Artificial Intelligence-powered tiered early warning framework addressing high false alarm rates for in-hospital mortality prediction - PubMed
4 hours ago
- #Early Warning System
- #In-Hospital Mortality
- #Artificial Intelligence
- AI-TEW is a two-stage early warning framework designed to reduce false alarms in in-hospital mortality (IHM) prediction.
- Stage 1 uses a robust machine learning model validated on 174,292 ED visits across three hospitals, achieving AUROCs of 0.84 to 0.91.
- Stage 2 implements tiered risk stratification, increasing PPV from 9.8-18.8% to 32.5-40.5% while maintaining high NPV (>98%).
- A knowledge-based filtering layer using LLMs interprets patient-specific risk factors, enhancing contextual understanding and reducing spurious alerts.
- AI-TEW integrates improved predictive efficiency with interpretable, knowledge-informed filtering to mitigate class imbalance in emergency risk prediction.