I am trying to recreate the temporal dataset you used in your paper. I noticed in your preprocessing folder under the 'create_temporal_dataset.ipynb', that you used 'master.csv' file, please can you explain how this file was generated.
Also, in the datasets folder, you have another file 'temporal_mimic_cxr_dataset.py' where you used 'mimic_annotation.json' file.
Can you explain which of the longitudinal data you used, the one from the preprocessing folder or the one from the datasets folder? How was the 'master.csv' file and 'mimic_annotation.json' file created?
Thanks for your intererst in our work. We are refactoring our codebase and will release the preprocessing code to curate this longitudinal dataset as soon as possible.
Hi authors,
I am trying to recreate the temporal dataset you used in your paper. I noticed in your preprocessing folder under the 'create_temporal_dataset.ipynb', that you used 'master.csv' file, please can you explain how this file was generated.
Also, in the datasets folder, you have another file 'temporal_mimic_cxr_dataset.py' where you used 'mimic_annotation.json' file.
Can you explain which of the longitudinal data you used, the one from the preprocessing folder or the one from the datasets folder? How was the 'master.csv' file and 'mimic_annotation.json' file created?
dataset = TemporalMIMICCXRDataset( annotation_file="/home/r15user2/Documents/CXR_dataset/temporal_CXR/mimic_annotation.json", dataset_dir="/home/r15user2/Documents/CXR_dataset/mimic_data/2.0.0/files", split="train", train_data_pct=1, tokenizer=tokenizer )