Predicting COVID-19 Severity in Pediatric Patients using Machine Learning

Strategic Insights -

A new study from Tehran, Iran, has unveiled a significant breakthrough in predicting COVID-19 severity in pediatric patients. The research, conducted by a team from the Tehran University of Medical Sciences, employed machine learning algorithms to analyze the clinical characteristics of 588 pediatric patients with confirmed COVID-19. The findings suggest that machine learning, particularly ensemble methods, can improve risk stratification for pediatric COVID-19 by identifying key predictors of disease severity.

Key Takeaways:

  • A retrospective analysis of 588 pediatric patients with confirmed COVID-19 was conducted to evaluate the performance of machine learning algorithms in predicting disease severity.
  • Various machine learning models were trained and assessed, with a SuperLearner ensemble model implemented to enhance predictive accuracy.
  • Random Forest exhibited the highest performance, achieving an accuracy of 90.1%, sensitivity of 90.2%, and specificity of 90.1%.
  • The SuperLearner ensemble further improved predictive performance, demonstrating the lowest mean risk estimate.
  • Key predictors, including oxygen saturation, respiratory parameters, and specific laboratory markers, played a crucial role in distinguishing severe from non-severe cases.
  • The research suggests that integrating these predictive models into clinical practice could support early identification of high-risk patients and optimize clinical decision-making.
  • Setareh Mamishi, Pediatric Infectious Disease Research Center, Tehran University of Medical Sciences, Tehran, Iran, led the research team.
  • Additional authors include Babak Pourakbari, Sepideh Keshavarz Valian, Shima Mahmoudi, Reihaneh Hosseinpour Sadeghi, Mohammad Reza Abdolsalehi, Mahmoud Khodabandeh, and Mohammad Farahmand.

Statistics:

  • 588 pediatric patients with confirmed COVID-19 were included in the study.
  • Machine learning models achieved an accuracy of 90.1% (Random Forest model).
  • Sensitivity of 90.2% and specificity of 90.1% were reported for the Random Forest model.
  • The SuperLearner ensemble model demonstrated the lowest mean risk estimate.
  • Oxygen saturation, respiratory parameters, and specific laboratory markers were identified as key predictors of disease severity.

Sources:

  • Predicting COVID-19 severity in pediatric patients using machine learning: a comparative analysis of algorithms and ensemble methods. Scientific Reports, 2025;15(1):29118.
  • Pediatrics Week. August 30, 2025; p 178.