This sure looks like a product expo to me.

Haha bonus points for the owl

I already like it

It’s not even a formula! It’s just the definition of a derivative

Oooh now I need to know the joke 😂

If I ever write an r package for DAGs, I call dibs on dagnabbit

Ooh this is very nice!

Manchester Indian food is top notch. I recommend eating it until you nearly pass out, then walking a few miles to a football match and chugging two Fosters before you take your seat

Ha! For sure. We’re trying to understand the factors that clinicians use when they intuitively conclude a pediatric patient is likely to come back to the hospital. spoiler: it’s not stuff that’s in the EHR anyway

Nice! We're undertaking something like this now for pediatric asthma readmission. Most of the clinicians I've spoken to don't even pay attention to what the EPIC model tells them re: chance of readmission. We want to know why and whether we can make it better.

It’s become ridiculous at this point. At least with Microsoft buying into OpenAI, you know it’ll slowly get even crappier

I know! I wish it were true. What it will do, however, is give you a more realistic picture of what the data offer, even though the coefficient estimates will be slightly biased

No, LASSO can be thought of as variable selection and regularization in one. It's optimized to make calibrated predictions, but isn't going to rescue any causal claims