AI-powered healthcare

With healthcare systems around the world burdened by staff shortages and tightening budgets, the sector is increasingly looking to Artificial Intelligence for support. Tietoevry Care has multiple solutions under development.

A study from Finland published in early 2023* predicted that implementing Artificial Intelligence (AI) for the country’s health- and social care would save EUR 770 million each year. By allowing AI to perform laborious and time-consuming administrative tasks, thousands of doctors and nurses would be able to focus more on patient care. Through rapid processing and analysis of vast datasets, AI can accelerate disease diagnosis. When patients receive timely and correct treatment, outcomes are improved and the societal cost is lower.

Tietoevry Care has been at the forefront of AI development in healthcare for several years. Our Lifecare data platform – with its advanced analytical capabilities – is at the core of this work. The platform enables healthcare professionals and researchers to access comprehensive patient information from a single source, eliminating the need for manual searches across multiple platforms.

The era of predictive care has arrived

One of our pioneering AI initiatives is a collaboration with Helsinki University Hospital (HUS). The joint team has been developing algorithms and data lake capabilities to expedite the diagnosis of three groups of rare diseases that are notoriously difficult to diagnose.

The potential of the research project has been bolstered by a legal change in Finland from January 2024. Medical professionals will now be able to proactively contact citizens whom AI algorithms have identified as having a disease risk.

"The results from the HUS research have been so promising that we are now starting with clinical validation. Our goal is to continue advancing our predictive models and productize the solution," says Niina Siipola, Head of AI and Data Solutions at Tietoevry. "We're also exploring opportunities outside of Finland, utilizing patient data from other systems. The possibilities ahead are very exciting."

Another key focus area for Tietoevry Care is around the use of Generative AI, where new data is created based on patterns learned from existing data. Generative AI can quickly read extensive medical records or literature to create an answer to a clinical query.

"Large Language Models took huge steps forward in 2023. But they are still not mature enough for clinical use cases, so we have started by using them with non-sensitive data,” says Siipola.

The case in question is at HUS Children's Hospital, where Tietoevry used Generative AI principles to process 1,300 documents into a treatment-guidelines search tool. Healthcare professionals can query the tool and be served up a summary of the information they need.

The solution leverages Tietoevry’s data lake – based on the Microsoft Azure cloud – while ChatGPT 4.0 Turbo summarizes the documents and provides direct links to the source material. The approach streamlines the information-retrieval process for caregivers, who previously needed to input precise keywords into a conventional search tool and then sort through the results.

Artificial Intelligence deciphers doctors’ notes

In another product development team, Tietoevry Care has been using Natural Language Processing (NLP) to retrieve information from patient records.

The work enables healthcare professionals to search for commonalities within a specific patient’s clinical notes. As these notes are written by multiple doctors – often over a period of many decades – there may be inconsistencies in the use of terminology. NLP resolves these.

"In the Finnish language, for example, there are many different ways to say that a patient has a history of smoking. We’ve trained the NLP model to find all these terms within a patient’s records," explains Siipola. "The language model we built for smoking was ready at the end of 2023. Now we’re creating models for alcohol consumption, narcotics usage and prescription medication."

The NLP project is also helping to provide new data sets for the Tietoevry data lake. By structuring clinical knowledge, AI is a useful tool in identifying trends and creating reports on specific health topics.

"We’re constantly looking for new AI use cases in healthcare, so we encourage our customers to come forward with ideas. I would love to see us developing even more prediction models, as these enable people to get the right care at the right time," concludes Siipola.

* A. Larsio. Powering the social health and social care system with data. Sitra, May 2023