Machine Learning-Based Prediction of Acute Kidney Injury in Patient Admitted to ICU with Sepsis: A Systematic Review of Clinical Evidence

Sepsis is a highly prevalent condition in intensive care units, with one of its severe complications being acute kidney injury (AKI). Sepsis-associated acute kidney injury (SA-AKI) can be a reversible process if timely recognition and adequate treatment are provided to the patient. This systematic review (SR) summarizes the current clinical evidence of machine learning (ML) based prediction models. After conducting the literature search, 9 publications meet the inclusion criteria of the SR, categorized into three groups: prediction of SA-AKI occurrence, prediction of persistent AKI in septic patients, and prediction of mortality in SA-AKI patients. In summary, based on the current clinical evidence, ML-based methods show great potential for future clinical applications. They have the ability to outperform conventional scoring systems (such as SOFA and SAPS II), indicating their promising role in clinical practice.