UW team's hearing loss detection hack could change lives globally

Ninety-nine percent of infants in the United States are screened for hearing loss at birth. But when you look worldwide, that figure drops to just 40 percent. The failure to diagnose hearing issues early on can have major implications for those kids. But a team of University of Washington scientists has created a way to address that.

Shyam Gollakota is a professor in UW’s Paul G. Allen School of Computer Science and Engineering and the director of its Mobile Intelligence Lab. KUOW’s Kim Malcolm talked to him about his team’s cheap and accurate testing hack.

This interview has been edited for clarity.

Shyam Gollakota: As you mentioned, every kid in the U.S. gets a newborn hearing screening test. Unfortunately, that's not true with most of the rest of the world. The reason for it is that the medical device equipment to do that is extremely expensive, which means that kids who are born in India or Kenya don't really have a newborn hearing test.

We have smartphones, we have earphones, we have all these devices. The question we asked is, can we reuse these cheap $2 to $3 earbuds to detect these hearing losses accurately?

Kim Malcolm: A very provocative question. What did you find?

We were able to do a large study. We used wired earphones attached to a smartphone. By running algorithms on the smartphone, we were able to hear these very faint sounds emitted by the ear, and accurately use that to detect hearing loss to a standard similar to that of an FDA-approved medical device.

So, these are measurements of something happening inside the ear that you can pick up with the smartphone and the earbuds?

Yes. Unlike a healthy adult, you cannot really ask a newborn if they can hear different audible tones. You can’t just ask them to raise their hand. So, it turns out that the current practice uses the fact that a healthy ear generates sounds of its own. This is really fascinating because we typically think of the ear as something which receives sounds, but it turns out that the ear also generates sounds. So, if you have a healthy ear, the hair cells in your inner ear are going to vibrate in response to external sounds. We can detect these sounds from healthy ears. We cannot detect them when there is some kind of hearing loss issue. And we were able to do this by just using these $2 to $3 earphones.

Is this test difficult for a layperson to run? Because it sounds like you want to scale this up.

All we need to do is basically put the earphone in your ear, and then turn on the app, which plays really faint sounds. The whole test takes around 20 to 30 seconds to run.

So it sounds like a parent or a family member could run the test.

That's our ultimate goal. Right now, tests have been done with graduate students as well as nurses and clinicians. But from a usability perspective, it can potentially also be done by a parent or over a tele-health visit, and then a physician can analyze the results. The whole thing potentially costs less than $10. This is exciting because if you're trying to put together a universal hearing screening program in India, Kenya, or any of these countries where the resources are limited, you might as well not put too much money into paying for medical devices and maintaining them, but rather use those resources instead on human capital, for example, for hiring nurses, hiring people who can provide follow-up.

Which leads me to my next question. Say a family finds that their child does have some impact on their hearing that needs to be addressed. What do they do if they're living in a country with fewer resources?

So that's exactly the point. The first step we are providing here is detecting that there is actually an issue. The second step is potentially follow-up. You need medical care and follow-up to make sure that you have the right resources in terms of, potentially, hearing aids and so forth. That's where this becomes a policy issue. How do you divide up the finite amount of resources? Potentially you can put more resources into the follow-up rather than the initial diagnostics itself.

So, what's next in this process?

Here at the University of Washington, we created what's called the TUNE program. There's a pretty broad team spanning the UW computer science department, global health departments, Seattle Children's, the University of Nairobi, and the Kenyan Ministry of Health as well. The goal here is to help create technologies that can create the universal newborn hearing program in Kenya. We've also made the whole code for this project open source for anyone across the world to download and use in their own context.

What kind of difference could this make in a child's life if they got this early detection?

There are multiple studies that say that earlier detection of hearing loss could really help in terms of neurological development, and also in making sure that the kid doesn't really see any difference in terms of how they're perceiving the world itself. That's the reason why the United States is in a much better state for people who have hearing loss than the rest of the world. I do think that by enabling early detection of hearing loss for millions of kids across the world, we can really make a big difference in their life.

Why are you so interested in this work? What draws you to it?

Growing up in India, my grandmother, who lived with us, had hearing loss. Since we were in a lower-income family, we couldn't afford to have hearing care for her until later in her life. So, this is definitely a problem that is pretty close to me. It's really amazing that I finally have the opportunity to work on this problem and use my skills to make a tiny difference.

What do you think she would have made of the work that you're doing now?

I think she would have loved it. I think she would have basically asked me for the next step, which is for a cheaper hearing aid, for example.

Listen to the interview by clicking the play button above.