Real Ear Measurement (R.E.M.)

@user490 said:

Does research support some reason that hearing tests covering more frequencies in the 250-8k range would not provide benefits? With digital aids curves could be better smoothed which might clean up the signal enough to benefit some.

When I let our engineers consult with a local university helping Sonic Innovations develop the first 11-channel digital hearing aid, the Natura, 11 channels were a big innovation. Previous digital hearing aids had 4 channels. We were using 22 bit chips in our products, which were the cutting-edge chips at that time. I asked why they were limiting themselves to only 11 channels and was told that the data from more channels would be expensive to collect and of limited value in improving speech recognition.

Today’s DSP chips for hearing aids can process 128 channels of data. Now that would be a smooth audiogram curve, but it would take an entire day to test 128 frequencies, and after fitting all of those, 99% of the users would not be able to detect the difference.

Originally, hearing aid companies used the channel count as a marketing pitch. Our hearing aid has 16 channels instead of a measly 11. Then it was 22 and finally 24. The Phoank Sphere i90 uses a 20-channel chip. But it’s not about curve smoothing; it’s about processing data for Autosense. Autosense is supposed to provide a better hearing experience by adjusting gain across frequencies based on the current hearing environment. And it does for some people. However, adding more channels won’t make it work better and create the just-noticeable difference that makes it worth more money.

Because sound processing chips are very close to their practical limits in improving hearing, companies are moving to AI, which can process millions of possible changes for an environment and choose the best one on the fly. AI is going to make the next couple of generations of hearing aids much better. Later generations of AI driven hearing aids will blow out our ears, make our eyes bleed and lead to a slow, painful death in a bid to wipe out everyone with defective hearing in order to improve the gene pool. But until then, we’ll all hear better.

You’ve asked about research before. Obviously, you feel that is a great source of information. I’ve published 32 research papers and have written 134 scientific articles. I don’t bother looking at the “research” most of the time because 90% plus of it has little practical value. I want in-the-moment improved hearing, not a hearing aid that was designed to fit the research.

Setting aside that fitting prescriptions are synthetic constructs from research, the curves cannot be fit precisely with the number of handles currently available. Is it close enough? Could more handles allow for better outcomes? Adjusting gain on the fly or denoising does not preclude improving the fit to match the loss at the same time. Brains adapt some but relying on that flexibility without also developing algorithms that more closely match the hearing loss may not be ideal.

Gotta love these Posts, it’s quite amazing how the thread “flows” like a pendulum swing!

@user490 said:

the curves cannot be fit precisely with the number of handles currently available. Is it close enough?

You can get up to 20 or 24, so you can get a little bit more detailed, one would think this would be enough for speech, however is it enough for music?

Ignoring the question of whether there is utility, do we not have the technology to use REM to match to much smaller resolution and make hearing aids with similaly sized to even smaller handles, think thousands or more. It would probably be too annoying for a human to adjust but a computer could iterate and make the adjustments to best fit the curve. Even now it is unclear why a human has to do the initial adjustments.

Perhaps so many handles would too power demanding but we’re discussing possible hearing benefits not the practicality of making the product.

It individualised the fitting for each ear’s unique ear canal acoustics ensuring that soft sounds are audible, loud sounds are comfortable and average speech isn’t too compressed by soft and loud speech so loudness perception is as natural as possible and speech isn’t distorted. It does this in real-time in the ear. Without REMs you cannot accurately measure any of these things. In verifying the acoustics of the device, you are ensuring many things in the chain of events required for a successful fitting are occurring in an objective manner and addressing many of the individual differences between ears (and individuals) that cannot be predicted by a manufacturer’s software. In short, it eliminates a lot of guesswork and establishes a good starting point for a fitting. In most cases people don’t vary more than 2 - 3 dB away from this starting point as their goal endpoint, although first time device users may have an automatic adaptation manager activated to slowly get them there over time.

Studies have consistently shown that patient satisfaction and real-world outcomes with hearing aids are superior when REMs are done properly to not using them at all. Just as you would expect custom-made shoes to fit your feet better and be more comfortable than a one-size fits all option.

