Zero-Knowledge Risk Oracle for predictive diagnoses of childhood neuropsychiatric disorders from sparse electronic health records
Project description
- Info:
Zero-Knowledge Risk Oracle
- Author:
ZeD@UChicago <zed.uchicago.edu>
- Description:
Estimation of the risk of future diagnoses of neuropsychiatric disorders (particularly autism) in early childhood, based on the diagnostic codes recorded during doctor visits. The prediction pipeline is based on inferring optimal stochastic generators for diagnostic code sequences, and detecting subtle deviations that drive up risk of an eventual neuropsychiatric diagnoses. The out-of-sample AUC score on the Truven dataset of insurance claims (close to 3 million children in out-of-sample data) is just over 80%, for both males and females.
Usage:
from ehrzero import ehrzero
ehr.predict_with_confidence(SOURCE,n_first_weeks)
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.