Lexicomp

Because they work by predicting the next word that should logically follow the last words used, they cannot be expected to determine whether a drug interaction is present with 100% accuracy.

The weakness of using Lexicomp to check for drug interactions is the time it takes to put a medication list into a form the program can check by selecting one medication at a time from a drop-down list.

Once the list is in, Lexicomp software will always perform the correct check for interactions 100% of the time.

Large language model strength is language, meaning that a messy list with some misspellings and missing spaces can easily be copied or dictated and adjusted by the LLM and parsed correctly into a format that Lexicomp software can then run a check on.

So in my opinion, a better way to test a large language model would involve having the AI parse the list into LexiComp's preferred format, then compare the speed and accuracy of a pharmacist checking drug interactions manually with LexiComp with one leveraging the LLM to quickly get the list into LexiComp.

This way, the pharmacist stays in the loop performing the medication management task while the AI model serves as leverage by taking on the tedious work of entering a medication list into the software.

Right.

Because once you know what they are taking, you have to run it through reference databases like Lexicomp or Apocrates.

We have to check for interactions.

So let's talk about the absolute worst case scenario.

That really hit home to me about the lack of data we have on dosing in our patients.

And I don't think that most of us as clinicians are aware that the data we're looking up on Lexicon might be based on so little information or even abstracted from adult data.

Yeah, I'm happy to start.

So I agree.