Phonak’s autoREM is very accurate if done properly for vented and closed fittings (open I’m not so sure). Unitron’s was not good at all the last time I checked. Oticon’s is fair but I’d rather do manually. It all depends on how much effort the manufacturer has put into developing their autoREM process.

@Hearing_Potential, I’m curious about this: in theory AutoREM provides a closed loop: the fitting software adjusts the gain and then, via REM, can determine if the results are correct. With enough iterations, it should be possible for the fitting software to achieve a near perfect result.

Is the issue that the fitting software tries to do this in one or very few iterations and that’s why it’s imprecise?

(I have never had AutoREM done on me, so I don’t know how it operates)

Who are you asking. Me? I think your dodge is an answer.

Absolutely it is the case that your individual cochlea is a black box until you die, because opening it up to have a look will damage it. That doesn’t mean that we don’t know a lot about the function of the cochlea. Why do you think that a finer frequency spacing will be better? Why do you think that you are the only person who might have considered that rather than thinking that there might be reasons that it actually wouldn’t be better? Amplifying dead regions seems like it might be a problem, but why do you think no one other than you has asked that question?

Oh fun, I’ll bite. QuickSIN, HINT, ABR, ECochG, ASSR, P1, ACT, DPOAE, TEOAE, tympanometry, acoustic reflex thresholds, acoustic reflex decay…

For me, the benefit of REM has been quite clear.

I learned in an environment where it was quite divided. Everything I was learning from reading and otherwise was teaching that REM was very valuable and critical to a successful hearing aid fitting. Everything I was told by the people around me doing the work and the reps from the hearing aid manufacturers was that REM was hardly effective and that hearing aids fit to REM were much too loud. For some time, I tried to follow what the people around me would tell me. But I would try to learn more, and especially when people did not like their hearing aids, I would implement REM. I don’t know that I can properly explain the difference I saw. I had many people coming back and telling me that these were now the best hearing aids that they ever had. Many of these people did not come from the people I directly worked with, but from a huge variety of places: well-known hospitals, highly regarded providers, and from other states altogether. I had people start crying because it was the best that they had heard in decades of wearing hearing aids. I had people get mad and ask me why no one did this before.

Over some years, I’ve had new people come to work with me and many had the same prior experiences with REM that I had before I started using it consistently: That they were told it was too loud and generally a waste. That REM is pushed by people who have no idea what they are talking about and don’t actually fit hearing aids to “real” people.

I would spend time showing them through helping them gain experience using it and demonstrating the results with people. At least one of them remarked that they thought the way things worked were that first you set the hearing aids to first fit, and then the person came back very upset about a whole bunch of things, and you would spend that appointment calming them down and making adjustments. He said it was so weird here because most people just came back happy. No, not everyone.

I realize that it is a tool and not magic. But it’s the closest thing to magic I’ve seen with hearing aids so far. And until someone gives me a tool that does better, I will continue to believe in the merits. I’m not against someone providing something better. And it doesn’t mean that you can’t try other things when there continues to be issues.

There is a lot of room for improvement in hearing aids. Users and providers would welcome them. I know I can’t wait for bluetooth to be easier to use and more consistent of a connection.

@mindsil said:

I know I can’t wait for bluetooth to be easier to use and more consistent of a connection.

In my experience, the need for REM is dependent on the brand’s fitting software and the skill of the fitter. Not all fitting software is created equally. And fitters sure aren’t.

Bluetooth stability, range and connectivity appear to also be brand-dependent along with the phone type and brand. I don’t have any Bluetooth issues with my Phonak Sphere i90s. They’re connected to a Motorola phone, and this Bluetooth is the best I’ve ever had. The TV streaming range is incredible.

Oh for sure, I didn’t even touch on CI test battery!

Oh, that’s really hurt…! <wink, wink>

@PaulC said:

I didn’t even touch on CI test battery!

Do you have the credentials for CI?