firasgit / metra Goto Github PK

Inspired by your work, I am now attempting to apply the model proposed in your article to some transfer learning tasks. However, due to the limited size of my dataset, utilizing a pre-trained model such as MIMIC would likely improve the performance of my task. On one hand, my objective is not to reproduce the experimental results of your article. On the other hand, acquiring the MIMIC dataset (MIMIC-CXR and MIMIC IV) and training a model from scratch can be quite time-consuming. Therefore, I would like to kindly request if you could share the trained model (MeTra (CP+CXR)) used in your article. It would be greatly appreciated if you could provide me with access to the trained model, as it would significantly expedite my research.

Does not make sence to give names like "MeTra (Medical Transformer)" or "Medical Diffusion" to the structures, which were already published and known before. Those fancy names for non-novel architectures sounds as if the model is original, but in reality they were only applied for medical data.

Great work! Reading the code, but can't find information about how mimic-cxr-ehr-split.csv got generated. Is it possible to share the idea how this file get generated?

From the Figure 1 in the publication[1], the model was trained on a train/val/test split of 4396/ 472/1257.

However, the MedFuse preprocessing pipeline[2], which has been reused for this paper with no major modification, reports 4885/540/1373. Crucially, this is because the authors made a small mistake that is still existent in MeTra.

Here is the error in Metra

    if cfg.dataset.task == 'in-hospital mortality': # should be 'in-hospital-mortality'
           end_time = cxr_merged_icustays.intime + pd.DateOffset(hours=48)

which will include cxr samples > 48h after admission. After fixing this error, the MedFuse authors report a train/val/test split of 4485/488/1242 which is somewhat more inline with the reported split in MeTra. I could not find any preprocessing steps that would otherwise explain this gap.

Thus the question: What split correspond to the performance reports[3] in the publication? I ask this because the code does not match the reported train/val/test split

[1] https://www.nature.com/articles/s41598-023-37835-1/figures/1
[2] https://github.com/nyuad-cai/MedFuse/tree/6f827589afd89562813cc5aa915762d054c29efc
[3] https://www.nature.com/articles/s41598-023-37835-1/figures/2

hello, great work! I am trying to run it as a baseline but I realized the evalutation ipynb file mentioned in the repository is not visible. it says the script shouldbe under /classification/eval but no such directory exists.