Article Details

Abstract

With a massive population and severe urban traffic congestion, China faces a critical problem where emergency medical response times often exceed the golden rescue time window, leading to unnecessary loss of life. At present, air medical rescue in China is still in its developmental stage and faces significant challenges. The most urgent issue is how to match dynamic, multi-point rescue demands with limited aircraft capacity while meeting the high timeliness requirements and different medical priority levels. This study proposes a dynamic task allocation model and a trainable framework based on Multi-Agent Reinforcement Learning (MARL). Through targeted design, the dispatching and matching problem is transformed into a stochastic game process from a "hospital cluster" perspective. By solving the optimization problem within this framework, a robust optimization policy model is generated. Experimental results demonstrate that the proposed model achieves higher allocation benefits by considering demand dynamics, task priorities, and limited aircraft availability. Specifically, the model achieves higher task completion rates within time windows, shorter average waiting times, and better completion rates for critical missions within time windows. These results suggest that the proposed dynamic task allocation model is an effective approach to improving the efficiency of air medical rescue.

Keywords

air medical rescue; medical priority; dynamic task allocation; MARL

References

1. [1] You Shaochun, Zhang Xiaoli, Wu Feifei, etc The current development status of electric vertical takeoff and landing aircraft and its application prospects in aviation medical rescue [J]. Medical and Health Equipment, 2024, 45 (2): 82-86
2. [2] Liu Yan, Yan Qiyu, Hou Aijun Research on the Evolution, International Comparison, and Development Path of China's Low Altitude Economic Policy [J]. Logistics Technology, 2026, 45 (2): 1-11
3. [3] Guo Aibin, Gao Wen, Liu Bin, etc Research on the Construction Practice and Operation Mode of Air Ground Integrated Medical Rescue System [J]. Chinese Journal of Emergency Medicine, 2021, 41 (5): 438-443
4. [4] Yang Kaidong Research on Post Disaster Helicopter Rescue Path Planning with Time Window [D]. Liaocheng: Liaocheng University, 2024
5. [5] Zhu Xingguo, Tian Hui, Li Jingjing, etc Thoughts on the Construction of Aviation Emergency Rescue Service Platform [J]. Journal of the Chinese Academy of Electronic Sciences, 2022, 17 (10): 1021-1026
6. [6] Hao Zehang, Li Yanjun, Cao Yuyuan, etc EC135 P3H Helicopter Emergency Medical System Modification Design [J]. Medical and Health Equipment, 2025, 46 (11): 24-30
7. [7] Ding Ziyu, Shen Yang, Wang Muyuan, etc Research on the Allocation of Aviation Emergency Rescue Tasks in Earthquake Scenarios [C]//Proceedings of the Annual Meeting of the Emergency Rescue Branch of the Chinese Society of Aeronautics and Astronautics (2023-2025) Beijing University of Aeronautics and Astronautics Press, 2023: 16-26
8. [8] Xia Zhenghong, Pan Weijun Research on Multi target Allocation and Trajectory Planning for Multiple Rescue Helicopters [J]. Science, Technology and Engineering, 2013, 13 (34): 10226-10231
9. [9] Gaohai Peak Design of Task Scheduling Model in Aviation Emergency Rescue [J]. Integrated Circuit Applications, 2025, 42 (1): 104-106
10. [10] Zhang Qingsong, Jin IKEA, Qin Shuaixing Research on Optimization of Helicopter Collaborative Rescue Dispatch [J]. Chinese Journal of Safety Science, 2019, 29 (10): 180-186
11. [11] Zhang Kaiqian Research on Task Allocation of General Aviation Emergency Rescue Based on Market Mechanism [D]. Nanjing: Nanjing University of Aeronautics and Astronautics, 2018
12. [12] Qin Rui, Li Xiaoyu Research on Multi UAV Firefighting Task Allocation and Path Planning [J]. Flight Mechanics, 2026, 44 (1): 54-60
13. [13] Sun Pengyao, Huang Yanyan, reside in Jielin Task allocation method for unmanned aerial vehicle cluster based on improved dual population genetic algorithm [J/OL]. Command, Control and Simulation, 2026:1-17 https://link.cnki.net/urlid/32.1759.tj.20260119.1501.008.
14. [14] Huang Jin, Li Xinyang, Gao Zhen, etc Multi UAV logistics task allocation based on improved genetic simulated annealing algorithm [J]. Science, Technology and Engineering, 2025, 25 (30): 13179-13186
