who are a bit skeptical about this approach of looking for biomarkers with metabolomics, what is your rationale to push this study specifically?

Yeah, I think so.

This study is fairly large, and we were also able to look at sex and race differences.

So on the computational side, there were many different models tried, training, testing, and repeating that thousands of times, making sure that things were stable and This was largely driven by my PI, Kurt Harris, because I remember coming into him the first time and telling him, hey, I have a list of 500 features that are predictive.

I was super excited, super proud of myself.

I was like, you know what, they're stable, I believe in blah, blah, blah.

He's like, yeah, we're not validating 500, get it down to 10 or less.

a lot of data.