Table 1_Early risk stratification for carbapenem resistance among Pseudomonas aeruginosa i

Jianhong Yu's research dataset, published on figshare in April 2026, contains clinical and laboratory data for developing a machine-learning model. The data includes 1,666 patients from a primary center and an external validation cohort of 471 patients. It was used to create an interpretable model for early risk stratification of carbapenem-resistant Pseudomonas aeruginosa infections.

Use Cases

• Training a 'rule-in' diagnostic model based on routine complete blood count parameters and demographic features.
• Evaluating model interpretability using SHAP analysis based on features like hemoglobin and platelet-to-lymphocyte ratio.
• Assessing clinical utility via decision curve analysis across threshold probabilities.
• Conducting sensitivity analysis for class imbalance using techniques like SMOTE.

Strengths

• Model performance is validated on an independent external cohort of 471 patients.
• The random forest model achieved an external validation AUC of 0.898 and specificity of 0.985.
• Dataset includes 32 initial variables, with LASSO regression identifying 7 key predictors.
• Model calibration was excellent, with a Brier score of 0.073 in external validation.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the primary dataset is unknown, which may limit suitability assessment.
• The model is described as having modest sensitivity (0.600 externally), limiting its use as a standalone screening tool.

Provenance

Source

figshare, author Jianhong Yu.

Collection Method

Retrospective analysis of patient data from a primary medical center.

Time Range

null

Freshness

Last updated 2026-04-16 05:18:52; freshness should be verified.

Geography

null