Classification and Regression Tree Predictive Model for Acute Kidney Injury in Traumatic Brain Injury Patients

Area of Science:

• Nephrology
• Neurology
• Critical Care Medicine

Background:

• Acute kidney injury (AKI) is a common complication in hospitalized patients with traumatic brain injury (TBI).
• AKI in TBI patients is associated with increased morbidity and mortality.
• Predictive models are needed to identify TBI patients at high risk for AKI.

Purpose of the Study:

• To develop a visual and convenient decision-tree-based model for predicting AKI in TBI patients.
• To compare the predictive accuracy of a decision tree model with LASSO regression.

Main Methods:

• A cohort of 376 TBI patients was analyzed.
• Data included demographics, vital signs, lab results, and treatments.
• AKI prediction models were built using Classification and Regression Trees (CART) and LASSO regression.
• Model performance was evaluated using Receiver Operating Characteristic (ROC) curves and Area Under the Curve (AUC).

Main Results:

• The incidence of AKI was 10.4% in TBI patients.
• LASSO regression identified glucose, serum creatinine, cystatin C, serum uric acid, and fresh frozen plasma as predictors.
• The decision tree model highlighted glucose, serum uric acid, and cystatin C.
• The decision tree model achieved an AUC of 0.892, slightly higher than the LASSO model's 0.854.

Conclusions:

• A decision tree model is a valuable tool for predicting AKI in TBI patients.
• The visual, tree-based flowchart assists clinicians in identifying at-risk patients.
• Early identification facilitates the development of appropriate therapeutic strategies for AKI in TBI.