This folder includes that files for feature selection and preparation data needed for matrix construction.
This folder includes that files for feature matrix construction from raw data.
The content for each file used in the code:
e2p: mapping from encounter id to patient id
icd9toicd10: mapping from icd 9 code to icd 10 code, based on the table at https://www.health.govt.nz/nz-health-statistics/data-references/mapping-tools/mapping-between-icd-10-and-icd-9
position: mapping from encounter id to the position of the encounter in the patient's feature matrix
oudce: clinical evnets information for OD patients
ouddx: diagnosis information for OD patients
oudlab: lab tests information for OD patients
oudmed: medications information for OD patients
ouddemo: demographics for OD patients
comce: clinical evnets information for negative patients
comdx: diagnosis information for negative patients
comlab: lab tests information for negative patients
commed: medications information for negative patients
comdemo: demographics for negative patients
labdict: mapping to position of lab test in feature matrix
cedict: mapping to position of clinical event in feature matrix
dxdict: mapping to position of diagnosis in feature matrix
meddict: mapping to position of medication in feature matrix
racedict: mapping to position of race type in feature matrix
This folder includes the files of training and testing the predictive model.
Using the permutation method to get the features importances.
According codes are put in the doc files in seperate folder.
Xinyu Dong, Jianyuan Deng, Wei Hou, Sina Rashidian, Richard N. Rosenthal, Mary Saltz, Joel H. Saltz, Fusheng Wang, Predicting opioid overdose risk of patients with opioid prescriptions using electronic health records based on temporal deep learning, Journal of Biomedical Informatics, Volume 116, 2021, 103725, ISSN 1532-0464, https://doi.org/10.1016/j.jbi.2021.103725.