Early Prediction of Plastic Bronchitis in Pediatric MPP Patients via Interpretable ML

A retrospective cohort of 307 pediatric patients with Mycoplasma pneumoniae pneumonia (MPP) who underwent bronchoscopy intervention from April 2023 to June 2025. The dataset was used to develop and validate machine learning models for the early prediction of plastic bronchitis, with the best model achieving an AUC of 0.905 on a test set. The study was authored by Pei Wang and published on figshare.

Use Cases

• Training binary classification models to predict plastic bronchitis risk based on inflammation and nutrition indices.
• Developing feature importance analyses using SHAP to identify key clinical predictors like retinol-binding protein 4 and D-dimer levels.
• Building a clinical decision support tool, such as the provided web-based risk calculator, for early patient triage.

Strengths

• Model performance is quantitatively reported, with the best model achieving an AUC of 0.905 on a test set.
• The study includes 307 patient records, providing a defined cohort size for model development.
• Feature selection and model validation methods (LASSO, Boruta, 5-fold cross-validation) are explicitly described.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the underlying data is unknown, which may limit suitability assessment.
• The dataset is very small at 19.7 KB, indicating limited scope and likely a summary or model output rather than raw patient data.

Provenance

Source

figshare, author Pei Wang.

Collection Method

Retrospective cohort study of pediatric patients undergoing bronchoscopy.

Time Range

April 2023 to June 2025.

Freshness

Last updated 2026-04-23 05:34:37; freshness should be verified.