In 2013, Duke University athletes Ivonna Dumanyan and Gabrielle Levac both found themselves sidelined by sports-related injuries. During their rehab, they commiserated over the constant onslaught of preventable injuries they’d experienced over the course of their careers.
“That’s basically how injury works,” Dumanyan tells Hypepotamus. “You get hurt, you develop a compensation, you hurt something else, take the time to recover, and you compensate differently then get hurt again. It keeps going.”
With their combined backgrounds in biomechanics and machine learning, Dumanyan and Levac realized that they could create a sensor to monitor an athlete’s movements in a manner similar to methods used by lab technicians, doctors, and physical therapists.
“The type of elite care and athletic training rooms we were getting at Duke is not available to most people. And it wouldn’t be available to us after graduation,” says Dumanyan. “Anyone who’s active and who’s experienced preventable injury could benefit.”
Currently, experienced physical therapists examine the way you move in person. They can determine if you have a certain weakness on your knee or hip. After physically examining an injured athlete, they can then create a plan of exercises designed to restore proper range of motion.
The duo co-founded fitness tech startup Fathom AI to commercialize the wearable biometric sensors these health professionals use to help athletes prevent injury, optimize recovery by increasing mobility, or rehab old injuries.
“We can’t be a PT that’s putting their hands on your body and watching you move, but we can replicate the types of measurements that those motion capture spaces are getting,” says Dumanyan.
Similar to the way movies like Avatar get made, Fathom AI uses motion capture data inputs to analyze human movement for injury prevention. They worked with UNC and Duke to use their motion capture laboratories with high-speed cameras and force plates.
The technology grabs the athlete’s motion data from three quarter-sized sensors — two for your ankles and one for the small of your back — that adhere to your skin like a Band-Aid.
The machine learning-powered sensors run on a suite of analytics that helps Fathom AI measure the way your joints move, as well as the force you produce and absorb when do so.
From those core movements, the sensors processes data about the compensations that the athlete makes.
“We also derive information about how you fatigued over time, how your form changes at different speeds or while moving in different planes of motion. We extract nearly 180 different variables and metrics that we monitor,” says Dumanyan.
However, Dumanyan says, the company does not plan to focus on the rehabilitation aspect itself at this time.
After the data capture process takes place through the sensors, the machine learning component translates the data into what your body needs at the moment and creates a hyper-personalized exercise plan, just as a physical therapist would.
As your body changes, so does the exercise routine, which includes muscle activation, stretching, foam rolling, and more.
The startup claims this routine reduces recovery time by up to 30 percent.
“It’s all about moving better and building strength more intelligently to balance out imbalances in your body to prevent injuries,” says Dumanyan. “The products that we have today are 20 percent driven by our vision and 80 percent driven by the actual needs that we heard from athletes.”
The Durham-based startup operates on a D2C model. The sensor kit is available to customers on Fathom AI’s site with an accompanied monthly fee. There’s also a freemium version of the app with limited features.
Fathom AI has raised $3.2 million in a seed round to date.
To scale their product, Fathom AI is currently partnering with other fitness apps to help incorporate their intelligent data system and create more robust strengthening exercises for their users. They will be partnering through API with three large digital fitness content creators in 2020.