GitHub - jygao97/CAMP: pytorch implementation for the CAMP model in ICDM 2019 "CAMP: Co-Attention Memory Networks for Diagnosis Prediction in Healthcare"

pytorch implementation for the CAMP model in ICDM2019 paper "CAMP: Co-Attention Memory Networks for Diagnosis Prediction in Healthcare"

Step1: Create a folder named "data" and two sub-folders named "ccs" and "mimic". Then download the CCS-single-level file and CCS-multi-level file from https://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp into the "ccs" folder. Download the mimic files "PATIENTS.csv", "DIAGNOSES _ICD.csv" and "ADMISSIONS.csv" from https://mimic.physionet.org/ into the "mimic" folder.
Step2: Use processMIMIC.py to extract patient historical records and demographics from MIMIC-III tables. After this step, we get "mimic.seqs", "mimic.profiles" and "mimic.pids" in the "mimic" folder.
Step3: Use preprocess.py to split the dataset into three parts: training set, validation set and testing set. After this step, we get "mimic.train", "mimic.valid" and "mimic.test" in the "mimic" folder.
Step4: Use mkHierarchy.py to extract the taxonomy from the CCS-multi-level file. After this step, we get "mimic.forgram" in the "mimic" folder.
Step5: With "mimic.train", "mimic.valid", "mimic.test", and "mimic.forgram" ready in the "mimic" folder, please use run.py to run our model, including training, validation in each epoch, test and evaluation.

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
README.md		README.md
mkHierarchy.py		mkHierarchy.py
model.py		model.py
preprocess.py		preprocess.py
processMimic.py		processMimic.py
run.py		run.py
utils.py		utils.py