Pediatric Pneumonia Clinical Phenotypes and Prediction Model, 305 Patients

305 pediatric patients with community-acquired pneumonia were clustered into three clinical phenotypes using k-prototypes machine learning. The dataset, created by Meng Xiao and last updated in 2026, includes microbiological, inflammatory, and clinical features from a retrospective cohort between November 2022 and October 2025. A prediction model for the Mixed-Infection phenotype achieved an AUC of 0.917 on a test set.

Use Cases

• Training phenotype classification models based on microbiological and inflammatory markers.
• Developing early prediction tools for complex pneumonia cases based on white blood cell count, lactate dehydrogenase, and procalcitonin.
• Studying clinical heterogeneity in pediatric pneumonia based on integrated host-pathogen data.
• Benchmarking unsupervised clustering methods (k-prototypes) on clinical patient data.

Strengths

• Includes data from 305 pediatric patients, providing a specific clinical cohort.
• Phenotype proportions are explicitly quantified (Mycoplasma-Dominant 37.7%, Mixed-Infection 28.2%, High-Inflammation 34.1%).
• Prediction model performance is reported with specific metrics (AUC=0.917, accuracy=91.3%).

Limitations

• Row count and column-level documentation are unknown, which may limit suitability assessment.
• The cohort is selected from patients who underwent bronchoalveolar lavage, which may limit generalizability to all pediatric CAP patients.
• Description metadata is limited; actual data quality and structure require manual inspection after download.

Provenance

Source

figshare

Collection Method

Retrospective cohort study of patients who underwent bronchoalveolar lavage.

Time Range

November 2022 to October 2025

Freshness

Last updated 2026-05-21 05:42:38