Four-Gene Signature for Early Diagnosis of Refractory Pediatric Pneumonia

349 pediatric patient records form a dataset for developing a diagnostic model for refractory Mycoplasma pneumoniae pneumonia. The study, authored by Qiuting Wu and last updated in April 2026, identifies a four-gene signature from peripheral blood mononuclear cell expression profiles. It includes a training cohort of 295 patients and an external validation cohort of 54 patients.

Use Cases

• Train a multinomial logistic regression model to classify patients into Normal, GMPP, or RMPP categories using the IGHM, NEAT1, IL32, and ACTG1 gene expression features.
• Validate diagnostic model performance using the reported macro-average AUC metrics from the training and external validation cohorts.
• Analyze the differential expression of the eight candidate genes identified via single-cell RNA sequencing to understand immune response in refractory pneumonia.
• Assess the clinical utility of the four-gene signature for early intervention by replicating the three-category classification achieving 88.8% accuracy.

Strengths

• 349 patient records with clear cohort division for model training and validation.
• Model validated externally on a separate cohort of 54 patients and internally with 1000 bootstrap resamples.
• Diagnostic model achieved a macro-average AUC of 0.987 on the external validation set.

Limitations

• Small dataset scale at 1.5 MB, with patient count limited to 349 children from a single study.
• Geographic and demographic bias is likely present, as the population source is not specified.
• Data is presented in a PDF report, requiring extraction to obtain the underlying gene expression and patient classification tables.

Provenance

Source

figshare, authored by Qiuting Wu.

Collection Method

Data gathered from a clinical study using single-cell RNA sequencing and RT-qPCR on peripheral blood mononuclear cells from pediatric patients.

Time Range

null

Freshness

Last updated in April 2026.

Geography

null