It’s that scene in “The Little Mermaid,” except instead of Ursula stealing Ariel’s voice with a seashell, it’s a head and neck surgeon, asking politely to borrow your voice to help those on ventilators get theirs back.
It’s LiRA, a UNC technology startup currently collecting video selfies from volunteers. The goal is to use your facial expressions to enhance its algorithms – data it will ultimately crunch to create a pilot lipreading technology its founders say could make a real difference – particularly amid a pandemic putting so many on ventilators.
The idea comes from a situation that co-founder Andrew Prince said is heartbreaking Prince recently completed his otorhinolaryngology residency – head and neck surgery training – at UNC. A patient, who couldn’t speak due to the tube that was helping her breathe, was being told she had cancer.
“I’m trying to lipread … all of a sudden, she is screaming,” Prince recalled. “We can see she is screaming … but it is silent. Because she can’t speak. It’s hard to watch.”
If LiRA realizes its mission, however, patients could soon find their voice. At least that’s the goal and why Prince is hoping the website collecting voice samples will go viral. The more voices his team samples, the better the tool will be at allowing voiceless patients to be part of their medical conversations.
LiRA got its start in 2019. That’s when Prince and his co-founders, Nga Nguyen, Alison Schaefer, Andres Tello and Dina Yamaleyeva, met during a six-month long entrepreneurship class that ended about when the pandemic began. They bring a mix of talents, from biomedical engineering to surgery. The pandemic turned them into a family of sorts.
It’s their first startup, but possibly not their last.
Prince always knew he wanted to work in health care. But he also wanted to do it in more ways than just the operating table – an urge solidified by the patient in the bed having a literal panic attack because she couldn’t communicate.
“I love being a surgeon,” he said. “I love operating. But being in medicine is so frustrating because there are so many problems. There are so many things that could be fixed with things that are already out there.”
He hopes LiRA is his first step in connecting those dots. The New Orleans native said he’d like to solve more problems in the future – and that he plans to take advantage of every resource the Triangle offers. Already, with LiRA, he’s entered a slew of entrepreneurship competitions, from winning second place at the Tulane Business Model Competition (a prize of $30,000) to a first place finish at the recent Covintus Tech Tank pitch competition (a prize of $25,000). He’s obtained the support of NC IDEA. Altogether, the firm has amassed close to $170,000 from grants and competitions.
In the meantime, he’s hoping to get more vocal samples. Volunteers read select sentences – from “I can’t breathe,” to “My chest hurts,” to the camera, and the LiRA team analyzes how their faces move, using the data to “train” the algorithm. If all goes well, a demo will debut in the next two months. Then comes early clinical testing, he said.
Eventually, the applications could go beyond just people on breathing tubes to anyone in need of major medical intervention.
Already, the team has been in touch with major hospitals, as well as speech language pathologists, who may be interested in piloting the technology.