A world-first nationwide AI model will boost decision-making in healthcare

Aging populations and rising demand for services are putting a strain on healthcare systems around the world. Artificial intelligence is being explored to address these issues, but there still aren’t any comprehensive solutions that bring together both health data and information systems at the levels of patients, operations and society. 

Now, researchers from ELLIS Institute Finland and three Finnish universities will address these challenges in the FINe-Health Foundry, which will leverage Finland’s unique national health databases for new types of AI-assisted care and health systems operations and planning solutions. The project will build the world’s first nationwide healthcare foundation model, a ‘Swiss Army knife’ AI model that can support doctors making clinical decisions in real-time and create scenarios at the population-level based on different interventions. The FINe-Health Foundry has received nearly five million euros from Business Finland for an initial three-year period.

Finland is uniquely positioned to advance AI for healthcare: there is high-quality nationwide health registry data going back decades, legislation that covers privacy and responsible data use, and secure computational capacity with the LUMI AI Factory supercomputer. There is also strong AI expertise through main partners ELLIS Institute Finland, which unites the top machine learning researchers across all Finnish universities, and FIMM – Institute for Molecular Medicine Finland, University of Helsinki.

Multimodal health data in Finland includes everything from medical images to electronic health records and genomic data from a large part of the Finnish population. With this input, the new foundation model will be able to predict disease risk for over 200 conditions for every living person in Finland, says professor Andrea Ganna of FIMM and the University of Helsinki. Decision-making support will come from AI agents running within virtual laboratories, which are platforms that combine human expertise with simulations.

FINe-Health Foundry is one of the first big actions for the ELLIS Institute’s strategic focus area of health. “Our research in the Foundry will move machine learning from using correlations to cause-and-effect reasoning, while integrating expert knowledge from healthcare practitioners. The resulting foundation model will accelerate and improve healthcare decision-making,” says professor Samuel Kaski, director of ELLIS Institute Finland.

Global demand for governed, real‑time solutions for healthcare is massive. The Foundry’s foundation model aims to provide not only better, targeted care, but also the elements for healthtech products and support for evidence-based health policy and innovation. Holistic integration of AI into the Finnish public healthcare system could produce savings of over one billion euros a year, according to estimates. “Our goal as a society needs to be to both improve healthcare and make it more efficient,” says Kaski, “and that requires contributions from the whole ecosystem. Our contribution with the Foundry is new decision-making principles and methods. Together with an AI application layer and virtual laboratory platform, these can be leveraged in the healthtech domain and beyond.”