Pediatric ED Models Predict Sepsis With High Accuracy
Alpern et al. (JAMA Pediatrics, 2025) developed and validated machine-learning models using electronic health record data from five PECARN health systems to predict pediatric sepsis with organ dysfunction within 48 hours, using the first four hours of ED data. The gradient tree boosting model achieved AUROC 0.94 and positive likelihood ratios up to 6.18, indicating strong discrimination and potential for earlier identification.
Key Points
- 1Developed ML models using first four hours of ED EHR to predict pediatric sepsis.
- 2Achieved AUROC 0.94 for gradient boosting, with positive likelihood ratios indicating strong discrimination.
- 3Enable earlier identification before organ dysfunction, potentially reducing delays to treatment in ED.
Scoring Rationale
High novelty and robust multicentre validation support impact, tempered by implementation challenges, fairness differences, and limited generalisability to other ED settings.
Sources
Public references used for this report.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Ad Tech problems
