Article content
For the past 15 years, digitalisation in Estonian healthcare has meant combining healthcare data into online repositories and making it accessible to different providers. But just having more data at one’s fingertips will only benefit patients if it is being used properly, according to Madis Tiik, a family physician, entrepreneur, and longtime digital healthcare advocate in Estonia.
“The benefits won’t come through digitalisation or integration, but through the secondary use of the data,” he said. However, according to Tiik, the processing of harnessing electronic healthcare data is realising, as Estonia has begun introducing AI-based services.
From analysing health data to assessing drugs
Among these is Andmevaatur (Data Viewer), a tool TEHIK, the Estonian Health and Welfare Information Systems Centre, developed. Data Viewer makes healthcare data available to physicians in a format that can be used to get an overview of a patient’s state of health, compiling data related to analyses, diagnoses, examinations, surgeries, immunisations, and other risk factors. TEHIK introduced the tool two years ago and last month said that more than 500 healthcare providers already use it.
The Data Viewer tool is a “good example of how we can reuse data collected from different healthcare providers and documents,” said Tiik. “It gives the caregiver a very concrete picture of the dynamics of patient data.”
Tiik said it also demonstrates how AI can be paired with Estonia’s digital healthcare backbone to improve care and noted that several other such services have already been introduced or are on the way. Since 2018, for example, Tervisekassa, the Estonian Health Insurance Fund, has supported the Drug Interaction Assessment Database, which alerts physicians prescribing drugs about other drugs the patient might be taking, and if the two therapies might not be a good combination. The tool also recommends alternative therapies to a medication prescribed, Tiik noted.
In 2020, the Estonian Health Insurance Fund also rolled out its healthcare decision support system. “This offers reminders about missing treatments and medication or to measure blood sugar in a diabetes patient, for example, if the patient hasn’t had one in a year,” said Tiik. It also provides updated guidelines to inform clinical decision-making, all guided by an AI algorithm.
Last year, the Estonian Health Insurance Fund supported the innovation project Pre-visit, which Tiik described as a clinical pathway for patients. By interacting with the module and answering questions about one’s symptoms and health history, the tool can assess how urgent a patient’s problem is and direct them to the best solution. In addition, the tool relies on AI to ask follow-on questions, Tiik said and frees up time for healthcare providers while providing them with information ahead of a visit.
Another service, set to debut in July, will be a search engine called DynaMed that will enable them to pull up the most recent medical evidence, clinical findings, and guidelines to support the clinical decision-making process, all in a structured way. A fifth AI-based service will launch later in the year, relying on pharmacogenetic data from the Estonian Biobank at the University of Tartu to provide drug metabolism to clinicians to help them better prescribe treatments.
“It will be built on an automatic decision support algorithm that will help doctors when we prescribe drugs by adding pharmacogenetic data on potential drug reactions to the patient’s profile,” said Tiik. However, he noted that patients would first consent to add the data.
15 years of digitalisation pays off
For Tiik, the increasing number of AI-based services demonstrates that Estonia’s early focus on healthcare digitalisation is paying off.
“We have been collecting data for 15 years, and now we have enough data to feed these AI solutions.”
“We can reuse the collected information, building sophisticated, personalised services,” he added.
Tiik noted that Estonia had yet to build these AI-based services from scratch and instead has worked with international vendors to tweak their solutions for a local context with the help of TEHIK and other partners. In addition, because Estonian healthcare data is centralised, it makes integrating such services and rolling them out to users more accessible. He noted that some countries have also purchased licenses to the same AI-based products but have different standardised and integrated data levels. “This is the Estonian way of doing things,” Tiik said.
But Estonia was only sometimes a digital enthusiast, he noted. When Tiik was the CEO of the Estonian e-Health Foundation over a decade ago, he met resistance from decision-makers. “They said we didn’t have enough data, the data wasn’t of good enough quality, that they didn’t have the money to buy expensive licenses,” he said. Tiik credited the appointment of Rain Laane, a former regional director for Microsoft, to the head of the Estonian Health Insurance Fund in 2017 as setting the wheels in motion to deliver the AI-backed services available today.
“He has supported these things, but it takes time, with procurements and integration,” he said.
Prevention above all
Tiik’s main project at the moment is Pre-Visit. It relies on a symptom checker product sold by Infermedica, a company with offices in the US and Poland. The Estonian Health Insurance Fund does not yet allow all physicians in Estonia to use the service because of the license fees, about €300 per month per practice. While Pre-Visit does come with a price tag, Tiik argued that it improves productivity and that a pilot with the Tallinn Institute of Technology has showcased both its economic benefits as well as acceptance by patients, data that may encourage the Estonian Health Insurance Fund to make Pre-Visit more widely available.
Tiik is now involved in further developing the tool by linking it with a risk assessment to make preventative risk calculations as early as possible. He noted That all these tools aim to catch diseases early when they are easier to resolve.
“Today, we mainly fail with disease prevention because we find people with diseases too late,” said Tiik. “We can do it better,” he said. “To do that, we can have decision support to find good candidates for prevention and to develop the right prevention measures for the different patient groups,” he added. “This is what I am working on.”