Dr. Michael Gao

Okay.

And similarly, on the clinical side, to use that blood pressure example of sort of 180 over 100, that must be high blood pressure, right?

Well, not necessarily.

What if you just broke your arm and you're in the ER and you're in lots of pain?

I guarantee your blood pressure will be high.

White coat.

Right.

But it's not high blood pressure.

And so kind of very similarly understanding the context before and after a specific measurement is actually really key to interpreting the clinical data.

And so the way that transformers work is that instead of just searching for the letters B, A, and K,

it's converting all the words using the words before and after it or the sequences before and after it doesn't have to be words it can be numbers um into you know kind of a digital representation that that you can actually map to things like a physical bank versus a promise you can bank on versus a river bank

And that context also helped us understand, pick out specific abnormalities in clinical data that were sort of not otherwise explained by the context around that value.

Yeah.

So, you know, first I left the hospital, went home and, you know, told my then fiance that she thought she was marrying a doctor.

But guess what?

You're an entrepreneur now.

I'm now an entrepreneur slash unemployed.

And but luckily.

um, you know, was able to be fairly convincing, not just for my now wife, but for, you know, kind of other people who joined along the journey.

And so I think actually the first thing to build was, um, really, um, you know, kind of a data science team, um, because, you know, I'm, I'm good at data science for a doctor.