Automated Continuous Acute Kidney Injury Prediction and Surveillance: A Random Forest Model

Area of Science:

• Critical Care Medicine
• Nephrology
• Health Informatics

Background:

• Acute kidney injury (AKI) is a common and serious complication in intensive care units (ICUs).
• Early detection and intervention are crucial for improving patient outcomes in AKI.
• Existing methods for AKI prediction often lack continuous monitoring capabilities.

Purpose of the Study:

• To develop and validate a predictive model for acute kidney injury (AKI) of any severity.
• To enable continuous surveillance and management of AKI in ICU settings.
• To improve clinical outcomes through early identification of high-risk patients.

Main Methods:

• Retrospective cohort study utilizing data from adult ICU patients at Mayo Clinic.
• Random forest classification algorithm employed for AKI risk score prediction.
• Model trained on 4572 patients and validated on 1958 patients, with data analyzed every 15 minutes.

Main Results:

• The AKI prediction model achieved an area under the receiver operating characteristic curve of 0.88 on validation.
• The model demonstrated 92% sensitivity and 68% specificity for detecting any AKI.
• It identified 30% of AKI cases at least 6 hours prior to clinical diagnosis, with higher accuracy for severe AKI stages.

Conclusions:

• A validated AKI prediction model using random forest can continuously monitor ICU patients.
• This model facilitates the identification of high-risk individuals for targeted preventive strategies.
• The tool supports proactive AKI management and enrollment in clinical trials.