Artificial intelligence could predict headaches 24 hours in advance

People who continue to suffer from headaches and fatigue following a concussion may in future be warned a day before these symptoms appear. This is the conclusion of a study conducted by researchers at the Norwegian University of Science and Technology (NTNU), as reported by Medical Xpress.

The research, published in the Journal of Head Trauma Rehabilitation, shows that these symptoms are not random, but follow certain physiological patterns that artificial intelligence models are able to recognise.

"There is strong evidence that the symptoms experienced the following day do not occur randomly in people dealing with persistent problems after a concussion," said Antonios Danelakis, a researcher at NTNU.

How the study worked

The study involved 20 people with persistent post-concussion symptoms, who were monitored over a period of 28 days. Participants recorded their symptoms daily and took measurements of their heart rate, skin temperature, and muscle tension in the shoulders and neck.

Using this data, the research team tested two artificial intelligence models designed to predict the days on which participants were likely to experience moderate or severe headaches. The better-performing model was then also used to estimate the onset of fatigue episodes.

The predictions achieved moderate to good accuracy, with up to 24 hours' notice before symptoms actually appeared.

What the technology could offer in practice

The researchers believe these findings could form the basis for smart applications or devices, such as smartwatches, capable of sending patients alerts before symptoms arise. This would allow those affected to better organise their daily activities, whilst also giving doctors additional information to aid recovery monitoring.

According to recent studies cited in the article, more than one in five people continue to experience symptoms more than three months after a concussion, which can affect their professional life, studies, and social wellbeing.

Further research needed

The study's authors make clear that this represents only a preliminary proof of concept for symptom prediction, rather than a technology ready for clinical use. Further research involving larger patient groups is needed, along with the development of warning systems that meet data privacy and security requirements.

The NTNU team is already preparing a new study, scheduled for spring 2027, which aims to include at least 200 people with persistent post-concussion symptoms.