Something a bit worrying to note about using Ai in healthcare.

I’ve had two specialist appointments recently, both using ai to transcribe. Both sent report letters with inaccuracies about my diagnoses and past medical history. Even my GP was like, “huh, that directly contradicts what I put in the referrals.”

I have followed up both and requested amendments (which were done) but if I hadn’t, these inaccuracies could have significantly damaged ongoing care, further treatment or insurance claims.

Human error has always been a factor, but both doctors were clearly using the ai software and assuming what it spat out was correct. They made no other notes during the appointments to cross-reference and double check. This is how Very Bad Things can happen.

Two friends have told me in the last week they had similar issues happen. One had an incorrect diagnosis listed before they had a procedure done. The other noted viral not bacterial infection (although they did at least get the medication they needed). I feel like I’m being a pain in the bum going over everything and requesting corrections, but I’m seeing so many mistakes, to the point where any human reading them would immediately say “that doesn’t even make sense”. I worry for those who don’t or aren’t capable of checking these things. Sure, using ai might save the docs 10 minutes per patient in the ER but is that really worth the risks?
@bloodflowersburning Where are doctors using ai like this? And how did you check the transcribes?🤔
@theron29 Genuine question or scepticism? I’m in Aotearoa NZ. Doctors were from two different specialist medical departments. Both used ai software to record consultation and take notes. The report letters sent to my GP contained multiple discrepancies about conditions discussed, and referenced in GP referrals. If they had checked before sending they would have realised mistakes had been made. My GP questioned the content, which was how I became aware. I can provide several specific examples but would rather not on a public forum to a stranger. However, both letters were re-assessed and sent again with corrections on request. Hope this helps.

@bloodflowersburning Genuine question (from central EU). (AI Scepticism is expected to come a bit later on... 🙂 ).
Doctors are not using AI here, yet. I guess this tech had to be certified and tested before it was admitted into doctor's realm?

Although this does not seem to be the worst usecase scenario where&how to use AI, your detailed explanation gives doubts whether the tech is actually ready now for anything *this* important.... 😏

@theron29 Agreed. Not the worst case scenario. For me personally it could have caused issues with further treatment, getting reimbursed by insurance, and caused confusion when needing ongoing care with other providers. So more an avoidable inconvenience and extra paperwork rather than a dangerous outcome in this example. I hope that’s the worst possibility across the board, and that people check their notes carefully to catch any inconsistencies.

Mistakes in medical notes have always happened, unfortunately it’s inevitable. Only time will tell if this becomes more of an issue if/when ai transcription is used in medical settings more frequently and if it generates a higher number of errors as opposed to human note taking. What I think is essential is that we still retain a human buffer to assess factual accuracy, rather than simply assuming (hoping?) the software can do it better.

For more info, the software Heidi AI Scribes has been endorsed for use within Health NZ. https://www.tewhatuora.govt.nz/health-services-and-programmes/digital-health/generative-ai-and-large-language-models#naiaeag-endorsed-tools

Generative AI and Large Language Models - Health New Zealand | Te Whatu Ora

Building the future of health

@bloodflowersburning @theron29

As an AI proponent, generally speaking - even I find this disturbing. Who issues the ICD code for any incident? It had better not have been the machine.

@tuban_muzuru
This is more a transcribe and summarise tool than a make decisions tool—at present. The risks I can see here are if decisions are made based on the incorrect information, or in my case, I wouldn’t have been able to pursue insurance coverage/compensation as the letter stated incorrect diagnosis.

If we move on to machines making medical decisions, without human input, I think we’re stuffed.

@theron29

@bloodflowersburning @theron29

Well sure - yet more chalk on the board for trying to shove AI into a totally inappropriate application.

@theron29
Not true. Doktolib provides this fearure and doctors are using it to do transcripts of their consultations. (I know this to be true for Germany, can't say anything about other countries in the EU.)
@bloodflowersburning