A survey of cardiologists in Denmark shows that they are ready to incorporate artificial intelligence (AI) into their clinical work. This applies to both younger and older doctors, who are generally optimistic about the opportunities AI offers. Now we just need to show that it will also positively affect patients, says a researcher.
AI is advancing everywhere – including in hospital corridors.
For cardiologists, this means new tools that can quickly review electrocardiograms and X-rays and point out signs that might otherwise be difficult to detect.
And the stakes are high: faster and more accurate assessments can ultimately mean fewer errors, earlier treatment – and potentially lives saved.
Technologies that once seemed far in the future are now being tested directly in clinical practice. If they work well, they could change the way cardiologists work on a daily basis.
Such an AI-based algorithm is currently being tested in all departments of cardiology in the Capital Region of Denmark and Region Zealand, and if the results are positive, it could pave the way for Danish cardiologists to soon welcome a new digital colleague.
It is therefore natural to ask cardiologists what they think about what will soon become part of their working lives – and whether they believe that the technology can actually improve the treatment patients receive.
This is exactly what the researchers did.
“On the whole, doctors are optimistic and positive about using AI in their work. The real battle over AI will come when we get the study results that show it also benefits patients,” says Søren Albertsen Rand, doctor and PhD student from the Department of Cardiology at Rigshospitalet in Copenhagen.
The research has been published in the Danish Medical Journal.
How the AI technology doctors will use actually works
One researcher developing the AI algorithms currently being tested in the Danish healthcare system is Søren Brunak, Professor at the Novo Nordisk Foundation Center for Protein Research at the University of Copenhagen.
The algorithms work by extracting information from patient records, combining it with new tests and measurements and then calculating the probability of various outcomes for the patient.
The model is a machine-learning algorithm trained on thousands of earlier patient cases. It recognises patterns that people often overlook, such as combinations of blood tests, symptoms and medical history that together may indicate an increased risk.
For instance, it can realistically estimate the risk of complications in the year after a treatment such as balloon angioplasty.
The model calculates the likelihood of readmission, the development of heart failure or the need for renewed intervention. These are risk indicators that would typically require a doctor to manually review many sources.
The plan is for the algorithm to be integrated into doctors’ existing information technology systems and to provide continuous support during examinations and in decisions about individual patients.
What 60 Danish cardiologists really think about AI
In the study mentioned above, Søren Albertsen Rand and colleagues asked 60 cardiologists working in departments where the large-scale trial of AI is being tested to complete a questionnaire about their attitudes towards AI in their work.
The study is a cross-sectional study, which means that the researchers only measure attitudes at one point in time. It cannot, however, reveal how doctors’ views on the technology may evolve as they gain more experience with it. In addition, all participants come from departments already trialling AI, which may mean they are more positive than doctors elsewhere in Denmark.
The study only measures doctors’ expectations and perceived preparedness – not whether the technology actually improves diagnoses, treatments or patient care. This will require separate clinical trials in which AI is compared directly with existing practices.
The questionnaire focused on how much doctors trust the technology, whether they feel sufficiently prepared and whether they are ready to use it in clinical practice.
The questions were answered between September 2024 and December 2024.
The researchers divided the doctors into those with more or less than 10 years of clinical experience.
“The biggest concern is, of course, whether AI can help doctors detect diseases earlier or faster; otherwise, patients will not benefit from it. The second concern is whether doctors are ready for it, because if they are not, the clinical outcome will not improve,” explains Søren Albertsen Rand.
Trust – but also clear reservations
The study shows that doctors have confidence in the technology’s accuracy and are ready to use it in their work.
An important scientific question is whether the model has been trained on a population that is sufficiently similar to Danish patients. If the training data come from other countries or from different types of hospitals, the accuracy may decrease when the algorithm is used in practice.
Research from abroad shows that AI models can be highly accurate in controlled test environments, but performance may shift when AI is applied to new types of patients or clinical settings. Doctors’ confidence is therefore conditional on the algorithm functioning reliably in real-world Danish data.
“Overall, I am impressed with the result. I had expected there might be some resistance or reservations, if only among the older doctors, but we did not see that. The older doctors were just as ready to start using AI as their younger colleagues,” says Søren Albertsen Rand.
Ready for AI – but does it work in practice?
Søren Albertsen Rand believes that doctors are generally ready to work with AI, regardless of their specialisation.
“But we can only answer the question of whether we should use it once we have the results from the large-scale tests, in which the technology is being tested directly against current clinical practice,” says Søren Albertsen Rand.
The upcoming randomised trials will now be the deciding factor: can the technology not only predict but actually improve patient outcomes in the real world?
If the answer is yes, Denmark faces one of the biggest shifts in cardiac treatment in decades. If the answer is no, AI will have to go back to the drawing board.
