Related Experiment Videos
Predicting dynamic individual out-of-hospital cardiac arrest risks using explainable machine learning: a multicenter
Wenyi Tang1, Lingyun Zou1, Jun Xiao1
1Chongqing Key Laboratory of Emergency Medicine, Chongqing Emergency Medical Center, Chongqing University Central Hospital, School of Medicine, Chongqing University, Chongqing, China.
NPJ Digital Medicine
|May 14, 2026
Summary
Environmental factors significantly impact out-of-hospital cardiac arrest (OHCA) risk. A new XGBoost model integrating weather and patient data improves OHCA prediction accuracy, aiding emergency medical services.
Area of Science:
- Public Health
- Environmental Epidemiology
- Biostatistics
Background:
- Out-of-hospital cardiac arrest (OHCA) presents a major public health concern.
- Existing risk prediction tools often lack dynamic environmental factor integration.
- Accurate, real-time risk assessment for OHCA is crucial for timely intervention.
Purpose of the Study:
- To develop and validate a novel OHCA risk prediction model.
- To integrate individual patient data with environmental (meteorological) factors.
- To compare the performance of different machine learning models for OHCA risk prediction.
Main Methods:
- A multicenter nested case-control study design was employed.
- Data from 26,145 OHCA cases and 162,160 controls in Chongqing, China (Jan 2021-Jul 2024) were analyzed.
- Extreme Gradient Boosting (XGBoost) was utilized, integrating clinical and daily meteorological data.
Main Results:
- The XGBoost model incorporating both individual and meteorological factors achieved an AUC of 0.810, outperforming models using only individual factors (AUC 0.789, P < 0.001).
- XGBoost demonstrated a 5.4-fold higher positive predictive value than logistic regression at 99% specificity.
- The model maintained high sensitivity (0.96) at the Youden-optimal threshold.
Conclusions:
- Integrating real-time meteorological data significantly enhances OHCA risk prediction models.
- The developed XGBoost model offers improved predictive accuracy, particularly at high specificity levels.
- This approach can inform situational awareness and bolster emergency medical services preparedness for OHCA events.
