Hearing Aid Recommendation Engine V2

Which is all an entirely fair and acceptable approach in an open market.

I can only imagine waht the US market is like WRT the over the counter device claims. I can see the inherent value of an independent tool in that respect, with the appropriate caveat that in this industry the new client largely can’t hear what they can’t hear: and without some appropriate interpretation and verification, they still won’t.

I know you aren’t approaching it from a position of malice either - it just strikes me as ‘computer (web/ai) says X’ might carry more weight than you realise in terms of recommendation. Especially if the AI begins to feed off it’s own suggestions and source material. There’s significant possibility of both inadvertent ghosting and placed recommendations driving source material - even if they have no basis in actual fittings or customer experiences.

Best of luck with it.

Thanks again for the engagement on this… I definitely see your points. They are valid, and I will do everything I can to ensure we don’t overreach with this. On this point specifically:

@Um_bongo said:

it just strikes me as ‘computer (web/ai) says X’ might carry more weight than you realise in terms of recommendation

I realize it will carry some weight, and that’s good, because we believe there are a lot of people making poor decisions in the OTC market place that we can help… not to mention the people going to Miracle Ear, etc, and spending 8 grand for what could be a 1.5 grand solution… again, when it comes to prescription aids, we’ll proceed very cautiously in terms of the wording around recommendations, but I think when it comes to helping people at least find good value, this tool will be helpful…

@AbramBaileyAuD said:

maybe after failing on their own and seeking professional guidance.

I bet a huge ratio of the patients that aren’t at all satisfied put them in a drawer with satisfaction that they’ve proved hearing aids don’t work.

WH

Even more the reason to pursue this imho. Thanks WH.

@DoctorNo said:

Tried it a few times. Gave me completely different devices depending on single parameters and when I added a full set of requests the only device it gave me was one that lacked half the requested features even though I currently use a HA that has all of them.

Thanks for giving it a go and for providing feedback!

Could you share the query by chance? I’ll take a look.

@DoctorNo said:

Would ask that the output be a ranked list (and the only output I’ve gotten are the three cost levels of the exact same HA, which is silly) with 8-10 options as opposed to 3.

I have been meaning to add this. It’s shown in my debugging mode, but not in the public facing version. Will add next time I’m working on it!

@MDB said:

I know it’s not the intent, but what I’d find useful is something that could synopsize some of the advice professionals have given on the forum.

I’m actually working on something like that … which draws on information from multiple sources (our articles, videos, and the forum) to answer questions… more on that soon

Sample from my testing:

This is an interesting thread, have you finally got fed up with Matthew Allsop?

Will you call the bot Matthew in his demise

Its an AI version of Matthews reviews and guidance… Virtual Matthew

@DoctorNo said:

Severe to profound. Good directional control. Best for conversations in loud social environments. AI. Rechargeable.

I got everything but “AI” with the Phonak Bolero M PR M90… The issue is that we aren’t tagging the Infinio as appropriate for profound hearing loss, even though with adequate setup (UP receiver) it could accommodate some degree of profound loss (phonak claims it can handle a flat 95 dbHL loss, though that would be firmly on the edge)… and of course, profound only starts at 90… so I’m just being conservative with what I say is OK for profound loss. The same query with just severe gives you the Sphere… If we get to the point of cataloging fitting ranges and accepting audiograms, then we could get way more accurate with queries like this.

@threluja said:

Its an AI version of Matthews reviews and guidance… Virtual Matthew

Ah I see. Yes, a virtual version of the whole team

It was the next option available (Sphere wins by a small margin though due to overall score):

@rfmparis said:

I would think that many users might want to search for Costco hearing aids (or their equivalents) since they are so dominant in the industry.

When I specified that “I want to purchase a Costco hearing aid” only one result was returned (Rexton Reach R-LI T 80) with an incorrect price.

The system returns the Rexton and Jabra aids a lot because we’ve tested them, and given the value their overall scores are great… Having said that, I had the pricing wrong on Philips and Rexton (apparently they raised the prices from 1499 to 1599 and this got updated in our articles, but not in our product database), but this is now fixed.

Jabra 30 and Sennheiser need to be added to our product database. We’ve covered both extensively, but I was waiting for Steve to review the Jabra 30 before adding the product to our database, which has taken longer than expected. Costco devices are a real pain to get a hold of. And the Sonite R … is that nationwide now? I was waiting to see if that would happen before adding it.

@rfmparis said:

It would be helpful to see all four models currently offered with their correct pricing.

I’ll make the system to handle that. Thanks!

Resurfacing this thread…

I have a new experiment that feeds off the forum and HearingTracker articles and videos… It’s not just a product recommendation engine, but can actually answer other questions too, like “what is hidden hearing loss”…

Give it a try4

That’s not bad actually: it’s more balanced if not yet nuanced.

There’s an over representation of the Jabra product, simply because the forums seem to recognise Costco products over other sources, plus the extra amounts of reviews. Despite the equivalent GN ‘private market’ product not really making the same impact.

That tells me the engine effectively favours the ‘most reported’ performance rather than real world performance. Or to put it another way, it’s harvesting a collective subjective opinion on the market rather than objective data.