CURE-CKD Bayesian Network for predicting 40% eGFR decline over a year

A personalized data-driven model of longitudinal laboratory data for patients with CKD or at risk for CKD that simulates missing values over time and predicts≥ 40% decline in eGFR over a 1-year period.

This is a model trained and tested on UCLA deidentified data and externally tested on Providence Health System through the CURE-CKD registry.

Note: the license was updated to reflect the non-commercial terms of the underlying training data.