Artificial intelligence (AI) is playing an increasingly important role in health research, both as a tool in research processes and as a technology that can improve patient care.

We have spoken with Professor Stine Korreman, researcher in radiation therapy at the Department of Clinical Medicine, about how her research group is developing AI models to improve treatment for cancer patients.

How do you use AI in your research?
We work with the treatment of cancer diseases through radiation therapy, and we use AI to analyze CT, MRI, and PET scans to identify tumors and healthy tissues. Our models are trained on many years of patient data, making it possible to develop systems that can help delineate areas in scan images and support medical assessments in treatment.

How has the technology evolved in your research field in recent years?
AI and deep learning have played a major role in the development, and especially since 2016-2017, we have seen a significant shift in the use of AI, and there has been almost an explosion in the number of scientific articles about AI in health research over the past five to six years.

Explain specifically how the technology can be used in clinical practice?
Until now, doctors have manually reviewed scan images, where they have to draw tumors, organs, and healthy tissue on scan images. It is a time-consuming process that can take several hours per patient, and which can also vary greatly between different doctors. AI can provide a uniform starting point that doctors can then fine-tune.

What is the potential for the technology?
The greatest potential lies immediately in the saved resources. AI can save thousands of work hours per year by automating the most time-consuming parts of the process, freeing up more time for patient contact and ensuring faster and more efficient treatment. There is also the prospect that AI can improve patient treatments in some areas, but this has a somewhat longer time horizon.

How far are we from being able to use it in clinical practice?
AI is already being used for some tasks, for example to identify healthy structures such as bones, organs, and muscles, but when it comes to identifying tumors, there is still some way to go. We have just completed a trial for head and neck cancer, and we are soon starting a large national trial for breast cancer.

What perspectives do you see for AI in healthcare?
AI can save doctors many hours of manual work and ensure a more uniform assessment of patients' scans. We expect that AI can become a natural part of workflows in treatment, so doctors can, for example, spend more time on patient contact instead of reviewing images.

How can Denmark and AU position themselves in the international AI competition?
Denmark has a unique position in AI development within healthcare because we have one of the world's best organized and most comprehensive health databases. We have digital patient records and health registries that make it possible to develop AI solutions based on a large data foundation. Many other countries, including the USA and China, work with AI in health, but often on more fragmented datasets. I therefore believe that Denmark can become a leading player in the development.

What is the next step for AI in your research?

We will continue to test our AI models in clinical trials and improve collaboration between researchers, doctors, and decision-makers. We hope that our work can contribute to AI becoming an integrated part of patient care.

When can we expect AI to become a permanent part of clinical practice?
The time saving from using AI is such a big advantage, and therefore I believe that implementation will happen relatively quickly, as AI is primarily used as a tool to support doctors' decisions. We are already testing AI in national trials, and if they show good results, I think we will see widespread implementation within the next few years.

How do you address the ethical dilemmas of AI in healthcare?
We are in the process of starting a collaboration with the Center for Interacting Minds, which researches AI in healthcare to ensure that our research also takes into account ethical aspects related to the implementation of AI. AI is not just a technological development - it is a fundamental change in the way we work in healthcare. Therefore, it is crucial that we implement AI with the same scientific rigor as other medical technologies.

What are the biggest challenges for the development of AI in health research?
One of the major challenges is access to data. GDPR and other regulations make it difficult, for example, to share health data across regions and research institutions. There is a tendency to regard all health data as extremely sensitive, even in anonymized form, and this can inhibit innovation. If we don't find a good balance between data security and innovation, we risk falling behind in the race.

Contact

Professor Stine Korreman
Aarhus University, Department of Clinical Medicine - Danish Center for Particle Therapy
Phone:  
stine.korreman@clin.au.dk