Photo Credit: Monsitj
The following is a summary of “Acute Kidney Injury Prediction Model Using Cystatin-C, Beta-2 Microglobulin, and Neutrophil Gelatinase-Associated Lipocalin Biomarker in Sepsis Patients,” published in the March 2024 issue of Nephrology by Susianti et al.
Acute kidney issue (AKI) is a common issue in sepsis patients, affecting nearly half of all patients with severe sepsis. No effective treatment for AKI in sepsis, so prevention is critical.
Researchers conducted a prospective study exploring the biomarker models to predict AKI in patients with sepsis.
They compared Cystatin C, Beta-2 Microglobulin, and Neutrophil Gelatinase-Associated Lipocalin (NGAL) levels in patients with sepsis with and without AKI. Using Orange Data Mining for machine learning (ML), 130 samples with 23 lab parameters were analyzed to develop the biomarker prediction model.
The results showed that both SVM and Naïve Bayes models were used for ML.The SVM model showed 50% sensitivity, 94.4% specificity, 71.4% negative predictive value (NPV), and 87.5% positive predictive value (PPV). The Naive Bayes model had 83.3% sensitivity, 77.8% specificity, 87.5% NPV, and 71.4% PPV. The SVM-MLhad higher AUC and specificity but lower sensitivity, whereas the Naive Bayes model had better sensitivity in predicting patients with sepsis.
Investigators concluded that the Naive Bayes ML model is effective in predicting AKI in patients with sepsis.
