Skip to main content
. Author manuscript; available in PMC: 2024 Oct 24.
Published in final edited form as: Proc Conf. 2024 Jun;2024:7193–7210. doi: 10.18653/v1/2024.naacl-long.399

Figure 7:

Figure 7:

Macro-averaged risk prediction performance evaluated on synthetic labels and averaged over 5 random seeds for choosing the which time-point in the EHR to use prior to the diagnosis label. Error bars represent the standard deviation of the random seeds. Here, BERT and Longformer refer to Clinical BERT and Clinical Longformer.