What question did this study set out to answer?

This study aims to develop and validate a machine learning-based diagnostic system for identifying pediatric respiratory pathogens.

June 1, 2026npj Digital MedicineOpen Access

Development and validation of a machine learning-based diagnostic system for 22 pediatric respiratory pathogens: a large-scale multicenter study

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial demonstrates improved pathogen identification in pediatric respiratory infections, suggesting enhanced treatment strategies.

This study aims to develop and validate a machine learning-based diagnostic system for identifying pediatric respiratory pathogens.
Multicenter study involved 134,500 hospitalized children across three clinical centers.
Integrated 42 clinical and laboratory features from electronic health records for accurate pathogen identification.
Prospective validation conducted on an independent cohort of 1338 children.
Pathog-PDx distinguished 22 pathogen subtypes and outperformed conventional models.
Achieved high classification performance for influenza virus (AUC = 0.95; Sn: 0.88; Sp: 0.86).
Mean AUCs of 0.88 for various pathogens of respiratory tract infections.

Su et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark

Randomized trial demonstrates improved pathogen identification in pediatric respiratory infections, suggesting enhanced treatment strategies.