no code implementations • 28 Jun 2021 • Matthew Barren, Milos Hauskrecht
More specifically, our method improves prediction performance of a low-prior task by jointly optimizing a shared model that combines the loss of the target event and a broad range of generic clinical events.