Accurate prediction of sepsis from pediatric emergency department to PICU using a machine-learning model
Genaue Vorhersage von Sepsis in der pädiatrischen Notaufnahme bis zur PICU mithilfe eines Machine-Learning-Modells
Frontiers in Pediatrics · 10.10.2025 · 13
DOI: 10.3389/fped.2025.1610187 ↗Auf einen Blick
In einer retrospektiven Studie sagte ein Machine-Learning-Framework pädiatrische Sepsis früh und mit hoher Genauigkeit voraus. Die externe Validierung bestätigte die Übertragbarkeit.
Fragestellung
Soll ein Machine-Learning-Framework mit statischen und dynamischen EHR-Daten eine frühe Sepsis-Erkennung bei pädiatrischen Patient:innen unterstützen?
Methode
Retrospektive Studie mit interner und externer Validierung; Vergleich von XGBoost- und GRU-basierten RNN-Modellen über ein 12-h-Fenster vor der klinischen Diagnose.
Ergebnisse
GWCMC n = 1.697, MIMIC-III n = 827. CTWH + MGP-XGBoost: AUROC 0.915 zum Diagnosezeitpunkt; externe Validierung AUROC 0.905. Medianer Vorlauf 6,2 h; κ = 0,82.
Schlussfolgerung der Autor:innen
Die Autor:innen folgern, dass ein Dual-Modell-Ensemble mit interpolationsbasierter Vorverarbeitung und interpretierbarem Machine Learning eine robuste frühe Sepsis-Erkennung ermöglicht.
Kernaussagen
- CTWH + MGP-XGBoost erreichte AUROC 0,915 zum Diagnosezeitpunkt
- Externe Validierung im MIMIC-III-Datensatz: AUROC 0,905
- Medianer Vorlauf der Alarme: 6,2 h vor der Diagnose
Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.
Kurz gefasst
Fachartikel in der Zeitschrift Frontiers in Pediatrics, online erschienen am 10.10.2025. Verfasst von Xuan Shi und 10 weiteren Autor:innen. Beteiligte Einrichtung: Pediatric Emergency Department, Guangzhou Women and Children's Medical Center, Guangzhou Medical University, Guangdong Provincial Clinical Research C…. Die Arbeit wurde bisher 6-mal zitiert (Crossref) und nennt 25 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Themen: Sepsis
Abstract (Originalsprache)
Background
Timely identification of pediatric sepsis remains a critical challenge in emergency and intensive care settings due to the heterogeneous clinical presentations across age groups. Existing scoring systems often lack temporal resolution and interpretability. We aimed to develop a real-time, machine learning–based prediction framework integrating static and dynamic electronic health record (EHR) features to support early sepsis detection.
Methods
This retrospective study included pediatric patients from Guangzhou Women and Children's Medical Center (GWCMC; n = 1,697) and an external validation cohort from the MIMIC-III database ( n = 827). Irregular time-series data were imputed using a correlation-enhanced continuous time-window histogram with multivariate Gaussian processes (CTWH + MGP). We compared the predictive performance of XGBoost and gated recurrent unit (GRU)-based RNN models over a 12-h window prior to clinical diagnosis. Model outputs were validated internally and externally using AUROC, AUPRC, and Youden index, with SHAP-based interpretability applied to identify key clinical features.
Results
The CTWH + MGP-XGBoost model achieved the highest AUROC at diagnosis time (T = 0 h; AUROC = 0.915), while the GRU-based model demonstrated superior temporal stability across early windows. Top contributing features included lactate, white blood cell count, pH, and vasopressor use. External validation confirmed generalizability (MIMIC-III AUROC = 0.905). Simulation of real-time alerts showed a median lead time of 6.2 h before clinical diagnosis, with κ = 0.82 agreement against physician-confirmed cases.
Conclusions
Our results suggest that a dual-model ensemble combining interpolation-based preprocessing and interpretable machine learning enables robust early sepsis detection in pediatric populations. The system supports integration into EHR platforms for real-time clinical alerts and may inform prospective trials and quality improvement initiatives.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Pediatric Emergency Department, Guangzhou Women and Children's Medical Center, Guangzhou Medical University, Guangdong Provincial Clinical Research Center for Child Health, Guangzhou, China.
- Ewell Technology Company, Hangzhou, China.
- Medical Department of Guangzhou Women and Children's Medical Center, Guangzhou Medical University, Guangdong Provincial Clinical Research Center for Child Health, Guangzhou, China.
Zitieren
APA
Shi, X., Wang, X., Yang, H., Fan, X., Liu, G., Song, Y., Peng, Q., Wang, Q., Sun, X., Ma, W., & Li, P. (2025). Accurate prediction of sepsis from pediatric emergency department to PICU using a machine-learning model. Frontiers in Pediatrics, 13. https://doi.org/10.3389/fped.2025.1610187
Vancouver
Shi X, Wang X, Yang H, Fan X, Liu G, Song Y, et al. Accurate prediction of sepsis from pediatric emergency department to PICU using a machine-learning model. Frontiers in Pediatrics. 2025;13. doi:10.3389/fped.2025.1610187
BibTeX
@article{Shi2025,
author = {Shi, Xuan and Wang, Xuying and Yang, Haomei and Fan, Xiaowei and Liu, Guangming and Song, Yongling and Peng, Qiuyan and Wang, Qiang and Sun, Xin and Ma, Wencheng and Li, Peiqing},
title = {Accurate prediction of sepsis from pediatric emergency department to PICU using a machine-learning model},
journal = {Frontiers in Pediatrics},
year = {2025},
volume = {13},
doi = {10.3389/fped.2025.1610187}
}
EUVIMED