Spotting a concussion can be tricky. After a potential head injury, you can ask if the person feels dizzy or has a headache — but that relies on self-reporting, which isn't always accurate.
What if there were a way to take the guesswork out of it?
That's the problem Trent Guess, an associate professor at the University of Missouri College of Health Sciences , and Jacob Thomas, a Mizzou doctoral student, have set out to solve. They've developed a portable system that uses machine learning to track body motion and detect possible signs of a concussion — offering a more objective way to identify injuries in real time. While most concussion-testing machines are expensive and only available in specialized labs, the Mizzou Point-of-Care Assessment System is affordable and portable, making it more accessible to the general public in a variety of settings.
In a recent study, the portable system — which combines a force plate, a depth camera and an interface board — was used to test the movement, balance and reaction times of 40 college athletes, 20 of whom had recently been diagnosed with a concussion.
The system's portability allows it to be useful in a variety of clinical settings.
By using machine learning — a process where a computer looks for patterns in data to make better-informed decisions — the portable system was able to quickly learn the difference between someone who is healthy and someone who has a concussion.
For example, those with a concussion were more likely to have slower reaction times, walk slower and struggle more with maintaining their balance, particularly when asked to complete tasks with eyes closed or while counting backwards from 100 in intervals of seven.
"Going forward, we can have individuals take the assessment when they are healthy to establish a baseline. This is where the machine learning comes into play," Guess said. "Then, if they sustain a possible concussion in the future, taking the same assessment again and comparing the results with the baseline data helps us know if the individual's movement and motor control were impacted by possible cognitive damage."
Guess, whose background is in engineering and biomechanics, is the director of the Mizzou Motion Analysis Center, a state-of-the-art gait lab in the College of Health Sciences.
"Mizzou is a great place for this research because of the opportunities I have to collaborate with clinicians who have expertise in orthopedics, physical therapy, sports rehabilitation and exercise science," Guess said. "In their clinical settings, they often work with a variety of patients and see the need for a better way to assess potential concussions, and with my engineering background, I knew I could help develop a portable system to help meet the demand."
While college athletes were the subjects of this particular study, the Mizzou Point-of-Care Assessment System could also help assess potential concussions for individuals working as first responders, military personnel or in other professions with a higher likelihood of head injuries.
"This portable system can also be a helpful tool in measuring an individual's recovery after a concussion," Guess said. "The last thing you want is to send someone back out on the field or into a work environment when they are not ready. Hopefully, our efforts can help people stay as safe as possible."
While a small number of these portable systems are currently being tested for research purposes, Guess hopes they can one day be manufactured at a larger scale before being used in a variety of clinical and athletic settings.
"A machine learning approach to concussive group classification using discrete outcome measures from a low-cost movement-based assessment system" was published in Medical Engineering and Physics.
