Klebsiella Pneumoniae Liver Abscess Recurrence Risk Predictors from a Multicenter Study

A multicenter retrospective study from 2016 to 2024 included 829 patients with Klebsiella pneumoniae liver abscess (KPLA). The dataset, created by Liyong Zhang, was used to develop an XGBoost model for predicting recurrence risk, integrating 24 candidate clinical and laboratory variables.

Use Cases

• Training machine learning models for clinical risk prediction based on features like age, diabetes status, and lab indicators.
• Validating feature selection methods for medical datasets using techniques like LASSO regression and the Boruta algorithm.
• Developing interpretable clinical tools using SHAP analysis on predictors such as fibrinogen level and SOFA-2 score.
• Building web-based clinical decision aids for early identification of high-risk patients with liver abscesses.

Strengths

• Dataset includes 829 patient records from three tertiary hospitals, providing a multicenter perspective.
• Model validation reported AUC values up to 0.936 on the training set and 0.904 on a quasi-external validation set.
• Feature selection identified 9 key predictors, including age, type 2 diabetes, and procalcitonin level.

Limitations

• Row count for the underlying data is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect geographic and institutional bias inherent to the three specific hospitals in the study.

Provenance

Source

Affiliated Hospital of Chengde Medical University, Kailuan General Hospital, and the First Hospital of Qinhuangdao.

Collection Method

Multicenter retrospective study.

Time Range

2016–2024

Freshness

Last updated 2026-05-20 05:44:05; freshness should be verified.

Geography

Data collected from hospitals in China; specific locations not detailed.