search

liuqidong07 / LEADER-pytorch

[arXiv'24] The official implementation code of LEADER.
https://arxiv.org/abs/2402.02803
MIT License
10 stars 2 forks source link

Topk filtering for Diagnosis or Procedures #6

Open liyifo opened 1 week ago

liyifo commented 1 week ago

Hello, in data processing, you have filtered out diagnoses or procedures except for the top k, but have the drugs corresponding to the filtered diagnoses not been filtered out. Is there a problem with the processed data?

liuqidong07 commented 1 week ago

Hello, thanks for your interest in our work. Actually, we refer to the preprocessing code of GAMENet and SafeDrug. Could you please refer to the specific code in our repo? Best,

liyifo commented 6 days ago

Hello, thanks for your interest in our work. Actually, we refer to the preprocessing code of GAMENet and SafeDrug. Could you please refer to the specific code in our repo? Best,

Thank you for your reply. Here is the code I am referring to: SafeDrug/data/processing.py line 162 diag_pd = filter_2000_most_diag(diag_pd)

liuqidong07 commented 6 days ago

Hello. In my view, this line will filter out redundant medication, diagnoses, and procedures. https://github.com/liuqidong07/LEADER-pytorch/blob/75bac886bfce983b0fe6b8856e53f1abb13a436a/data/mimic3/utils.py#L236 You can check this function. Best,