Extended Data Table 1.

Detailed statistics of datasets

We built a comprehensive pretraining dataset (NYU Notes) with two site-specific variants (NYU Notes - Manhattan/Brooklyn) as discussed in the Methods section. For readmission prediction, we also built a finetuning dataset (NYU Readmission) with two site-specific variants (NYU Readmission Manhattan/Brooklyn), one structured-data variant (NYU Readmission - LACE), and a deployment test set (NYU Readmission - Deployment) that was sampled in real-time as part of our prospective trial. To test the breadth of NYUTron’s applicability, we added 4 tasks (NYU Mortality, NYU Binned LOS, NYU Comorbidity, NYU Insurance denial) with their respective structured-data variant (NYU Mortality - SAPS2+APACHE2, NYU Binned LOS - Lisbon Portugal, NYU Insurance Denial - Claim forms). NYU Comorbidity has no structured-data variant because the task is to impute comorbidity index with the lack of structured icd codes. Finally, we have a Named Entity Recognition (NER) dataset for testing how well NYUTron generalizes to different clinical predictive tasks using non-NYU data.