Dynamic Mortality Prediction in Critically Ill Children During Interhospital Transports Using Explainable AI

Feb 5, 2025·
Zhiqiang Huo
· 1 min read
Abstract
This study introduces PROMPT (Patient-centred Real-time Outcome Monitoring and Mortality PredicTion), an explainable AI-driven model designed to provide real-time mortality risk prediction for critically ill children during interhospital transport to PICUs. The model continuously evaluates patient risk, supporting clinical decision-making and improving transport safety.
Type
Publication
npj Digital Medicine

🚑 AI-Driven Mortality Prediction for Interhospital Transport in Pediatric Intensive Care

We are thrilled to share our latest research on real-time mortality risk prediction in critically ill children during interhospital transport to paediatric intensive care units (PICUs). This study, published in npj Digital Medicine, introduces an explainable AI-powered model designed to enhance clinical decision-making in transport medicine.

🔬 Key Findings

  • Real-time, dynamic risk prediction – Continuously evaluates mortality risk based on vital signs, interventions, and transport conditions.
  • Explainable AI with SHAP analysis – Identifies key clinical contributors to risk estimation, enhancing trust and usability.
  • Higher accuracy than traditional methods – The PROMPT model outperforms PIM3, achieving AUROC of 0.83.
  • Scalable deployment – Designed for integration into ambulance monitoring systems and clinical decision-support tools.

📢 Clinical Impact

This research provides clinicians with real-time, AI-driven insights to support early interventions, reduce preventable adverse events, and optimize transport outcomes. The PROMPT model paves the way for next-generation predictive analytics in pediatric intensive care transport.

📄 Read the full article here: Publication Link

Pediatric Critical Care Mortality Prediction Machine Learning in Healthcare Health Data Science ICU Transport Monitoring