Prediction of atelectasis in Mycoplasma pneumoniae pneumonia using a SHapley Additive exPlanations-interpretable machine learning model - Scorecard - MDSpire

Prediction of atelectasis in Mycoplasma pneumoniae pneumonia using a SHapley Additive exPlanations-interpretable machine learning model

  • By

  • Jia Sun

  • Wang Tengfei

  • Mengsi Li

  • Mian Wang

  • May 11, 2026

  • 0 min

Share

Clinical Scorecard: Forecasting Atelectasis in Pediatric Mycoplasma pneumoniae Pneumonia Using an Interpretable Machine Learning Approach with SHAP Analysis

At a Glance

CategoryDetail
ConditionAtelectasis in Pediatric Mycoplasma pneumoniae Pneumonia
Key MechanismsMachine learning models (KNN, SVM, NN) predicting atelectasis risk using clinical data.
Target PopulationPediatric patients with Mycoplasma pneumoniae pneumonia (MPP).
Care SettingPediatric hospital settings.

Key Highlights

  • Neural network model showed best performance (AUC 0.89, accuracy 0.82).
  • KNN and SVM models also performed well (AUC 0.88, specificity 0.87).
  • SHAP analysis identified NEU.pct, SAA, and CRP as key predictive variables.
  • Early identification of atelectasis is critical for timely intervention.
  • Machine learning offers a non-invasive tool for predicting atelectasis risk.

Guideline-Based Recommendations

Diagnosis

  • Utilize machine learning models for risk prediction of atelectasis in MPP.

Management

  • Implement early intervention strategies for high-risk patients identified by predictive models.

Monitoring & Follow-up

  • Regularly assess inflammatory markers and clinical symptoms in patients with MPP.

Risks

  • Consider risks associated with invasive procedures like bronchoscopy.

Patient & Prescribing Data

Children diagnosed with Mycoplasma pneumoniae pneumonia.

Focus on non-invasive predictive tools to guide management and reduce complications.

Clinical Best Practices

  • Incorporate machine learning tools in clinical decision-making for pneumonia management.
  • Use SHAP analysis to interpret model predictions and enhance clinical understanding.

Related Resources & Content

Original Source(s)

Prediction of atelectasis in Mycoplasma pneumoniae pneumonia using a SHapley Additive exPlanations-interpretable machine learning model

  • by Jia Sun, Wang Tengfei, Mengsi Li, Mian Wang

  • May 11, 2026

Related Content

Usefulness of ultrasound in diagnosing constipation in children: a narrative review

Case Report: Acute myeloid leukemia with mixed immunophenotypic features and TCF3::ZNF384 fusion in a pediatric patient with CNS involvement: a diagnostic and therapeutic challenge

AD Not Linked to Early CV Risk Markers

Repeated assessments from childhood to early adulthood did not identify consistent cardiometabolic risk patterns by disease activity or severity.