Using Machine Learning to Predict a Patient's Prognosis

Having a constant feedback at the beginning is really important. We will commonly use things like mortality or death as an end point because it's just a one or zero in the model. The issue is that often times death isn't the right outcome that matters. And then, you know, once we get the right features and outcomes, i think we can really trust a lot of the data scientists to come up with the best model.