15. [15] Cheng Jiamin Research on forest firefighting helicopter scheduling strategy based on fire deduction [D]. Tianjin: Civil Aviation University of China, 2023
16. [16] Wang Tianjiao, Li Songwei, Li Hongjun, etc Helicopter Optimization Scheduling Method Based on Heuristic Algorithm [J]. Information and Computer (Theoretical Edition), 2019 (15): 41-42
17. [17] Wang Feng, Fu Qingpo, Han Mengchen, etc LeKMPSO algorithm for solving the collaborative multi task reassignment problem of heterogeneous unmanned aerial vehicles [J]. Control Theory and Applications, 2024, 41 (6): 1009-1017
18. [18] Pan Qiuxia, Lin Zhongming, Huang Wu, etc Research on Task Allocation Algorithm for Heterogeneous UAV Groups of Light and Small Logistics in Urban Areas under Low altitude Economy [J]. Enterprise Technology and Development, 2025 (2): 26-31
19. [19] Wang Haoyan, Wang Mingming, Shi Xinyue Cluster Spacecraft Task Allocation Method for Multi Task Objectives [J/OL]. Flight Mechanics, 2026:1-8 https://doi.org/10.13645/j.cnki.f.d.20260126.001.
20. [20] Li Mengke, Yao Guanxin, Xu Jing, etc Research on the problem of simultaneous pickup and delivery of mixed fleets based on Q-learning super heuristic algorithm [J/OL]. Computer Engineering and Applications, 2026:1-8 https://doi.org/10.13645/j.cnki.f.d.20260126.001.
21. [21] Zhang You'an, He Ziqi, Li Bochen, etc Dynamic task allocation for heterogeneous unmanned aerial vehicles based on task evaluation feedback [J]. Aviation Weapons, 2024, 31 (6): 78-85
22. [22] Hu Kaiming, Zhuang Yi, Xu Tao Multi agent resource adaptive allocation algorithm for dynamic demand [J]. Computer and Modernization, 2025 (12): 11-18, 25
23. [23] Xie Haiwen Design of distributed collaborative task allocation strategy for multi-agent systems [D]. Beijing: North China University of Technology, 2025
24. [24] Zhao X Y，Zong Q，Tian B L，et al. Fast Task Allocation for Heterogeneous Unmanned Aerial Vehicles Through Reinforcement Learning [J].Aerospace Science and Technology，2019，92：588-594.
25. [25] Wang Y J，Pan W J，Liu K Y. Multi-Agent Aviation Search Task Allocation Method [J/OL].IOP Conference Series: Materials Science and Engineering，2019，646：012058.https://iopscience.iop.org/article/10.1088/1757-899X/646/1/012058.
26. [26] Shen Y，Wang X B，Wang H J，et al. A dynamic task assignment model for aviation emergency rescue based on multi-agent reinforcement learning [J].Journal of Safety Science and Resilience，2023，4（3）：284-293.
27. [27] Zhang Huaqing, Hao Mingrui, Jiang Jixiang, etc Multi agent batch task allocation for emergency tasks [J]. Control Theory and Applications, 2025, 42 (11): 2242-2251
28. [28] Ministry of Health of the People's Republic of China Chinese Health Industry Standard Ambulance: WS/T 292-2008 [S]. Beijing: People's Health Publishing House, 2008
29. [29] Zhou Haiyan, Yang Jing, Fu Xiaomin, etc Research progress on time management strategies in the treatment of patients with acute severe trauma [J]. General Nursing, 2025, 23 (18): 3439-3443
30. [30] Rainford D J，David G. Ernsting's Aviation and Space Medicine 5E [M].[S.l.]：Retired Consultant Physician and Consultant Adviser in Renal Disease (Royal Air Force)，2015.
31. Liu Xu, Wu Qiaoyue, Gong Hongshen, etc Analysis of the Characteristics of Outpatient Emergency Patients in Zhengzhou City [J]. Chinese Journal of Emergency Recovery and Disaster Medicine, 2021, 16 (3): 249-252
32. [32] Yu C，Velu A，Vinitsky E，et al. The Surprising Effectiveness of PPO in Cooperative Multi-agent Games [J].Advances in Neural Information Processing Systems，2022，35：24611-24624.
33. [33] Zhong Bin, Tian Jianqing The current situation and strategies of aviation medical rescue development in China [J]. Chinese Journal of Disaster Rescue Medicine, 2019, 7 (9): 531-535

Funding

National Innovation and Entrepreneurship Training Program (202513294007)

About the authors

Yue Xuanyi (2004-), male, Han ethnicity, from Xuzhou, Jiangsu Province, undergraduate studies, research direction: reinforcement learning; Corresponding author: Zhang Qiongni (1981-), female, Han ethnicity, from Taizhou, Zhejiang, associate professor, PhD, research direction: resource management, intelligent algorithms.