JMIR Publications

ABSTRACT

Background:

Acute kidney injury (AKI) is common in patients with community-acquired pneumonia (CAP) and is associated with increased morbidity and mortality.

Objective:

This study aimed to establish and validate predictive models for AKI in hospitalized patients with CAP based on machine learning (ML) algorithms.

Methods:

We trained and externally validated five ML algorithms, including logistic regression (LR), support vector machine (SVM), random forest (RF), extreme gradient boosting (XGBT), and deep forest (DF). Feature selection was conducted using sliding window forward feature selection (SWFFS) technique. Shapley additive explanation (SHAP) and local interpretable model-agnostic explanations (LIME) techniques were applied to the optimal model for visual interpretation.

Results:

A total of 6371 CAP patients met the inclusion criteria. The development of CAP-AKI was recognized in 1006 (15.8%) patients. The 11 selected indicators were sex, temperature, breathing rate, diastolic blood pressure, C-reactive protein (CRP), albumin, white blood cell, hemoglobin, platelet, blood urea nitrogen (BUN), and neutrophil count. The DF model achieved the best area under the receiver operating characteristic curve (AUROC) and accuracy in the internal (AUROC: 0.885, Accuracy: 0.896) and external validation sets (AUROC: 0.866, Accuracy: 0.833). Moreover, the DF model had the best calibration among all models. Additionally, a web-based prediction platform was developed to predict community-acquired pneumonia-associated AKI(CAP-AKI).

Conclusions:

The model described in this study is the first multi-center validated AKI prediction model that accurately predicts CAP-AKI during hospitalization. The web-based prediction platform embedded with the DF model serves as a user-friendly tool for early identification of high-risk patients. Clinical Trial: Not available.