Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
×
Google AI Medicine

Google is Developing AI That Can Hear If You're Sick (qz.com) 29

A new AI model being developed by Google could make diagnosing tuberculosis and other respiratory ailments as easy as recording a voice note. From a report: Google is training one of its foundational AI models to listen for signs of disease using sound signals, like coughing, sneezing, and sniffling. This tech, which would work using people's smartphone microphones, could revolutionize diagnoses for communities where advanced diagnostic tools are difficult to come by.

The tech giant is collaborating with Indian respiratory health care AI startup, Salcit Technologies. The tech, which was introduced earlier this year as Health Acoustic Representations, or HeAR, is what's known as a bioacoustic foundation model. HeAR was then trained on 300 million pieces of audio data, including 100 million cough sounds, to learn to pick out patterns in the sounds. Salcit is then using this AI model, in combination with its own product Swaasa, which uses AI to analyze cough sounds and assess lung health, to help research and improve early detection of TB based solely on cough sounds.

This discussion has been archived. No new comments can be posted.

Google is Developing AI That Can Hear If You're Sick

Comments Filter:
  • As far as I can tell, medical is the one field where A.I. may definitely benefit humanity, as it can even help work alongside new doctors to help them get better understandings, plus serve as a foothold into discovering new, or diagnosing very rare diseases not likely to be come across and diagnosed since it rarely gets seen.

    This is where A.I. may do the best good. I actually applaud this use.

    • by mmell ( 832646 ) on Friday August 30, 2024 @12:18PM (#64748940)
      It'll be notifying HR that Employ NNNNN is attempting to fraudulently claim "sick or medical leave, accounting category code N0283472" and should be issued a warning to log in at the workplace immediately or face disciplinary action.
      • It'll be notifying HR that Employ NNNNN is attempting to fraudulently claim "sick or medical leave, accounting category code N0283472" and should be issued a warning to log in at the workplace immediately or face disciplinary action.

        Extremely unlikely. Tools like this can't issue a definitive diagnosis either way. They're good for cheap, easy screening, so the potential positives they identify can be checked out in more detail.

      • ..in today's world is that it will hear you cough at work. decide you have some disease you've never heard of and the next thing you know your workplace will be sealed off and you and your workmates will spend the next several days in quarantine being poked and prodded until they finally figure out that the "funny cough" that got flagged was just because a crumb had gone down the wrong way.
        • Workplace sealed off? That costs money.
          Medical care? That too.
          Firing a sick worker - saves money.

          Outside of your conspiracy-theory fantasies, in the real world, they will take the option that saves money.

          • Outside of your conspiracy-theory fantasies, in the real world, they will take the option that saves money.

            Whose conspiring here? It was a light hearted crack at the nanny state. Companies can't quarantine people, only governments can and if you think governments will always take the option that will save money then I think you must be coming down with something so be careful if you cough or the CDC may be coming by.

      • It'll be notifying HR ...

        No, I'll be using voice-changing AI to give me a sore throat and an obvious case of 24hour-itis.

    • by zekica ( 1953180 )
      AI (current AI) is only as good as the training data. It can somewhat reason but it doesn't have a deep understanding a good doctor would have. It would be equivalent to you spending a month or more with some doctor as an apprentice without them ever explaining what their reasoning is, and letting you get to the conclusions yourself.
      • What you're saying has nothing to do with this article. You don't need "the understanding of a good doctor" for what we're talking about.

        "Good doctors" will be very happy to just to have a way to know that "a bunch of people in this area are coughing", because that info is incredibly valuable when tracking the spread of diseases.

        • I doubt very much anything medical can be mass-monitored that way under HIPAA laws. But the sort of pattern recognition AI is good at does seem an excellent match for personal health warnings.
          • I doubt very much anything medical can be mass-monitored that way under HIPAA laws

            HIPAA has nothing to do with identifying pockets of diseases/infections/whatever. Any identifiable information can be completely stripped out leaving you with 'X number of people in this community contracted measles'. This does not indicate who has measles, only that the concentration is in that particular community, and how many.

            That's something any doctor might want to know and especially for the CDC who tracks o
          • by machineghost ( 622031 ) on Friday August 30, 2024 @02:57PM (#64749364)

            Obligatory, IANAL, but I've worked for a healthcare start-up, and I think you're misunderstanding HIPAA if you think it categorically prevents researchers from using mass data with personal origins. As I understand HIPAA, it's not that you can't start with personal data ... it's that you have to make sure it's depersonalized before it gets exposed.

            Again, IANAL ... but an app which collects info from individual users ... and stores it without any way to tie it back to those individuals ... is in compliance.

          • This is how people think HIPAA works. It's how politicians and hospitals speak as if it works. It's how it probably should work.

            But it is not how HIPAA works.

    • by MobileTatsu-NJG ( 946591 ) on Friday August 30, 2024 @12:25PM (#64748954)

      As far as I can tell, medical is the one field where A.I. may definitely benefit humanity...

      The AI heard you cough, no coverage for YOU!

    • Doctors: "over my dead body."

      • I am not sure there. Simple test, further examinations still needed, their job is safe
        My wife is a doctor. We joke about this when she has to do a weekend shift. She proposes I go in her place and tell every patient to come back if it is not better in 3 days. It probably is the best advice for 75% of the patients. A lot of people go to the doctor for reassurance. (Doctor is very cheap here, 10 euro something.) Let's not think about what will happen to the 25% others. Now if an AI could reassure people, we
        • Think of it as a diagnostic tool providing automated triage.

          If it could filter out the 75% needing a response of "come back if it is not better in 3 days" and tell the other 25% to report to the exam room today how useful would that be?

          • You underestimate the persistence of those 75%. But I guess AI could divert those to an AI that is excellent at reassuring people. But... it is not a very human thing to do. If AI takes over, what is the meaning of life? Do you still have a place in this world if AI outsmarts you in every way? Lots of fundamental questions.
  • Perhaps AI can find when you masturbate, too!
  • by Walt Dismal ( 534799 ) on Friday August 30, 2024 @02:27PM (#64749298)

    My new AI startup listens to the sound of your farts and evaluates whether you are a popular person.

    • My new AI startup listens to the sound of your farts and evaluates whether you are a popular person.

      I know your post is a joke, but if they could develop an AI that could either listen to or "smell" your farts as an alternative to colonoscopy to screen for colon cancer, that would make millions.

"How many teamsters does it take to screw in a light bulb?" "FIFTEEN!! YOU GOT A PROBLEM WITH THAT?"

Working...