Development and validation of risk prediction models for acute kidney disease in gout patients: a retrospective study using machine learning
Abstract Background Limited research has been conducted on the prevalence of acute kidney injury (AKI) and acute kidney disease (AKD) in gout patients, as well as the impact of these renal complications on patient outcomes. This study aims to develop machine learning models to predict AKI and AKD in...
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMC
2025-07-01
|
| Series: | European Journal of Medical Research |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s40001-025-02939-z |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract Background Limited research has been conducted on the prevalence of acute kidney injury (AKI) and acute kidney disease (AKD) in gout patients, as well as the impact of these renal complications on patient outcomes. This study aims to develop machine learning models to predict AKI and AKD in gout patients, with the goal of deploying web-based applications to support clinicians in making informed, real-time decisions for high-risk patients. Methods A total of 1260 gout patients admitted to a tertiary hospital between January 2020 and January 2024 were included. The dataset was split into 80% for model training and 20% for testing model performance. Nine machine learning algorithms were evaluated, with performance assessed using metrics, such as AUROC, precision, recall, and F1 score. SHAP and LIME were used to visualize feature importance and interpret model predictions. The top-performing models were integrated into a web platform to identify patients at high risk of AKI and AKD. Results The incidence rates of AKI and AKD were 9.05% and 12.78%, respectively. Mortality rates were higher for patients with AKI (11.40%) and AKD (7.45%). The LightGBM model achieved excellent AUROC for predicting AKI (0.815) and AKD (0.873). SHAP visualizations revealed that the key predictors of AKI in gout patients were diuretics, serum sodium, and urate lowering; therapy agents, while predictors for AKD included age, diuretics, and AKI grade. SHAP force plots and LIME analyses provided individualized predictions. To facilitate clinical implementation, the model was simplified using the top 10 predictors while maintaining strong performance. Conclusions The significant incidence of AKI and AKD in gout patients warrants clinical attention. The web-based prediction model provide real-time predictions, helping clinicians identify high-risk patients and improve outcomes. |
|---|---|
| ISSN: | 2047-783X |