Image-based honey bee larval viral and bacterial diagnosis using machine learning

Authors: Copeland, Duan; Mott, Brendon; Kortenkamp, Oliver; Erickson, Robert; Allen, Nathan; Anderson, Kirk

Submitted to: Scientific Reports
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/13/2025
Publication Date: 8/21/2025
Citation: Copeland, D.C., Mott, B.M., Kortenkamp, O.L., Erickson, R.J., Allen, N.O., Anderson, K.E. 2025. Image-based honey bee larval viral and bacterial diagnosis using machine learning. Scientific Reports. 15:Article 30717. https://doi.org/10.1038/s41598-025-16261-5.
DOI: https://doi.org/10.1038/s41598-025-16261-5

Interpretive Summary: The Problem: Honey bees face numerous health challenges that contribute to concerning colony losses—up to 90% annually for commercial beekeepers. A significant challenge is accurately identifying diseases affecting bee larvae, particularly distinguishing between bacterial infections like European Foulbrood (EFB) and similar-looking viral infections. Accurate diagnosis of honey bee brood diseases requires years of specialized experience. When faced with uncertainty, beekeepers often resort to widespread antibiotic treatment across entire apiaries. This practice can promote antibiotic resistance among pathogens, disrupt beneficial gut bacteria essential for bee health, increase susceptibility to opportunistic infections, and be completely ineffective when the actual problem is viral rather than bacterial. The Accomplishment: We developed a proof-of-concept artificial intelligence system that can distinguish between bacterial EFB and viral infections by analyzing images of infected bee larvae. Our approach used deep learning and transfer learning techniques, combined molecular testing with image analysis to create an accurately labeled dataset, achieved 73-88% accuracy in diagnosing the correct disease type, and revealed specific visual features the AI uses to make its diagnoses. The Contribution: This AI diagnostic tool could significantly improve honey bee health management by reducing unnecessary antibiotic use in beekeeping, slowing the development of antibiotic resistance, preserving beneficial microbiome communities in honey bee colonies, providing beekeepers with faster, more accurate diagnoses, and creating a foundation for expanded disease detection capabilities.

Technical Abstract: Honey bees are essential pollinators of ecosystems and agriculture worldwide. With an estimated 50–80% of crops pollinated by honey bees, they generate approximately $20 billion annually in market value in the U.S. alone. However, commercial beekeepers often face an uphill battle, losing anywhere from 40 to 90% of their hives yearly, often by brood diseases caused by bacterial, viral, and fungal pathogens. Accurate diagnosis of brood diseases, especially distinguishing European Foulbrood (EFB) from viral infections with a superficial resemblance to EFB (EFB-like disease), remains challenging. Incorrect diagnoses often lead to prophylactic antibiotic treatment across entire apiaries, exacerbating antibiotic resistance, disrupting native gut microbiota, and increasing susceptibility to opportunistic pathogens. Correct field diagnosis of brood disease is challenging and requires years of experience to identify and differentiate various disease states according to subtle differences in larval symptomology. To explore the feasibility of an image-based AI diagnosis tool, we collaborated with apiary inspectors and researchers to generate a dataset of 2,759 honey bee larvae images from Michigan apiaries, molecularly verified through 16 S rRNA microbiome sequencing and qPCR viral screening. Our dataset included EFB cases and viral infections (ABPV, DWVA, and DWVB), which were augmented to 8,430 and 8,124 images, respectively. We leveraged transfer learning techniques, fine-tuning deep convolutional neural networks (ResNet-50v2, ResNet-101v2, InceptionResNet-v2) pre-trained on ImageNet to discriminate between EFB and viral infections. Our proof-of-concept models achieved 73–88% accuracy on the training/validation sets. When tested on an independent dataset from Illinois containing additional viral pathogens not present in training data, the models showed higher accuracy for EFB (72–88%) than viral infections (28–68%), highlighting both the promise and current limitations of this approach. Implementing AI-based diagnostic tools can reduce unnecessary antibiotic treatments and help maintain the microbiome integrity critical to colony health. However, expanding training datasets to include all major pathogens, healthy larvae, and diverse geographic regions will be essential for developing field-ready diagnostic tools.