One in three Nordic healthcare AI projects is already operational

More than 800 artificial intelligence initiatives are now documented across publicly funded healthcare in Sweden, Denmark, Finland, Norway and Iceland, and roughly one in three has already progressed into routine operation. A new Nordic mapping provides an unusually broad view of where healthcare AI has moved beyond experimentation and where health systems are still struggling to turn pilots into everyday practice.

The Nordic AI Health Map brings together initiatives across eight healthcare use cases and distinguishes between projects at concept, pilot and routine-operation stages. Data were collected during spring 2026, with the first version of the Nordic map released this summer and intended to be continuously expanded. The initiative is supported by health authorities, regions and national AI and digital-health organisations across all five Nordic countries.

For healthcare leaders outside the region, the most interesting number is not simply that more than 800 initiatives exist. It is that approximately two-thirds have apparently not yet reached routine operation. Even in digitally mature health systems with strong public infrastructure and relatively advanced health data environments, moving AI from development into normal care remains the central challenge.

From AI inventory to implementation evidence

Healthcare AI is often described through individual success stories. One hospital introduces an imaging algorithm, another tests generative AI for documentation and a third develops a predictive model for deterioration. That makes it difficult for policymakers and healthcare organisations to understand what is happening across an entire health system and, more importantly, which applications have survived the transition from pilot to practice.

The Nordic map attempts to provide that wider view. Initiatives are categorised according to their use case and maturity, allowing users to distinguish technologies still being explored from those actually embedded in everyday healthcare. Applications range from administrative support and clinical decision-making to disease prevention and other parts of care delivery.

The project is backed by Danish Regions and CAI-X in Denmark; Finland's national SOTE ecosystem and UNA; Stafræn Heilsa in Iceland; the Norwegian Directorate of Health and Norway's four regional health authorities; and AI Sweden together with the Swedish eHealth Agency. Funding comes through the Nordic Council of Ministers' eHealth Working Group.

That institutional involvement matters. Mapping AI systematically can help health systems identify duplication, compare experiences and determine whether a solution already operating elsewhere could be reused rather than developed again from scratch. It can also reveal where investment is concentrated and where promising projects repeatedly fail to progress beyond experimentation.

One-third in routine operation

According to the mapping, roughly one-third of the more than 800 identified initiatives have reached routine operation. That indicates AI is no longer confined to innovation programmes in Nordic healthcare, but it also highlights the scale of the implementation gap that remains.

The figure should be interpreted carefully. The map is not a clinical effectiveness study and does not establish that every operational AI application improves outcomes, reduces costs or saves clinicians time. Nor does the number of initiatives provide a direct comparison of AI maturity between individual Nordic countries. What it does provide is a much clearer picture of where technologies sit in the implementation lifecycle.

That distinction is increasingly important. Healthcare organisations have spent years demonstrating that AI can perform specific tasks under controlled conditions. The harder questions now concern integration into electronic health records, procurement, clinical responsibility, data access, reimbursement, workforce acceptance, monitoring and the ability to maintain algorithms after implementation.

An operational system also creates a different set of evidence requirements from a pilot. Once AI becomes part of routine healthcare, organisations need to know whether its performance remains reliable over time, whether professionals actually use it as intended and whether it delivers measurable value under real-world conditions. Moving into production is therefore not the end of implementation but the beginning of a longer governance process.

The Nordics could become a laboratory for scaling AI

The five Nordic countries have characteristics that make their experience particularly relevant internationally. Their healthcare systems are predominantly publicly financed, digital infrastructure is relatively mature and national or regional organisations can coordinate initiatives across large populations. At the same time, each country has its own governance structures, technology environments and implementation challenges.

That creates an opportunity to learn across borders. If comparable healthcare organisations can see where AI is already operating elsewhere in the Nordic region, they may be able to adopt existing approaches instead of launching another isolated pilot. The mapping initiative explicitly aims to support that exchange of knowledge and experience.

The timing is also relevant for Europe. Implementation of the AI Act and European Health Data Space will increasingly shape how healthcare organisations access data, evaluate AI and govern its use. Nordic organisations involved in the map argue that closer regional collaboration could help demonstrate how responsible AI implementation can work as those European frameworks move into practice.

The real value of the Nordic AI Health Map will therefore depend on what happens next. Counting projects is useful, but healthcare leaders ultimately need evidence about which applications deliver clinical, operational or financial value, why some reach routine use while others stall, and whether successful implementations can be transferred between organisations and countries.

More than 800 initiatives provide enough experience to start answering those questions. The next phase should be less about adding another dot to the map and more about identifying what separates the third that reached routine care from the majority that have not.