Health · Medical Research · 6 hrs ago
AI model predicts post-concussion headaches and fatigue a day ahead
Researchers tested AI to predict headaches and fatigue in people with symptoms that persist after a concussion.
The study involved 20 people who tracked symptoms for 28 days at St. Olavs Hospital in Norway.
Participants also recorded measures including heart rate, skin temperature and muscle tension each day.
The model predicted moderate to severe symptoms up to 24 hours ahead with moderate to good accuracy, and a foundation model performed best for headaches.
The findings suggest that symptoms may follow patterns detectable in health data, rather than occurring at random.
A future smartwatch or app alert could help people plan around symptoms, which can affect work, school and social activities, but further testing is needed before such tools are available.
An artificial intelligence model predicted moderate to severe post-concussion headaches and fatigue up to 24 hours before they occurred, with moderate to good accuracy.
The study involved 20 people with persistent post-concussion symptoms who logged symptoms and measured physiological signals daily for 28 days.
Measurements included heart rate, skin temperature and trapezius muscle tension.
Researchers said the findings suggest symptoms may follow patterns that AI can learn from data collected over time.
A warning system could help patients prepare for the next day and provide information to their healthcare professionals.
- Who
- Researchers at NTNU and the Norwegian Center for Headache Research studied 20 people with persistent post-concussion symptoms.
- What
- They tested AI models that use physiological data to predict headaches and fatigue up to 24 hours ahead.
- When
- The study was published on October 30, 2026. Participants logged symptoms and took measurements daily for 28 days.
- Where
- The study was conducted at the Department of Rehabilitation at St. Olavs Hospital in Norway.
- Why
- Researchers aimed to predict symptoms in people with persistent post-concussion issues and explore whether advance warnings could help patients plan.
This story does not have two clearly opposing sides.
This story does not have a timeline yet.
- Participants
- 20 people with persistent post-concussion symptoms
- Symptom logging
- Daily for 28 days
- Prediction window
- Up to 24 hours ahead
- Physiological measurements
- Heart rate, skin temperature and trapezius muscle tension
- Study site
- Department of Rehabilitation at St. Olavs Hospital, Norway
What people are saying
There are strong indications that the symptoms experienced the following day do not occur randomly in people struggling with persistent post-concussion issues
The symptoms can follow specific patterns. AI can learn these patterns by training on physiological data collected over time.