Why stop there? Want to get the medical literature to say what you want?
Here's 3 steps for how to make it happen:
1. Choose an evidence base with enough data sources that it's impractical for humans to read all of it, so you can't be double-checked easily
2. Put together a "systematic review" of the subject, using LLM's in both the screening step and the data extraction step
3. Fine tune your prompts for both the inclusion criteria and your endpoint of interest to get the result you want
