li-lab-mcgill / mixehr Goto Github PK

Hello,

Thanks for writing such an interesting paper and publishing your code!

Please could you add a license to this code base?

Thanks in advance,

Harry

Thank you for the great and inspiring work.

However, to reproduce the results, I encounter one problem that is the EHR codes. I actually wonder how you construct the EHR code. For example, in your paper, you mentioned that "For clinical notes (NOTEEVENTS.csv), we use the R library tm by following a simple pipeline. We filtered out common stop words, punctuations, numbers, whitespace and converting all of the remaining words to lowercase." The paper reports a total of 33,388 words are constructed in the clinical notes. But following your pipeline, I got a dictionary that is much larger than 33,388. Can you release the total of 53,432 EHR codes? Thank you.