Comparative study of an artifical intelligent-based mosquito surveillance method and three traditional mosquito vector surveillance methods

doi:10.12140/j.issn.1000-7423.2026.03.013

Abstract

Abstract:

Objective To compare the mosquito capturing effectiveness of an artifical intelligent (AI)-based mosquito surveillance method (the intelligent trap method) with three traditional vector surveillance methods, including the light trap method, the human-baited double net trap method and the BG-trap method, so as to provide insights into the application of AI-based intelligent systems in mosquito surveillance. Methods A residential community with more than 500 households and a high mosquito density was selected from Xiuzhou District, Jiaxing City. Mosquito surveillance was simultaneously performed with the light trap method, the human-baited double net trap method, and the BG-trap method one day per week during the period from May to September 2025, while the intelligent mosquito trap was run as a 24/7 service. The light trap index, mosquito biting index, mosquito attraction index and the adult mosquito density captured by intelligent traps during the same period were calculated. The mosquito-capturing effectiveness was compared between the intelligent trap method and three traditional mosquito monitoring methods with the Kruskal-Wallis H test, and the mosquito-capturing outcomes and trends in the mosquito density of these four mosquito monitoring methods, as well as the accuracy of the intelligent trap system for identification of mosquito species and genders were descriptively analyzed. Results A total of 4 620, 23, 250, and 3 564 adult mosquitoes were captured with intelligent traps, human-baited double net traps, BG-traps, and light traps during the monitoring period, including 299 (6.47% of total mosquitoes captured), 16 (69.57%), 247 (98.80%), and 88 Aedes albopictus (2.47%), respectively. At weekly monitoring days with three traditional monitoring methods, 2 female adult mosquitoes were captured with intelligent traps during daytime (15∶00-17∶00), with an average adult mosquito density of 0.05 mosquitoes per hour, and 219 female adults were captured during nighttime, with an average density of 9.95 mosquitoes per night, while a total of 21 female adults were captured with human-baited double net traps, with an average mosquito biting index of 0.95 mosquitoes per hour. In addition, a total of 150 female adults were captured with BG-traps, with an average mosquito attraction index of 6.82 mosquitoes per hour during the monitoring period, and a total of 1 104 female adults were captured with light traps, with an average light trap index of 25.09 mosquitoes per light per night. There was a significant difference in the average mosquito-capturing rate at the same period during daytime among the intelligent trap method (0.05 mosquitoes per hour), the human-baited double net trap method (0.95 mosquitoes per hour), and the BG-trap method (6.82 mosquitoes per hour) (H = 39.559, P < 0.05), with a lower average mosquito-capturing rate seen for the intelligent trap method than the human-baited double net trap method and the BG-trap method (both P < 0.05), and a lower average mosquito-capturing rate seen for the intelligent trap method than the light trap method at the same period during nighttime (9.95 mosquitoes per night vs. 25.09 mosquitoes per light per night; H = 6.377, P < 0.05). A similar trend was seen in the fluctuation of the adult mosquito density between the intelligent trap method and the light trap method, with both peaking in week 8 (June 24), at densities of 29 mosquitoes per night and 118.5 mosquitoes per light per night, respectively. In addition, the number of adult mosquitoes recorded in the backend database of the intelligent traps was higher than the manually collected count (4 620 vs. 3 339 adult mosquitoes) from May 1 to September 29, and the estimated accuracy of AI-based intelligent traps was 88.71% for identification of mosquito species and 90.03% for identification of mosquito genders. Conclusion Although the trend in fluctuation of the adult mosquito density is comparable between the intelligent trap method and the light trap method, its mosquito-capturing rate is significantly lower than those of three traditional methods, including the light trap method, human-baited double net trap method and the BG-trap method. In addition, the number of adult mosquitoes recorded in the backend database of intelligent traps differs from the manually collected count, which requires further optimizations.

Key words: Intelligent trap method, Human-baited double net trap method, BG-trap method, Light trap method

CLC Number:

R384.11

ZHONG Chenhui, LUO Tianbin, ZHENG Han, LUO Mingyu. Comparative study of an artifical intelligent-based mosquito surveillance method and three traditional mosquito vector surveillance methods[J]. CHINESE JOURNAL OF PARASITOLOGY AND PARASITIC DISEASES, 2026, 44(3): 402-407.

