Hospital to test AI 'copilot' for doctors that jots notes on patient care
- Reference: 1679402288
- News link: https://www.theregister.co.uk/2023/03/21/abridge_hospital_ai_doctor/
- Source link:
The technology, developed by Pittsburgh, Pennsylvania startup Abridge, aims to reduce workloads for clinicians and improve care for patients. Shivdev Rao, the company's CEO and a cardiologist, told The Register doctors can spend hours writing up notes from their previous patient sessions outside their usual work schedules.
"That really adds up over time, and I think it has contributed in large part to this public health crisis that we have right now around doctors and nurses burning out and leaving the profession." Clinicians will often have to transcribe audio recordings or recall conversations from memory when writing their notes, she added.
[1]
"Just to give you a sense of what status quo can look like right now, I might see patients and then not document until the weekend, and then I'm using chicken scratch that I put on a piece of paper to remind myself what we talked about. And so it's very lossy – I'm losing a lot of the details in this document," Rao said.
Take two aspirin and call me in the morning
Abridge's software automatically generates summaries of medical conversations using AI and natural language processing algorithms. In a short demo, The Register pretended to be a mock patient talking to Rao about suffering from shortness of breath, diabetes, and drinking three bottles of wine every week. Abridge's software was able to note down things like symptoms, medicines recommended by the doctor, and actions the clinician should follow up on in future appointments.
The code works by listening out for keywords and classifying important information. "If I said take Metoprolol twice, an entity would be Metoprolol, and then twice a day would be an attribute. And if I said by mouth, that's another attribute. And we could do the same thing with the wine example. Wine would be an entity, and an attribute would be three bottles, and other attribute every night."
[2]
[3]
"We're creating a structured data dataset; [the software is] classifying everything that I said and you said into different categories of the conversation. But then once it's classified all the information, the last piece is generative."
At this point, Rao explained Abridge uses a transformer-based model to generate a document piecing together the classified information into short sentences under various subsections describing a patient's previous history of illness, future plans or actions to take.
[4]
Screenshot of Abridge's app following the pretend medical conversation
Big Tech is desperately finding ways to incorporate generative AI into their services with a wave of startups racing to build products on commercial APIs like OpenAI's ChatGPT. However, these models are known to invent false information and can struggle to produce accurate content, making applications in industries like healthcare or law particularly risky.
Physicians can edit the notes further, whilst patients can access them in an app. Rao likened Abridge's technology to a copilot, and was keen to emphasize that doctors remain in charge, and should check and edit the generated notes if necessary. Both patients and doctors also have access to recordings of their meetings, and can click on specific keywords to have the software play back parts of the audio when the specific word was uttered during their conversation.
[5]
"We're going all the way from the summary we put in front of users and we're tracing it back to the ground truth of the conversation. And so if I have a conversation, and I couldn't recall something happening, I can always double-check that this wasn't a hallucination. There are models in between that are making sure to not expose something that was not discussed."
[6]Study: While text-generating AI can write like humans, it lacks common sense
[7]We know you all want to shove AI where the sun doesn't shine. And that's exactly where it's going – detecting prostate cancer
[8]Medic! Uncle Sam warns hospitals not to use outdated IPnet freely on their networks
[9]Well Holby damned! We've caught a virus: Brit medical soap operas team up for 'cyber' episode
Abridge's software has reportedly been trialed by over 2,000 clinicians and to help more than 200,000 patients. Now, the company is going to test it out in a real hospital starting with a small cohort of doctors.
Gregory Ator, Chief Medical Information Officer and Head and Neck Surgeon at The University of Kansas Health System, said he expected the trial to start soon.
"We need to figure out who are the ideal clinicians for this type of solution. First of all, we're going to validate that the datasets used [to train the software] are accurate and appropriate to Midwest healthcare, and deal with the type of problems that we deal with, and then we will progressively roll it out," he told The Register . ®
Get our [10]Tech Resources
[1] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2ZBnisVF-R0yxViKbCfFEtQAAAAQ&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZBnisVF-R0yxViKbCfFEtQAAAAQ&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZBnisVF-R0yxViKbCfFEtQAAAAQ&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[4] https://regmedia.co.uk/2023/03/20/abridge_demo_conversation.jpg
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZBnisVF-R0yxViKbCfFEtQAAAAQ&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[6] https://www.theregister.com/2020/11/20/machine_learning_language/
[7] https://www.theregister.com/2019/04/17/ai_prostate_cancer/
[8] https://www.theregister.com/2019/10/02/fda_ipnet_alert/
[9] https://www.theregister.com/2019/02/15/holby_casualty_brit_medical_soaps/
[10] https://whitepapers.theregister.com/
Re: Why AI???
