One of the cool things of my lab is that there are so many different people. There’s a good number of engineers, a good number of audiologists, but we also interact with people with a background in computational linguistics or physiology, sometimes basic physiologists. The name of my lab is Laboratory for Translational Auditory Research. Trying to translate basic findings into actual practice is really important for us. And that’s something that engineers and audiologists have in common: We like to see things work in real life.
One of the themes in my lab is the study of plasticity in the human auditory brain. It’s quite amazing. The brain is designed in such a way that you can throw at it a very degraded signal, and yet, humans can understand speech under extremely degraded conditions. That’s very fortunate for cochlear implant users in particular, because the signal they receive is extremely degraded.
One of the types of degradation that the signal undergoes in the case of postlingually deaf hearing users, is that the frequency map in the cochlea when these people had normal hearing, can be shifted and distorted with the map we impose with the cochlear implant. For example, the place that was stimulated by 100 hertz when these people had normal hearing may be 25 mm into the cochlea, and the new place that is stimulated by 100 hertz can be quite a distance away — 6 mm or 7 mm is a big distance in the cochlea.
So we’re interested in seeing how much people can adjust to that, and in finding whether that adaptation is complete or not. In cases when it is not complete, what we can do is, instead of asking them to adapt, we can adapt the signal processing to what they can hear. We’ve developed a tool where a patient tweaks a dial or changes something to make their frequency location table — that frequency map that I was talking about — sound as natural as possible.
The Role of Patient Feedback in Guiding Research
Patients frequently have important insights. They tell you things you wouldn’t have thought about. So I always pay close attention to whatever they have to say.
We were doing this experiment with one patient one day, and in this task, there is a large rectangle that is a grid, and depending on where he clicks on the grid he hears the same sentence processed with a different frequency table.
So he found the region — different from the standard one — that made speech sound more natural. This is a person who knows what natural sounds like, because he has very good hearing in the other ear. “This is the most natural,” he said — and that was for a frequency map that, unlike clinical ones that extend up to 8,000 hertz, this one went up to maybe 12,000 or 13,000 hertz. But he said, “It’s not the most intelligible. If I pick this other map –” the map only went up to 5,000 or so hertz, so it was in the other direction with respect to normal. He said, “– This sounds very distorted. It’s not good. But I can understand speech better with this.”
That turned out to be consistent with some mathematical modeling of speech perception we had done. And, of course, if you ask any experienced CI audiologist, they will tell you that the most natural is not necessarily the best always for speech perception. But this experience with this patient made this comment very sharp and focused in my mind. And it’s guided our thinking in the lab ever since.
Challenges in Translating Findings into Clinical Practice
It’s easy to convince your closest friends and colleagues that you should do this or that. But in terms of wide implementation, it’s hard.
If you find something that might be useful to provide a modest improvement to 10% to 20% of your population — like I think might be the case for this idea of adjusting frequency location tables — how much does that matter to different people?
To the 80% or 90% who don’t need that adjustment, it doesn’t matter at all. For a cochlear implant company, of course, they want to see their numbers higher. But they also want to provide audiologists with a streamlined fitting situation. Audiologists are under a lot of pressure to do a lot of stuff in a short amount of time that’s not very well reimbursed. So that’s one of the realities of life.
If you want to propose a procedure that adds 5 or 10 minutes to what they do, that’s a high threshold. You need to show that it’s worthwhile, that it’s justified. And the first thing is to convince cochlear implant companies that it’s worthwhile and should be included in their fitting software.
With respect to this particular issue of frequency location tables, companies have gone back and forth in terms of giving audiologists the flexibility to be able to change them, or not. Depending on the version of the software you’re looking at, it may or may not be possible.
But even when you have a version that makes it possible, there’s no guidance for the audiologist. We have these default settings for a map. If you want to depart from those, in what direction should that go? The tools to do that don’t exist, and that’s one thing we’re trying to create.
But it is very difficult for us and for everyone to make sure that what we develop is translatable into clinical practice. There are issues of how long it takes and how effective something might be overall.