THIS CONTENT IS BROUGHT TO YOU BY NTNU Norwegian University of Science and Technology - read more
Migraines leave traces throughout the body that researchers can now detect
“There are many inconsistencies with the diagnosis of migraines. It appears that AI could help us make the correct diagnosis,” says researcher.
Migraine may be a biological pattern that leaves traces throughout the body and that artificial intelligence can recognise. This is shown by a recent study from NTNU.
“There are many inconsistencies with the diagnosis of migraines. It appears that AI could help us make the correct diagnosis,” says Anker Stubberud. He is a physician and headache researcher at NTNU.
Researchers used AI to analyse information gathered from more than 43,000 participants during the county-wide health study HUNT.
The study shows that the condition can be identified through a complex interplay between genetics, clinical characteristics, and environmental factors.
“Migraines are a major health problem worldwide and have a considerable impact on quality of life. Accurate diagnoses are important so that people can receive the best possible treatment,” says Stubberud.
A major public health problem
Half of the world’s population experiences headaches. For some, the headaches are so severe that they are categorised as a migraine.
Migraine is characterised by symptoms including throbbing headaches, sensitivity to light and sound, vomiting, and nausea. The condition affects around 15 per cent of the Norwegian population.
Migraine is the leading cause of disability among women under the age of 50.
One challenge in identifying migraines is that the diagnosis is based entirely on symptoms.
“There are no blood tests or other biological tests that can be used as part of the diagnostic process. All we have is the person’s description of their symptoms. If the diagnosis is incorrect, the treatment may also miss the mark. We are therefore exploring whether we can identify biological markers that could help practicioners make the correct diagnosis,” says Stubberud.
This is where AI comes in.
Able to distinguish people with migraine from healthy individuals
The researchers first developed an AI model that could distinguish people with migraine from people without headaches. What was unusual was that the AI was not given access to the participants’ headache symptoms.
Instead, it used data such as age, sex, and general health information.
“The information included everything from constipation and medication use to back pain. We included 60 different variables. The fact that AI could identify migraine so accurately without knowing anything about the headache itself suggests that the condition leaves traces that extend far beyond the attacks,” says Stubberud.
He believes this could be a sign that migraine has a distinct biological signature that can be detected. The study also revealed several previously hidden subgroups of migraine.
Four different forms of migraine
Using a method in which AI independently searches for patterns in the data, the researchers analysed more than 12,000 people with different forms of headache. The analysis initially identified two clear main groups.
The first group consisted of 1,425 people, more than 90 per cent of whom met the criteria for migraine.
The second group consisted of 10,760 people. The vast majority had other types of headache that only partially macthed the criteria for migraine, or had other forms of headache altogether.
This suggests that AI was able to identify a distinct migraine profile independently of traditional diagnostic criteria.
The most interesting discovery came when the researchers took a closer look at the migraine group. They found four subgroups, each with different characteristics:
1. An all-male group
This subgroup stood out clearly because all of the participants were men. The finding may indicate that migraine in men has distinctive characteristics that are often lost when researchers analyse women and men together.
2. A group with prominent neck pain
People in this group reported significant neck pain in addition to migraine.
This could provide new insight into the well-known but still debated relationship between neck pain and migraine.
3. A group with extensive musculoskeletal pain, anxiety, and depression
This subgroup had a higher prevalence of widespread pain, mental health problems, and other health challenges.
This could represent a more complex form of migraine in which several biological systems are involved simultaneously.
4. A group with classic migraine
The final group had symptoms that more closely matched the established definition of migraine, including migraine aura, without the pronounced additional health problems that characterised the other groups.
Genetics supported the differences
The researchers also compared genetic patterns between the groups to investigate whether the subgroups were biologically different.
Traditionally, an established risk score is used to calculate the genetic risk of disease based on many small genetic variants.
In the NTNU study, the researchers also tested newer AI-based genetic risk models.
The results showed that AI was best at distinguishing between the different migraine groups.
“This strengthens the hypothesis that migraine is not a single disease, but rather a diverse group of different biological conditions,” says Stubberud.
Could pave the way for personalised treatment
The study suggests that diagnosis in the future could become more data-driven, serving as an aid alongside the doctor’s clinical assessment and the patient’s description of their symptoms.
“So far, much of the research on AI in healthcare has been based solely on numbers and statistics. AI ought to be tested on patients who seek actual medical help for headaches. This research is still lacking. These AI tools have not yet been tested in clinical practice,” says Stubberud.
He emphasises that if different forms of migraine do indeed have different biological causes, this could also have implications for treatment.
“Some patients may respond better to certain medications or preventive measures than others,” he says.
References:
Danelakis et al. Machine Diagnostics and Machine Phenotyping of Migraine: A HUNT Study, Neurology, 2026. DOI: 10.1212/WNL.0000000000218076
Stubberud, A. Artificial intelligence in headache care (Abstract), Nature Reviews Neurology, 2026. DOI: 10.1038/s41582-026-01226-7
———
Read the Norwegian version of this article on forskning.no
More content from NTNU:
-
Insect expert hunts for unknown life in New York
-
No, your family relationships don’t get worse even if you spend a lot of time on social media
-
Can artificial intelligence help save nature?
-
Elderly people living at home have less control, not more
-
Researchers are looking for solutions that will encourage people to leave their cars at home
-
Trade unions help raise wages and reduce the gender pay gap
