Your digital assistant for care

From speech-to-text for patient notes to predictive screening for rare diseases, Tietoevry Care is using AI to reduce the burden on care workers and improve health outcomes.

Many countries face overburdened health and social care systems, with aging populations driving an increased need for care. Public healthcare budgets are not growing at the same rate, while the availability of skilled workers may also be constrained.

The sector thus has a high need for tools that can help to provide cost-effective and high-quality care. Machine learning and artificial intelligence are critical in enabling this.

“We envision AI being a health and social care assistant, with care professionals able to use the technology to surface valuable information, write up patient notes and even make connections between various data points. Automating administrative tasks in this way frees up more time to spend on caring for patients,” explains Niina Siipola, Head of AI and Data Solutions at Tietoevry Care.

Tietoevry’s Lifecare Data Platform enables the multi-source data collection and management needed to power AI-driven health and social care services. The platform is widely used in the Nordics.

A key part of the offering is a solution that removes any potentially identifying data before Large Language Models (LLMs) are used. This capability is not a given among software providers working with healthcare data.

“We’ve been working with AI in healthcare for some six years now, so we’ve learned a lot about the customer needs and the technology challenges. There are also security and privacy regulations to consider,” says Siipola.

“We are very careful about how and where we use AI, as we want to avoid privacy issues when using real world data for AI development and products. Tietoevry Care’s regulatory compliant approach to security and privacy is underpinned by our long experience in this field,” she says.

One of the regulatory developments guiding Tietoevry Care’s work in Finland is a legal change that allows medical professionals to proactively contact citizens. This is relevant when machine learning tools pick up on disease markers that humans may overlook. The authorities can now reach out to person who has a clinically significant finding for a specific disease, unless that person has opted out of being contacted.

Tietoevry Care and Helsinki University Hospital (HUS) have been working together in this domain, developing the machine learning algorithms and data lake capabilities needed to diagnose three groups of rare diseases.

“We had the first validation round earlier this year and we are now continuing to refine some of the processes. Our expectation is that the solution will be in larger clinical validation before the end of 2024,” says Siipola.

Tietoevry Care and HUS (featured in this CNN healthcare report) are also participants in a new two-year research project to assess the potential of LLMs for different healthcare use cases. The project, which recently received more than one million euros in funding, brings together several healthcare organizations and private companies.

An important use case being explored is how AI can speed up the documentation process in both healthcare and social care settings. Tietoevry Care is currently piloting a new solution that uses AI speech-to-text technology, enabling care workers to dictate notes after patient visits instead of manually typing them up. The solution also structures the data and corrects typos.

“In any given day, a public care worker may spend up to five hours typing patient notes. Our AI assistant aims to make this process more efficient, so that care professionals can focus more on patients rather than on documentation. We anticipate our solution halving the time that doctors and nurses currently spend on writing patient journals,” explains Siipola.

Powered by Microsoft Azure’s AI capabilities, the solution is currently being tested by 50 care professionals to ensure it meets a wide range of needs.

“Finnish is probably one of the most challenging languages for LLMs, especially in the clinical context. But we believe this pilot solution can easily be adapted for the Swedish and Norwegian markets, where the language models are an even better fit for the need,” says Siipola.

In addition to supporting care workers with compiling patient notes, Tietoevry Care sees LLMs playing an important role in classifying free text customer feedback.

“Healthcare providers may receive thousands of customer feedback forms a year. This feedback often comes in free text format, which takes a lot of time to read and analyze. Tietoevry Care has developed an AI assistant that automates this process,” explains Siipola.

In a pilot project in Finland, the assistant was used to categorize more than 4,900 pieces of anonymized customer feedback. The AI analyzed the tone and multiple other aspects of the responses. It has proven to be an effective tool in reducing manual work, while using customer and employee feedback to continuously improve care services.

“The data structuring capability of AI has far-reaching benefits for care,” says Siipola. “Structured data is the foundation for seamless integration across different care systems. When data can be shared across platforms with ease, it facilitates research, better coordination of care resources and long-term service improvements.”

Structuring data with the help of AI is not just a day-to-day operational improvement – it’s a step toward a more intelligent and interconnected care system that improves outcomes for everyone.