Peter Attia

Yeah.

What were the outcomes?

CVT type outcomes.

So biomarkers, obviously.

Let's talk a bit about the epidemiology in this space.

So I think everybody listening to this podcast knows what epidemiology is, and we've talked a lot about the limitations of it and what a healthy user bias is.

But give us the landscape of how epidemiology has looked specifically at this question of the relationship between protein intake and outcomes of health.

What are some of the near unique or particular circumstances of epidemiology that lend itself to confusion here?

What do you think are the three most important biases that impact this particular question when asked through an epidemiologic lens?

And why do you think the editors at journals are unable to address that in the review process?

Where is AI in this process?

I mean, why are we not, or are we using LLMs to serve as peer editors?

And do we know if these AI agents, these peer review agents, I'll call them, are being trained on known fraudulent manuscripts?

Because we certainly have an abundance of things that were demonstrated to be frauds.

So it would be, I assume, a reasonable thing to do to start training these AI agents on that to start identifying the patterns?

Again, the answer is yes, but very much in its infancy.

Okay.

Who's leading the charge on this?

What do you think is the most compelling piece of epidemiologic data against the idea of exceeding the RDA?

Let's flip the question now again, which is, okay, I'm going to argue that lower protein is better because I've just demonstrated you are not going to starve to death at, let's just round up and call it one gram per kilogram of body weight.