dataset

[flying-lby]

Hello, I am deeply inspired by your work.
I would like to ask some questions about the dataset mimic-cxr-label-LLM_report-xinhuo-chexpertformat. How were the disease labels such as Atelectasis, Cardiomegaly, Consolidation, Edema, etc., generated? What do the values -1, 0, and 1 represent? Are these values used in the code?

Looking forward to your reply, I would appreciate it.

[laihaoran]

Thank you for your response. As mentioned in https://physionet.org/content/mimic-cxr/1.0.0/, disease labels such as Atelectasis, Cardiomegaly, Consolidation, Edema, etc., were extracted from the reports using the CheXpert labeler. In this context, 1 indicates the presence of the disease, 0 indicates its absence, and -1 indicates uncertainty. In our training, we did not use these values (1, 0, -1) as we are using zero-shot learning.