Figures/Tables 2

References 16

[1]	World Health Organization. World malaria report 2025[R]. Geneva: WHO, 2025: 213.
[2]	Achee NL, Grieco JP, Vatandoost H, et al. Alternative strategies for mosquito-borne arbovirus control[J]. PLoS Negl Trop Dis, 2019, 13(1): e0006822.
[3]	刘起勇. 病媒生物监测预警研究进展[J]. 疾病监测, 2018, 33(2): 123-128.
	Liu QY. Research progress on the vector surveillance and early-warning[J]. Dis Surveillance, 2018, 33(2): 123-128. (in Chinese)
[4]	刘起勇. 我国病媒生物监测与控制现状分析及展望[J]. 中国媒介生物学及控制杂志, 2015, 26(2): 109-113, 126.
	Liu QY. State-of-art analysis and perspectives on vector surveillance and control in China[J]. Chin J Vector Biol Control, 2015, 26(2): 109-113, 126. (in Chinese)
[5]	Falk M, Garriga J, Eritja R, et al. Augmenting community-driven vector surveillance with automated image classification: lessons from the Artificial Intelligence Mosquito Alert (AIMA) system[J]. Epidemics, 2025, 53: 100863.
[6]	Micocci M, Manica M, Bernardini I, et al. An easier life to come for mosquito researchers: field-testing across Italy supports VECTRACK system for automatic counting, identification and absolute density estimation of Aedes albopictus and Culex pipiens adults[J]. Parasit Vectors, 2024, 17(1): 409.
[7]	闫冬明, 黄坤, 赵春春, 等. 常用蚊虫监测方法和技术研究进展[J]. 中国媒介生物学及控制杂志, 2020, 31(1): 108-112.
	Yan DM, Huang K, Zhao CC, et al. Research advances in common methods and techniques for mosquito surveillance[J]. Chin J Vector Biol Control, 2020, 31(1): 108-112. (in Chinese)
[8]	裘炯良, 周力沛, 郑剑宁, 等. 机器学习技术与病媒生物种属鉴定[J]. 中华卫生杀虫药械, 2017, 23(5): 436-439.
	Qiu JL, Zhou LP, Zheng JN, et al. The technology of machine learning and its application on identification of medical vectors[J]. Chin J Hyg Insectic Equip, 2017, 23(5): 436-439. (in Chinese)
[9]	Hubble C, Sorensen M, Parker K, et al. Evaluating vectech IDX?: AI-driven identification for enhanced vector management[J]. J Am Mosq Control Assoc, 2026, 41(4): 230-236.
[10]	Abdi YH, Abdullahi YB, Abdi MS, et al. Using artificial intelligence in vector control: a new path for public health[J/OL]. J Vector Borne Dis, 2025-09-23. https://journals.lww.com/jvbd/abstract/9900/using_artificial_intelligence_in_vector_control__a.187.aspx.
[11]	郭玉红, 陈传伟, 朱丽, 等. 河南省永城市白纹伊蚊日活动节律初步研究[J]. 中国媒介生物学及控制杂志, 2016, 27(5): 484-486.
	Guo YH, Chen CW, Zhu L, et al. Study on daily activity rhythm of Aedes albopictus in Yongcheng City[J]. Chin J Vector Biol Control, 2016, 27(5): 484-486. (in Chinese)
[12]	国家市场监督管理总局, 国家标准化管理委员会. 病媒生物密度监测方法蚊虫: GB/T 23797—2020[S]. 北京: 中国标准出版社, 2020.
	State Administration for Market Regulation, Standardization Administration of the People’s Republic of China. Surveillance methods for vector density—Mosquito: GB/T 23797—2020[S]. Beijing: Standards Press of China, 2020. (in Chinese)
[13]	Silva M, Gouveia BR, Santos JM, et al. Enhancing entomological surveillance: real-time monitoring of mosquito activity with the VECTRACK system in rural and urban areas[J]. Biology (Basel), 2025, 14(8): 1047.
[14]	韦凌娅, 孔庆鑫, 王慧敏. 双层叠帐法与BG-trap诱捕器法在登革热应急蚊媒监测中的捕蚊效果比较[J]. 中国媒介生物学及控制杂志, 2019, 30(1): 65-68.
	Wei LY, Kong QX, Wang HM. Effectiveness of double mosquito net and BG-trap for emergency vector surveillance during dengue fever epidemics: a comparative study[J]. Chin J Vector Biol Control, 2019, 30(1): 65-68. (in Chinese)
[15]	Lee S, Kim H, Cho BK. Deep learning-based image classification for major mosquito species inhabiting Korea[J]. Insects, 2023, 14(6): 526.
[16]	Lin NF, Qian YX, Wei L, et al. An autonomous intelligent mosquito sentinel for field-deployed surveillance[J/OL]. Commun Eng, 2026[2026-05-14]. https://www.nature.com/articles/s44172-026-00685-6.

监测方法 Method	捕蚊总只数 Total no. mosquito	雌蚊只数 No. female mosquito	雄蚊只数 No. male mosquito	伊蚊只数 No. Aedes		库蚊只数 No. Culex		按蚊只数 No. Anopheles		阿蚊只数 No. Armigeres
监测方法 Method	捕蚊总只数 Total no. mosquito	雌蚊只数 No. female mosquito	雄蚊只数 No. male mosquito	雌 Female	雄 Male	雌 Female	雄 Male	雌 Female	雄Male	雌 Female	雄Male
智能捕捉器法 Intelligent trap method	4 620	2 036	2 584	143	156	1 893	2 422	0	6	0	0
双层叠帐法 Human-baited double net trap method	23	21	2	16	0	5	1	0	0	0	1
BG-trap诱捕器法 BG-trap method	250	150	100	147	100	3	0	0	0	0	0
诱蚊灯法 Light trap method	3 564	1 104	2 460	46	42	1 057	2 416	1	0	0	2