[1]Computer medicine has been documented for a long time now ....
[1] https://youtu.be/bn8rpBQSs-o
Re: Why AI???
I'm guessing that the principle is that an "AI" could get rid of a lot of "noise" - not actual noise, I mean all those bits of conversation that don't actually carry useful information. Plenty of those.
Of course, anyone who has used these "AI" systems knows that they make mistakes. Depending on the context, the rate of mistakes is sometimes very low, and sometimes very high. Worse, their mistakes are sometimes quite hard to catch.
On the other hand, though, doctors also make mistakes while doing this type of work. This is especially true if they are overworked, and, let's face it, most of them are, shamefully so.
So the big question is - is the "AI" going to make more and/or worse mistakes than a doctor?
I don't think the answer is obvious at all. There are fields where "AI" systems make loads of enormous mistakes, and others where they work about as well as a human or even somewhat better. I have no idea what category this falls in.
It's good that trials are being made. I just hope the trial results are taken seriously, and not "creatively interpreted" to make the "AI" look better than it is.
Re: Why AI???
"I'm guessing that the principle is that an "AI" could get rid of a lot of 'noise' ... all those bits of conversation that don't actually carry useful information. Plenty of those."
Amen to that. My side gig is reading doctor's full reports and summarizing for other doctors (and eventually lawyers - workers' compensation and similar lawsuits). Some doctors can WAY too wordy, and we paraphrase, but in a way that still captures the critical/core information/facts.
As for doctors making their own mistakes -- there is plenty of that also, which we have to fix on the fly. I blame doctors using text-to-speech software already (dictation, a time-honored tradition in medicine) without a human proofreader. AI might help this, but not 100%, so my job is more-or-less safe.
AI interpretation of surgeon & patient conversation
Surgeon: We will cut off your diseased foot, the left one ?
Patient: right.
Which foot will the AI summarise needs to be cut off ?
Re: AI interpretation of surgeon & patient conversation
If that exchange is the source of authority of the information on what foot needs to be cut off, then someone has made a grave mistake, long ago, that neither the AI nor the doctor can fix. Probably back when hospital procedures were devised.
Re: AI interpretation of surgeon & patient conversation
My thoughts exactly.
One has only to read the automatically-generated 'Closed Captions' on YouTube vids to see the misinterpretations made. These don't generally matter to me as they are the English Subtitles over English audio, and I can spot the speech-to-text mistakes and adjust accordingly. However, the written word on it's own would have led to ...confusion. Some of the 'hiccups' don't matter, but some do - there have been a couple I saw that meant the written was most definitely NOT what was spoken. Sorry, I can't quote refs...it was quite a while ago.
Leaving aside the obvious mistakes/wrong words in the machine translation, what about punctuation? A 'Let's eat, Grandpa! vs. Let's eat Grandpa!' issue could be quite serious in the medical world.....and the legal come to think of it.
Also, the NLP/speech-to-text has problems with accents.... Sure, en-something may be the language, but any accent can confuse the beast, it seems - the more pronounced the accent, the higher the error rate. And that's with 'broadcast-quality' or decent audio input, not the muffled low quality crap you're likely to get as source from whatever low-tech, cheap, shit device is used to record the audio in the first place.
Until the 'error-rate' improves, especially with accented speech or a non-native speaker (in ANY language), I for one dread my medical records getting transcribed by AI. This seems like a baaaaaddddddd idea and pretty shitty science, 'pushed' in the name of progress,
Yup, NOT a fan of 'AI', because - just like Tesla's Autopilot - it claims to be something it ain't.
It might work ...
As long as it doesn't start insisting it loves the doctor or the patient.
"three bottles of wine every week .... Wine would be an entity, and an attribute would be three bottles, and other attribute every night ."
If that's how it works the time saved in taking notes will be spent in fixing the results.
A GP prescribing 3 bottles of wine a night? Mine told me to stop that sort of thing!
Why AI???
Surely all this needs is a speech-to-text engine to write down what the doctor actually said, instead of trying to interpret what they said. Isn't that just as bad as trying to understand the written notes in whatever bad handwriting they have. Overkill, in more ways than one?