CT and Blood Markers Predict Severe Mycoplasma Pneumonia in Children

Researchers have built a prediction model that combines chest CT imaging features with routine laboratory indicators to identify children at risk of severe Mycoplasma pneumoniae lobar pneumonia early in their hospital admission. The study, published in the European Journal of Medical Research, enrolled 289 children with the infection retrospectively at a single center and split them into severe and non-severe groups to develop and validate an early risk stratification tool.

The team collected baseline clinical data, laboratory metrics and CT imaging features for each child, then used multivariate logistic regression to construct the prediction model. To make the model’s outputs interpretable, the researchers applied the SHAP (SHapley Additive exPlanations) framework after fitting the model, which quantifies how much each variable contributes to individual predictions. Model discrimination, calibration and clinical utility were assessed through internal validation.

The final comprehensive model contained eight clinical and imaging variables and achieved an area under the receiver operating characteristic curve of 0.75 in the training set and 0.77 in the validation set, a level the authors describe as moderate discriminative ability. Bootstrap internal validation with 1,000 resamples supported the result, yielding an optimism-corrected C-index of 0.723 and a mean absolute calibration error of 0.036, indicating that predicted risks stayed close to observed outcomes within the study sample.

SHAP analysis ranked lactate dehydrogenase (LDH), involvement of three or more lung lobes, and the CT attenuation value of consolidated lung tissue as the three largest contributors to the model’s predictions. According to the authors, CT attenuation provided supplementary imaging value beyond what clinical and laboratory indicators alone captured.

The authors emphasize important limitations. The study was retrospective and single-center, and the model has not been externally validated in other hospitals or patient populations. They therefore position it as an admission-based screening aid rather than a standalone clinical decision tool, and state that prospective, multicenter external validation is required before the model can be used routinely in clinical practice.

Subject of Research: Risk prediction modeling for severe Mycoplasma pneumoniae lobar pneumonia in children using CT imaging and laboratory indicators

Article Title: Analysis of influencing factors and construction of a risk prediction model for severe Mycoplasma pneumoniae lobar pneumonia in children based on CT imaging and laboratory indicators

Article References: Mu, H., Wang, W., & Zhang, Z. (2026). Analysis of influencing factors and construction of a risk prediction model for severe Mycoplasma pneumoniae lobar pneumonia in children based on CT imaging and laboratory indicators. European Journal of Medical Research. https://doi.org/10.1186/s40001-026-05320-w

Image Credits: AI Generated

DOI: 10.1186/s40001-026-05320-w

Keywords: Mycoplasma pneumoniae, lobar pneumonia, children, CT imaging, lactate dehydrogenase, risk prediction model, logistic regression, SHAP, machine learning interpretability, internal validation, pediatric pneumonia, Analysis