Dr. Faust, This is great positive information. Thanks. We all hope AI does what we all hope it will do. Hopefully it will make complexity less complex for those creating information and easier for those trying to understand that information.
I’m a telehealth provider (no brick and mortar practice anymore) with a part time “post-retirement” job. The company I work for uses the Ambience generative AI platform. It is an amazing and game-changing technology. I agree with you, Dr Faust!
I've been using Heidi AI Resident which really takes this concept and pushes it forward even further.
Not only does it take the recording and create the clinical note, but there is a generative function where you can ask it to create notes, out of work notes, matters of medical necessity based only upon the information from the conversation that you had with the patient.
This has made things like letters for emotional support, animals, school, out of work notes, reasonable accommodation notes, and patient summaries literally a sentence away.
The coolest part is you can actually modify and change the template that you want to use for any particular note. They are actively upgrading to have a community-based collection of templates and are extremely responsive.
Hipaa compliant too! You should check it out!! 600 yearly USD
Oooh yeah I want templates. That’s one thing I omitted from the article because it was getting long. My Epic macros/templates/shortcuts are not yet integrated. I’d love to see that.
This is great! Did you find many patients were reluctant to be recorded? There are so many people who worry about their privacy being violated. Of course, we know that it is no more likely that their HIPAA will be violated with AI than with docs charting, there is something inherently suspicious about adding an APP into the equation?
So far none expressed concerns. I decided not to ask a couple of patients due to sensitive nature of their issues, but ultimately I should probably ask those patients too. It’s all the same security!
I'm curious as to the effect on your clinical reasoning. I always found (I'm retired now) that creating and reviewing notes, although tedious, was an exercise in quality control for my reasoning. Errors of omission, incomplete ddx, inadequately supported hypotheses etc tended to jump out when subjected to that process of dictating/reviewing or typing. Does the AI solution change this?
That is an excellent point and I identify with exactly what you’ve said. In many ways, writing indeed is the process by which I discover what I believe, whether it is here writing Inside Medicine or charting.
It’s possible that this could be a problem for me with AI charting and so I’ll try to keep an eye on it. I think the editing process is enough to be a check on things but I will report back!
A few years back I had an experience that has me questioning depending on AI when it comes to medical charting. The result of my mammography came to me reporting a dramatic change in my breast density and upping the chance of breast cancer over time. I immediately called my doctor to ask if such a change was even possible and he told me no and that he would look into it. The next day he called saying it had taken him the better part of a day to get a hold of the radiologist who had dictated their notes but had not looked them over after the fact. For some reason it heard "dense" when that is not what the doctor said. The result could not be deleted but a notation was made and has been made on every subsequent result. AI "hears" the doctor but if Nuance Dragon or whatever software program that interprets speech could possibly get it wrong doesn't the doctor still have to take the time to carefully screen the report?
Wonder how long it will take for the legal folks to become the dominant force that “shapes” exactly how the AI program translates/organizes the recorded conversation into a medical record.
Sounds cool, but seems like a slippery slope. Next step: AI robot interviews the patient, interprets the history and generates differential diagnosis, orders lab tests and EKG and then interprets them, orders appropriate meds, notifies cardiologist of admission. You’re out of a job.
Dr. Faust, This is great positive information. Thanks. We all hope AI does what we all hope it will do. Hopefully it will make complexity less complex for those creating information and easier for those trying to understand that information.
That’s the dream. Maybe I’ll make a video demonstration at some point (with an actor playing a fake patient!)
I’m a telehealth provider (no brick and mortar practice anymore) with a part time “post-retirement” job. The company I work for uses the Ambience generative AI platform. It is an amazing and game-changing technology. I agree with you, Dr Faust!
Glad to hear it🙌
Fabulous! And another victory for AI last week, as it was used to suggest structures of new antibiotics for resistant bugs
That is a neat application!
The clearest and best business case example I’ve seen yet for AI. Wow, now I’m excited! Thanks, Dr. Faust.
Right?!! Someone said “they’ll just make you see more patients.” I don’t mind that, though! It’s the damn charting!
I've been using Heidi AI Resident which really takes this concept and pushes it forward even further.
Not only does it take the recording and create the clinical note, but there is a generative function where you can ask it to create notes, out of work notes, matters of medical necessity based only upon the information from the conversation that you had with the patient.
This has made things like letters for emotional support, animals, school, out of work notes, reasonable accommodation notes, and patient summaries literally a sentence away.
The coolest part is you can actually modify and change the template that you want to use for any particular note. They are actively upgrading to have a community-based collection of templates and are extremely responsive.
Hipaa compliant too! You should check it out!! 600 yearly USD
Oooh yeah I want templates. That’s one thing I omitted from the article because it was getting long. My Epic macros/templates/shortcuts are not yet integrated. I’d love to see that.
This is great! Did you find many patients were reluctant to be recorded? There are so many people who worry about their privacy being violated. Of course, we know that it is no more likely that their HIPAA will be violated with AI than with docs charting, there is something inherently suspicious about adding an APP into the equation?
So far none expressed concerns. I decided not to ask a couple of patients due to sensitive nature of their issues, but ultimately I should probably ask those patients too. It’s all the same security!
I'm curious as to the effect on your clinical reasoning. I always found (I'm retired now) that creating and reviewing notes, although tedious, was an exercise in quality control for my reasoning. Errors of omission, incomplete ddx, inadequately supported hypotheses etc tended to jump out when subjected to that process of dictating/reviewing or typing. Does the AI solution change this?
That is an excellent point and I identify with exactly what you’ve said. In many ways, writing indeed is the process by which I discover what I believe, whether it is here writing Inside Medicine or charting.
It’s possible that this could be a problem for me with AI charting and so I’ll try to keep an eye on it. I think the editing process is enough to be a check on things but I will report back!
Is this completely independent of Epic? Like is it platform agnostic and can be used for Cerner?
Yeah in fact the only downside is that I have to copy and paste it into epic from the web-based app
A few years back I had an experience that has me questioning depending on AI when it comes to medical charting. The result of my mammography came to me reporting a dramatic change in my breast density and upping the chance of breast cancer over time. I immediately called my doctor to ask if such a change was even possible and he told me no and that he would look into it. The next day he called saying it had taken him the better part of a day to get a hold of the radiologist who had dictated their notes but had not looked them over after the fact. For some reason it heard "dense" when that is not what the doctor said. The result could not be deleted but a notation was made and has been made on every subsequent result. AI "hears" the doctor but if Nuance Dragon or whatever software program that interprets speech could possibly get it wrong doesn't the doctor still have to take the time to carefully screen the report?
It's a pretty neat system, and it has a free mode. Not sure how it compares to your solution but worth checking out!
Wonder how long it will take for the legal folks to become the dominant force that “shapes” exactly how the AI program translates/organizes the recorded conversation into a medical record.
Yeah there’s always that concern. For now, it’s bliss. Long may it last.
Sounds cool, but seems like a slippery slope. Next step: AI robot interviews the patient, interprets the history and generates differential diagnosis, orders lab tests and EKG and then interprets them, orders appropriate meds, notifies cardiologist of admission. You’re out of a job.
I still think the human connection is desired by patients. But, we’ll see!
I certainly hope you’re right!