MosQNet-SA: Explainable convolutional-attention network for mosquito classification with application as a RESTful API for dengue and malaria risk mapping

Md Akmol Masud, Sanjida Akter, Nadia Sultana, Mohammad Shahidul Islam, Mohammed Abu Yousuf, Farzan M Noori and Md Zia Uddin

PLOS ONE, 2026, vol. 21, issue 4, 1-30

Abstract: Mosquito-borne diseases represent a significant global health challenge. Over 700,000 people succumb to mosquito-borne diseases annually, highlighting the important need for accurate and efficient mosquito classification systems. Current approaches face limitations in accuracy, computational efficiency, and interpretability, creating a gap that artificial intelligence can help address. This paper presents MosQNet-SA, a novel convolutional-attention network designed for mosquito classification that addresses these limitations through architectural choices. The proposed model incorporates a spatial attention mechanism and depthwise separable convolutions to enhance feature extraction while maintaining computational efficiency—achieving comparable performance with 10-fold fewer parameters than existing approaches. MosQNet-SA achieves 99.42% accuracy on a dataset of 1,000 images across three mosquito species (Aedes, Anopheles, and Culex), demonstrating strong performance compared to existing CNN architectures. The model’s explainability is enhanced through multiple methods, including Saliency, GradCAM, LIME, and Kernel SHAP, providing valuable insights into the decision-making process for public health practitioners. Additionally, we present a RESTful API implementation for real-time mosquito classification and disease risk mapping, demonstrating the practical applicability of our approach in public health surveillance systems.

Date: 2026
References: View complete reference list from CitEc
Citations:

Downloads: (external link)
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0344970 (text/html)
https://journals.plos.org/plosone/article/file?id= ... 44970&type=printable (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:plo:pone00:0344970

DOI: 10.1371/journal.pone.0344970

Access Statistics for this article

More articles in PLOS ONE from Public Library of Science
Bibliographic data for series maintained by plosone ().