[burke]

Anyone doubting this really has no idea what they're talking about.

Set up a "Health & Fitness" project in Claude (or whatever). Feed it:

* Basic data: height, weight, age, sex

* Basic metric snapshots from Apple Health or whatever: HRV range, RHR, typical sleep structure - go through everything and summarize it

* Typical diet (do you track it in MFP or Cronometer? Great, upload a nutrition report)

* Any supplements and medications you take

* Typical exercise habits

* Any health records you have - bloodwork results, interpreted imaging results, etc.

* Family history like you would describe it to a doctor

* Summary of any health complaints

* Anything else that seems relevant.

Then go through a few conversation loops asking it if there's any more information you could provide that would help it be more useful.

Then ask it things like "Given <health complaint>, what should I be doing more of? Less of?"—or "Please speculate about potential causes of <thing>".

Or, even if you don't have any particular health complaints you're working with, just being able to ask it questions like "What's one supplement I should consider starting or stopping today?" (and then obviously do some follow-up research...)

This is life-changing. Anyone skeptical of this has not tried it.

[LeafItAlone]

What are examples of things that Claude has told you to do?

[burke]

Troubleshooting some chronic inflammation issues (plausibly, like OP, an autoimmune issue)

It's suggested a few supplements that have helped a lot, helped me figure out dosing and timing, pointed me towards taking gut inflammation more seriously as a part of what's going on (and suggesting various tests and experiments to help prove/disprove that), explained correlations in various bloodwork results, the list goes on.

It's—of course—not perfectly trustworthy but a lot of things are either trivially verifiable or are low-risk experiments.

[reportgunner]

How do you know it's not barnum-forer effect ?

[burke]

Various suggestions that have improved objective outcomes.

Also, perhaps more often useful, is just "hey, tell me about ___ in the context of whatever I've told you about me" kind of thing".

Sometimes, yeah, the suggestions are very generically applicable and it's just a tool for thought.