all of that in parallel," Maar said.

The first trials involving artificial intelligence in health care are already underway. Projects in various areas of the sector require approximately €2 million in funding, which the government will discuss during April. The Estonia.ai initiative's other main areas are education and security.

--

Follow ERR News on Facebook and X and never miss an update!