Detecting concussions with the sound of your voice

More than 50% of concussions in the United States go undiagnosed, and about 70% occur in sports settings. Star players in pursuit of big wins often brush aside what might be serious brain injuries to stay in the game.

On top of that, identifying a problem during a football or soccer match can be challenging. Typically, an athlete is asked a series of questions – Do you have a headache? Are you dizzy? – and a decision is made about whether to sit out and, usually later, whether tests such as CT scans should be ordered.

Rahmina Rubaiat has plans to make getting a concussion diagnosis fast and easy. She examined hundreds of audio recordings of players saying specific words and sounds before and during football season. The collected data includes samples from participants who eventually experienced confirmed concussions and from uninjured control subjects. Voice changes imperceptible to the human ear but detected by artificial intelligence – in measures such as amplitude, frequency and vibration – were found to correlate with documented brain trauma.

Rubaiat is working to simplify the process with the goal of creating an app that would allow athletic personnel to capture one-word or single-sound baseline recordings and later, as needed, administer sideline tests. Results would include an indicator of degree – mild, moderate or severe concussion – to inform next steps and possibly prevent long-term cognitive impact.