What a big week on @watchhcn — @calleymeans on MAHA, @cremieuxrecueil on healthcare and data, plus the JAMA "can doctors make AI worse" debate and Lilly's $2.8B psychedelics bet.
Check out the full episode here!
AI will be doing 80-90% of what a doctor does today.
The truth is that doctors will become 10x more valuable over time because they will be orchestrating an army of agents, NPs, and PAs.
This team based use of physician time will make healthcare better, not worse.
It’s coming… whether we want it or not.
All doctors are proving with these “stories” are that doctors in the US need to do a better job.
NAFLD is one of the most under-diagnosed conditions in the US and changes clinical management of the patient. “Just lose weight” is awful doctoring.
Can’t wait for AI to improve care better than doctors who think they’re above patient autonomy.
I'm not anti AI (I use it every day)...but since I'm bedside with patients daily, I'll share how SOME patients are using LLMs and what I'm seeing with one visit to my office this past week.
This is why health care costs will probably go up short term before AI can bring them down.
Patient has labs come back with an ALT (liver enzymes) 5 points above upper limit of normal at his annual physical. His doc sends him a message most likely from overweight, to lose some weight.
The patient then goes online and an LLM tells them they might have livery injury, might be best to get labs and agrees with their doctor it might be due to weight.
Patient goes to another primary care office after his doctor refused to order a bunch of labs via just a portal message (looks like he requested a follow up visit if he wants to talk about them in much more detail beyond his message) and pt decides to talk to the new primary care doctor and the following was ordered after he tells an LLM recommended:
CMP
GGT
Hepatitis profile
Iron profile + Ferritin
ceruloplasmin
ANA
Antismooth muscle antibody
INR/PT/PTT
Then when all came back negative with same 5ish point elevation, the LLM program told him he needs an ultrasound. So back to his second primary care office with this where the doc does another visit and agrees to order it.
The ultrasound shows fatty infiltration.
So now the LLM convinces him he needs a fibro scan which came back similar, with mild fatty liver disease
He can't get into the second primary care doctor so comes to see me (third primary care office) where I reaffirm that yup...hes 100lbs overweight....nothing else compares to weight loss and we have a discussion about all the options.
When people are saying AI is going to decrease costs...help patients and doctors....this is just one example of a visit I had this past week with this type of patient who will go down the health anxiety rabbit hole.
This guy spent probably over $1,000 of his health insurance dollars on the additional workup and at least 3 extra doctor office visits.
Currently no LLM program is (IMO) willing to say something like, yes you are 100lbs overweight, strong probability it's just like your doctor said. Lose weight and let's recheck in the future and no additional labs needed at this time based on lack of symptoms or risk factors.
BCBS just for this one case probably spent over $1,500 on his care because he went down the LLM rabbit hole of what it might be.
Indeed. The same transition is reshaping education. I have spent years rebuilding systems around skills, not rote execution. The harder problem for medicine is rewriting curricula to train physicians for orchestration, not autonomous diagnosis.
NAFLD is one of the most under-diagnosed conditions in the US and changes clinical management of the patient. Just lose weight is awful doctoring.
All you’re proving with this “story” is that doctors in the US need to do a better job.
Can’t wait for AI to improve care better than doctors who think they’re above patient autonomy.
For folks who have enjoyed @watchhcn content… can you do me a giant favor? Subscribe to the channel on Youtube. They have a much better interface to browse old episodes and videos!
Also, please share a topic you want us to cover this week. Appreciate you all rooting for us!
https://t.co/1kZjsO8Z6o
AI will be doing 80-90% of what a doctor does today.
The truth is that doctors will become 10x more valuable over time because they will be orchestrating an army of agents, NPs, and PAs.
This team based use of physician time will make healthcare better, not worse.
It’s coming… whether we want it or not.
Interesting take. Same thing happened with dev tools. Automation gave engineers more leverage, didn't replace them. Healthcare follows the same logic, regulation just drags the timeline out.
I think psychedelics will be everywhere.
That’s both exciting and a problem.
For someone with PTSD or treatment-resistant depression, this can be a great treatment instead of being on meds forever.
But people are already microdosing with zero good evidence it works. These compounds affect your brain waves and I don’t think it’s safe to do recreationally.
We need to get the clinical use right before this turns into the next wellness trend that does more harm than good.