the categories for modality-base (Image, Table, Image+Table) and patient-based scope (none, single, group)

[nooralahzadeh]

Hello,

Thanks for this great work. I wonder if any identifiers in the dataset indicate the categories for modality-base (Image, Table, Image+Table) and patient-based scope (none, single, group)?

Thanks

[baeseongsu]

Hi @nooralahzadeh ,

Currently, we do not support direct identifiers in the dataset. However, you can try using the code from these links:

1. https://github.com/baeseongsu/ehrxqa/blob/724bff13a9a2e430ecd54f32e3ef3789bb7fcdb3/tests/test_utils.py#L45-L67
2. https://github.com/baeseongsu/ehrxqa/blob/724bff13a9a2e430ecd54f32e3ef3789bb7fcdb3/tests/test_utils.py#L132-L148

You can filter the dataset using the scope argument. The possible values for the scope are:

• TABLE-NONE
• TABLE-SINGLE
• TABLE-GROUP
• IMAGE-SINGLE-1
• IMAGE-SINGLE-2
• IMAGE-SINGLE-N
• IMAGE-GROUP-N
• MULTIMODAL-SINGLE
• MULTIMODAL-GROUP

The scope argument uses string matching at the start, so you can also use broader categories like "IMAGE-SINGLE" or "IMAGE-GROUP" (or even "IMAGE") to include all relevant subcategories.

Sorry for the inconvenience.

Best, Seongsu

[nooralahzadeh]

Thanks.