Validation of the Klinrisk Machine Learning Model for CKD Progression in a Large Representative US Population.

Area of Science:

• Nephrology
• Artificial Intelligence
• Health Informatics

Background:

• Early identification of high-risk chronic kidney disease (CKD) is crucial for effective patient management and improved outcomes.
• The Klinrisk machine learning model was developed to predict CKD progression.
• Validation in large, diverse US populations is essential.

Purpose of the Study:

• To validate the Klinrisk machine learning model for predicting CKD progression.
• To assess the model's performance across commercial, Medicare, and Medicaid insurance populations in the US.

Main Methods:

• Three cohorts of insured adults (commercial, Medicare, Medicaid) were established between 2007-2020.
• Patients with at least one serum creatinine test and specific eGFR ranges were included.
• Disease progression was defined as a 40% eGFR decline or kidney failure; model performance was evaluated using AUC and Brier scores.

Main Results:

• The Klinrisk model demonstrated strong predictive performance across all populations.
• In the commercial cohort, Area Under the Curve (AUC) ranged from 0.83 to 0.87.
• Performance in Medicare and Medicaid cohorts also showed high accuracy, with AUCs ranging from 0.80 to 0.86.

Conclusions:

• The Klinrisk machine learning model is accurate for predicting CKD progression.
• The model's effectiveness was confirmed in a large-scale validation across commercial, Medicare, and Medicaid populations.
• This tool can aid in the early identification of patients at high risk for CKD progression.