Sometimes it’s fun to browse the YCombinator website, and catch a glimpse of the future dystopia
70 whole pages? Of medical record?? And it’s “messy???”
Maybe I missed the startup that uses LLMs to compose a précis of your medical history that’s likely to be approved for treatment by the LLM that inevitably is asked to read it
If we can get to denying the treatment *before* someone produces 70 pages of records that would be even better.
@CartyBoston I can build you an LLM that will deny treatment with *zero* pages of medical record
It’s pre-A round, who wan
@CartyBoston I mean, seriously, I’m not trying to gatekeep or do the credentialism thing here but two of the founders hail from Deepmind and “I dropped out of med school” and none of this is at all anything resembling a good look
@arthegall Not really sure how you missed the LLM that then disputes the rejection.
@marcua the future is a sequence of LLMs arguing each other while human civilization slowly crumbles around them
@arthegall think of the marketplace opportunities
@marcua the moment that someone presents an LLM as a “stakeholder” in an internal company project, I’m gonna retire
@arthegall “Optimism Health. We wrote an algorithm that approves any treatment a doctor says is needed.”
@arthegall @hannu_ikonen “We use LLMs to analyze long (70+ page), messy medical records to determine why a patient is not eligible for a treatment.’
</fify>
Fairway Health - Process prior
authorization faster
We use LLMs to analyze long (70+ page),
messy medical records and determine if a
patient is eligible for a treatment.
@arthegall It's actually just one line of code: printf( "denied \n" );
@arthegall
Why do they need an llm when return false is probably enough for the MVP?
@baldur