I’m glad this is working for you! No judgement from me, but my brain works better as the constructor of narratives, rather than outsourcing this to AI. Though I reserve the right to capitulate and change my human mind in the future, I wrote this contrarian OpEd recently (ironically) for Doximity:
A no brainer. In the past, have you tried voice-to-text apps? I’ve had an appointment where the doctor spoke his notes the record. If so, how does this compare to the ai tool. Also 59 pages is a bit much. I like to have gpt turn text into a bullet point summary.
Second comment - from an engineer, not a doctor....
I use AI a lot for research. It is fast. It is all knowing in limited spheres. And it has little common sense. Physicians will have to learn to provide the history and context in their observations and feedback to patients. It is not difficult to comprehensively review the bidding at the end of the game but it requires a new skill and self discipline. And the patient will be much better informed for it.
Ten years ago my cardiologist in Portland, Maine started using a flesh and blood scribe. I immediately commented to him how much I appreciated being looked in the eye as we talked. In that visit he opened his computer once to check a lab value. After the visit the scribe stayed in the room to finish up needed paperwork, tests and appointments. I told her how much I appreciated her presence and asked her how he could afford her. She said that the increase in his productivity more than paid for her position. He retired a few years later and I have not seen a scribe anywhere since.
The comment about your son reinforces my opinion that doctors are often over skilled for what they are doing. This is true of many professions but there is a shortage of doctors and quicker more specific training might get more people treated.
The son was acting as a scribe, not a doctor. As for generalist docs vs. specialists, don’t you start with your gp and then see a specialist based on their recommendation?
I don't have the link handy, but I recently read that some medical system was using an AI scribe, and the scribe frequently hallucinated. I can see that this problem may go away in the future, but today it's a real concern.
It was mentioned in the Kaiser article as rare but something they were keeping an eye on.
Bottom line is that the AI will make errors and it is the provider's responsibility to review the AI note and correct errors before it goes into the medical record. But it is a lot easier to review the AI scribe note and correct minor errors (or hallucinations) than to create one from scratch and memory or notes.
I would love to see the difference an AI scribe would make in visits with my docs. I have a few complexities that would probably get better organized with an AI scribe.
My experience so far suggests that the AI scribe is significantly weaker than me in organizing the problems but unless a specific diagnosis is described during the visit, it is not going to know the patient has it. It does not have access to old notes or other parts of our EMR. For example, during a recent visit, the patient described shortness of breath and exercise intolerance. I had in prior visits determined that the cause of these symptoms were multifactorial and included HFpEF, deconditioning, etc. AI scribe didn't know why I started a diuretic.
I’m glad this is working for you! No judgement from me, but my brain works better as the constructor of narratives, rather than outsourcing this to AI. Though I reserve the right to capitulate and change my human mind in the future, I wrote this contrarian OpEd recently (ironically) for Doximity:
https://opmed.doximity.com/articles/we-need-human-physicians-not-ai-to-help-us-make-sense-of-chaos
A no brainer. In the past, have you tried voice-to-text apps? I’ve had an appointment where the doctor spoke his notes the record. If so, how does this compare to the ai tool. Also 59 pages is a bit much. I like to have gpt turn text into a bullet point summary.
How prevalent has the problem of AI hallucination been in your experience thus far?
How much time do you spend proofreading the AI’s work?
Second comment - from an engineer, not a doctor....
I use AI a lot for research. It is fast. It is all knowing in limited spheres. And it has little common sense. Physicians will have to learn to provide the history and context in their observations and feedback to patients. It is not difficult to comprehensively review the bidding at the end of the game but it requires a new skill and self discipline. And the patient will be much better informed for it.
Ten years ago my cardiologist in Portland, Maine started using a flesh and blood scribe. I immediately commented to him how much I appreciated being looked in the eye as we talked. In that visit he opened his computer once to check a lab value. After the visit the scribe stayed in the room to finish up needed paperwork, tests and appointments. I told her how much I appreciated her presence and asked her how he could afford her. She said that the increase in his productivity more than paid for her position. He retired a few years later and I have not seen a scribe anywhere since.
I have that Brian Eno album. :)
The comment about your son reinforces my opinion that doctors are often over skilled for what they are doing. This is true of many professions but there is a shortage of doctors and quicker more specific training might get more people treated.
The son was acting as a scribe, not a doctor. As for generalist docs vs. specialists, don’t you start with your gp and then see a specialist based on their recommendation?
Flying to PR this weekend. Will check out your music recs.
I remember your son from one of my appointments.
Does Dr. Oz qualify as AI?
Hopefully Dr. Oz is not using AI to search the internet. If so, I will have to redact his name from several of my earlier posts.
It was remarkable how many patients for many years asked me about my son after that brief time he spent as my scribe.
Oops , simple Google search of dr. oz and skeptical cardiologist easily brought this up
https://theskepticalcardiologist.com/2019/08/24/what-is-a-cardiologist/
wherein I call him "the prince of quackery"
I don't have the link handy, but I recently read that some medical system was using an AI scribe, and the scribe frequently hallucinated. I can see that this problem may go away in the future, but today it's a real concern.
It was mentioned in the Kaiser article as rare but something they were keeping an eye on.
Bottom line is that the AI will make errors and it is the provider's responsibility to review the AI note and correct errors before it goes into the medical record. But it is a lot easier to review the AI scribe note and correct minor errors (or hallucinations) than to create one from scratch and memory or notes.
I would love to see the difference an AI scribe would make in visits with my docs. I have a few complexities that would probably get better organized with an AI scribe.
The importance of history cannot be underestimated!
My experience so far suggests that the AI scribe is significantly weaker than me in organizing the problems but unless a specific diagnosis is described during the visit, it is not going to know the patient has it. It does not have access to old notes or other parts of our EMR. For example, during a recent visit, the patient described shortness of breath and exercise intolerance. I had in prior visits determined that the cause of these symptoms were multifactorial and included HFpEF, deconditioning, etc. AI scribe didn't know why I started a diuretic.