melinda315 / 4sdrug Goto Github PK

I am very interested in this article and I am greatly inspired after reading it. I git clone the code you posted on github and got the desired result after running it, thank you again for your open source spirit. After reading your code carefully, I found that some data files(like records_final.pkl voc_final.pkl ehr_adj_final.pkl ) in the dataset are consistent with the SafeDrug model in format, but the specific content of the data is quite different. In order to carry out further research, I would like to know how these data are processed, I would be very grateful if you could share with me, and if you could share the code of your data processing, then It could not be better. I promise they will be used only for research purposes.

Hello,

I am very interested in your brilliant work. However, I am very confused about the comparison between the proposed 4SDrug with other baselines. The baselines, such as SafeDrug, uses the diagnosis records but 4SafeDrug uses symptom records only. I want to know whether you use symptom records in baselines and how to use? Besides, I find the number of visits in your processed MIMIC dataset is different from the number in the paper SafeDrug. So, do you use the same processed data in the experiments of baselines and 4Safedrug?

Best Regards

I noticed that you have GAMENet in your baseline model, as far as I know the GAMENet model is useful for patient history information, how did you adapt the GAMENet model to get the results without entering historical data? If possible, I would like you to share the code of the baseline model and I assure you that the code will be used for research purposes only. Looking forward to your reply!