Each year, up to 3.8 million sports and recreation-related concussions are estimated to occur in the U.S., but the difficulty in diagnosing the condition can mean up to half of concussions go unreported.
Today, most diagnoses are based on self-reported symptoms and clinical evaluation, but a new diagnostic tool being developed by Thomas Bottiglieri, D.O., chief of the primary care sports medicine division in the Department of Orthopedics at NewYork-Presbyterian and Columbia, aims to provide a more accurate diagnosis based on a measurable novel biometric uncovered by him and his colleagues while doing research.
“At one point or another, physicians have to help athletes face retirement decisions. We consider whether an athlete has had too many concussions, or if they are at risk later in life for neurodegenerative disease,” Dr. Bottiglieri says. “But we currently have virtually nothing with which to inform these crucial decisions other than our subjective expert opinions.”
An ‘Accidental’ Biomarker Discovery
The current standard for diagnosing concussions is vestibular-ocular motor screening (VOMS), which evaluates five areas of vestibular and ocular motor function that are based on patient-reported symptoms. A positive VOMS test suggests potential impairments associated with concussion; however, the test lacks cut-offs that distinguish normal from abnormal results, leaving interpretation largely to the clinician’s experience. Recognizing these limitations, Dr. Bottiglieri collaborated with Dr. Linus Sun, who at the time was a neuro-ophthalmologist at NewYork-Presbyterian and Columbia, to conduct research with the goal of establishing clear thresholds for VOMS.
During patient evaluations, the researchers noticed a common biomarker in patients with more severe symptoms. “While testing their eyes, another signal unrelated to ocular motion was very consistently popping up in concussion tests,” Dr. Bottiglieri says. That signal was slightly altered stability in a person’s head, neck, and gaze that is related to proprioception, the body’s sixth sense that perceives its position and movement in space and is essential for coordination and balance.
In individuals with cerebellar injuries, the inability to stabilize the head leads to abnormal head movements and affects proprioception. “That stability is altered in neurodegenerative diseases such as essential tremor and ataxia,” he adds. “If that system can be disrupted in neurodegenerative diseases, we thought, maybe we can detect it in acute injury as well.”
If we can detect changes before imaging does or before there is a change in behavior, then maybe there are interventions we can implement earlier that impact outcomes.
— Dr. Thomas Bottiglieri
Using Artificial Intelligence for a More Accurate Concussion Diagnosis
Detecting that signal is nearly impossible to do with the naked eye. To create an accurate diagnostic tool, Christopher Driscoll, who was a graduate student working with Dr. Sun and is now a research coordinator in Dr. Bottiglieri’s lab, developed an algorithm that used machine learning to isolate the signature biomarker. Using data collected from more than 200 patients, including control patients who did not have a concussion, he was able to quantify an amplitude and bandwidth frequency that was consistent with concussion.
The algorithm is the basis of a new concussion detection software tool the team is developing called ProScope. To administer the test, patients wear an eye tracking headset and focus on a visual target. “They may think we’re testing their eyes, but we’re actually assessing their head and neck stability and movements,” says Dr. Bottiglieri. ProScope then captures and analyzes these movements to produce a digital signal that translates to a metric that indicates normal or abnormal neurophysiological function. The tool can detect the biomarker with 80% to 90% sensitivity.
“It’s impossible to fake the signal because you can’t intentionally create the movement at the patterns that we recognized,” says Dr. Bottiglieri. “We are measuring an unconsciously controlled pathway, which is very different from other concussion tests, which can be manipulated by players who just want to get back on the field.”
Dr. Bottiglieri and team plan to expand enrollment in their trial to validate the tool further, with plans to publish their initial results later this year. Because the assessment only takes three minutes to complete and requires minimal training and equipment to administer, he believes the technology can be developed commercially to be used in doctors’ offices, athletic training rooms, or on the field, and foresees a future where it could potentially be used to detect other neurodegenerative diseases.
“If we can detect changes before imaging does or before there is a change in behavior, then maybe there are interventions we can implement earlier that impact outcomes,” says Dr. Bottiglieri. “It could really be life-altering for some patients.”
Columbia University filed a provisional patent application for the ProScope technology in 2025.
