Prediction of AKI in ICU Patients Based on Interpretable Machine Learning

journals.sagepub.com January 9, 2025

The machine learning model described in this study is capable of accurately predicting the onset of AKI in ICU patients up to 24 hours in advance. Validated within the MIMIC-IV and MIMIC-III databases, the model demonstrates commendable performance and has the potential to improve AKI outcomes in the ICU by providing early warnings and actionable feedback.

The Medical Information Mart for Intensive Care (MIMIC) databases were used to construct a dataset of critically ill patients.

Predictive models were constructed using five machine learning algorithms based on MIMIC-IV data, and the best predictive model was selected by multiple model evaluation metrics.

A total of 18,186 patient data were included in this study.

The analysis combining calibration and decision curves demonstrated that the eXtreme Gradient Boosting (XGBoost) exhibited superior performance among the five algorithms, achieving an area under the receiver operating characteristic curve of 0.88.