Simulation of intelligent triage of earthquake casualties by medical rescue teams based on reinforcement learning
10.7644/j.issn.1674-9960.2025.01.003
- VernacularTitle:基于强化学习的医学救援队地震伤员智能分类仿真研究
- Author:
Juan WU
1
;
Junyao JING
;
Bin JING
;
Bin WU
;
Nana SUN
Author Information
1. 空军军医大学(第四军医大学)唐都医院,西安 710032
- Keywords:
earthquake;
casualty;
triage;
reinforcement learning;
simulation
- From:
Military Medical Sciences
2025;49(1):15-21
- CountryChina
- Language:Chinese
-
Abstract:
Objective To design and develop an intelligent triage model for earthquake casualties that is intended for medical rescue teams based on reinforcement learning and the feasibility of this model is verified via computer simulation.Methods The process,difficulty,and requirements of the triage of the injured during an earthquake medical rescue were analyzed.The Markov decision process was used to formally describe the problem.Subsequently,a triage model was designed and developed based on reinforcement learning.Finally,the effectiveness of the model was verified through simulation experiments.Results Compared with conventional triage strategies,this intelligent triage model showed significant advantages in terms of mortality rates and waiting time.Under experimental conditions,casualties decreased by nearly 50％,and the waiting time for both nonoperative casualties(T-class)and operative casualties(S-class)casual-ties also decreased.Conclusion The intelligent triage model can autonomously learn triage strategies,reduce the ca-sualty rate while lowering the waiting time for the injured,thereby effectively improving the efficiency of treatment of earthquake injuries.
