A machine learning-based prediction model for sepsis-associated delirium in intensive care unit patients with sepsis-associated acute kidney injury

Abstract

Sepsis-associated acute kidney injury (SA-AKI) patients in the ICU often suffer from sepsis-associated delirium (SAD), which is linked to unfavorable outcomes. This research aimed to develop a machine learning-based model for early SAD prediction in SA-AKI patients. Data was sourced from the Medical Information Mart for Intensive Care IV (MIMIC-IV) and eICU Collaborative Research Database (eICU-CRD). Various models, including logistic regression, extreme gradient boosting (XGBoost), random forest, k-nearest neighbors, support vector machine, decision tree, and naive Bayes, were constructed and evaluated. The XGBoost model emerged as the best, with an internal validation AUROC of 0.775 and an external validation AUROC of 0.687. Unlike traditional delirium assessments, this model enables earlier SAD prediction and is suitable for patients who are hard to assess conventionally.

Keywords: MIMIC-IV database; eICU-CRD database; machine learning; predictive model; sepsis-associated acute kidney injury; sepsis-associated delirium.

Figure 1.

Research of SA-AKI and without SA-AKI patients flowchart.

Figure 2.

The association between acute kidney injury staging and the incidence of delirium in sepsis patients.

Figure 3.

Research of machine learning-based prediction models for SAD in ICU patients with SA-AKI flowchart.

Figure 4.

(A) Kaplan–Meier survival curves of 30-days hospital mortality for SAD and Non-SAD groups in the MIMIC-IV database. (B) Kaplan–Meier survival curves of 60-days hospital mortality for SAD and Non-SAD groups in the MIMIC-IV database. (C) Kaplan–Meier survival curves of 90-days hospital mortality for SAD and Non-SAD groups in the MIMIC-IV database. (D) Boxplots of hospital length of stay for SAD and Non-SAD groups in the MIMIC-IV database.

Figure 4.

(A) Kaplan–Meier survival curves of 30-days hospital mortality for SAD and Non-SAD groups in the MIMIC-IV database. (B) Kaplan–Meier survival curves of 60-days hospital mortality for SAD and Non-SAD groups in the MIMIC-IV database. (C) Kaplan–Meier survival curves of 90-days hospital mortality for SAD and Non-SAD groups in the MIMIC-IV database. (D) Boxplots of hospital length of stay for SAD and Non-SAD groups in the MIMIC-IV database.

Figure 5.

(A) Cross-validation plot for the penalty term. The dashed lines represent the lambda.min and lambda.1se. (B) Plots for the LASSO regression coefficients over different values of the penalty parameter.

Figure 6.

(A) The receiver operating characteristic (ROC) curves of the LR, SVM, XGBoost, RF, DT, NB, and KNN models on the internal validation set. (B) The ROC curves of the LR, SVM, XGBoost, RF, DT, NB, and KNN models on the external validation set. (C) Calibration curves of the LR, XGBoost, SVM models. (D) Decision curves of the LR, XGBoost, SVM models.

Figure 7.

The receiver operating characteristic (ROC) curves of the SOFA, SAPS II, APS III, Multi-Score and XGBoost model on the internal validation set. Multi-score include SOFA, SAPS II and APS III.

Figure 8.

(A) Feature importance ranking plot of the XGBoost model (top 15 features). (B) SHAP summary plot of the XGBoost model (top 15 features). Abbreviations: Vent: mechanical ventilation; gcs: Glasgow coma scale; seda: sedation; resp_rate: respiratory rate; Na: sodium; aniongap: anion gap; bun: blood urea nitrogen; P: phosphate; glu: glucose.

Figure 9.

(A–O). Partial dependence plots of features. Y-axis represents SHAP values; X-axis represents actual clinical parameters for continuous variables, and for binary variables (e.g. MV, gcs, sedation, SOFA), ‘0’ indicates absence and ‘1’ indicates presence of the condition. Abbreviations: vent: mechanical ventilation; gcs: Glasgow coma scale; seda: sedation; resp_rate: respiratory rate; Na: sodium; aniongap: anion gap; bun: blood urea nitrogen; P: phosphate; glu: glucose.