Input data format

[rawan-qu]

Hello,
Great work!
Can you please share the format/structure of the pickle files which include the input data being loaded in dataloader.py?
Thanks.

[jackzhu727]

A synthetic data generator has been added. You can generate a dataset the same as our real-world EHR data.

Thanks for your advice!

[cswpy]

Hi, I am following up on this issue. The synthetic data generator in the repo is rather difficult to comprehend. While I am sure the generated data fits the input format, I have no way of knowing the (clinical) meaning of the columns. I am trying to understand it because I want to use the model for my own data, so it is critical for me to understand the input data format. It would be greatly appreciated if you could specify what the input data should look like. Thanks!

[jackzhu727]

Please refer to our experiment on MIMIC and eICU. They are publicly available.
https://github.com/NYUMedML/GNN_for_EHR/tree/master/data/