The problem with current detection methods is timing. The standard techniques — alcohol wash, sticky board counting, and manual inspection — are invasive, labor-intensive, and deliver results one to three days after sampling. By then, mite populations may have grown past the point where low-impact treatments remain effective. When infestations are caught late, beekeepers often resort to synthetic miticides such as amitraz or coumaphos, which accumulate in beeswax, carry documented resistance risks in multiple countries, and can harm colony health. When mite loads are identified while still low, beekeepers can instead turn to organic acids, which at correct dosage leave no honey residue and have no documented resistance. The economic threshold for action is 2 to 3 mites per 100 adult bees; an alert issued at 1 to 2 mites per 100 bees gives a beekeeper roughly four to six days before that threshold is crossed.

Previous automated approaches have each tackled one piece of the problem. A vision-based system by Bjerge and colleagues achieved 91 percent classification accuracy using a convolutional neural network on hive entrance video, but required a custom GPU workstation and a dedicated optical tunnel, pushing total cost above $500. A microscopy-based method by Babic and colleagues reached 94 percent accuracy but requires physical bee samples and offline laboratory processing. Acoustic research showed that varroa-stressed colonies produce measurably different sound signatures, particularly in the 200–500 Hz band corresponding to disrupted wing-beat and fanning behavior, and environmental studies demonstrated that carbon dioxide concentration and temperature at the hive entrance correlate with mite-induced stress. The new system is, to the author’s knowledge, among the first to fuse all three sensing modalities in a single, affordable, edge-deployed device.

The hardware is deliberately modest. A Raspberry Pi 5 runs all inference locally, paired with a 12-megapixel Arducam mounted horizontally at the hive entrance, an I2S MEMS microphone, and an SCD41 sensor that measures true carbon dioxide using non-dispersive infrared technology rather than cheaper proxy estimates. A custom PCB routes the sensor buses, and a battery-backed power module keeps the unit running through outages. The complete bill of materials came to $189.68, and a proposed mass-production redesign built around a Raspberry Pi Zero 2 W would cut that to an estimated $49.50. The device needs no internet connection, storing records on a microSD card and serving a dashboard to any browser on the local Wi-Fi network.

The software pipeline runs a four-stage inference cycle every 30 seconds. Stage A is a YOLOv8n bee detector that draws bounding boxes around every bee in each video frame, achieving a mean average precision of 0.976 at 512-by-512 resolution. Stage B is a second YOLOv8n model that inspects each bee crop for visible mites — a crop-first architecture that proved essential, because at full-frame resolution a mite occupies roughly 0.3 percent of the image, making direct detection unreliable. Cropping first lifted the F1 score from 0.774 to 0.805. Stage C is a CNN-LSTM that classifies 30-second audio clips as healthy or stressed, using 43 acoustic features including MFCCs, spectral centroid, and chroma features; an earlier 157-feature version overfit at just 74 percent accuracy before the streamlined feature set raised test accuracy to 93.6 percent. Stage D is a Random Forest that classifies colony risk from temperature, humidity, and carbon dioxide readings.

The fusion layer is where the system earns its keep. Each modality’s probability score is combined through confidence-weighted voting, with weights proportional to each sensor’s validation precision: 0.28 for vision, 0.34 for audio, and 0.38 for environmental data. Because the three signals arise from biologically independent mechanisms — visible mites on bee bodies, altered colony acoustics, and metabolic drift in hive air — spurious readings in one modality are unlikely to coincide with spurious readings in the others. The result is dramatic: the false alarm rate falls from 27.0 percent for vision alone to 0.5 percent for the fused system, a reduction confirmed by McNemar’s test at p = 0.0176. Fused precision reached 0.995 with an F1 score of 0.957, and removing any single modality degraded performance, confirming that each contributes independently.

Field validation spanned nine colonies in three infestation categories, three hives each, with mite loads confirmed by alcohol wash sampling. The three healthy hives ran for 30 days apiece and produced median fused scores between 0.0196 and 0.0217, with zero threshold-positive windows and zero alert episodes across 12,920 total observation cycles. Three mildly infested hives carrying 1 to 5 mites per 100 bees also ran 30 days each; their median scores shifted upward to between 0.0759 and 0.0889 without ever crossing the 0.50 alert threshold, showing the system responds to sub-threshold mite pressure without crying wolf. Three heavily infested colonies, each carrying 8 to 10 or more mites per 100 bees, were monitored for one day and produced medians above 0.51, with 34 to 38 alert episodes per hive and threshold exceedances in roughly three-quarters of all inference cycles.

The 25-fold increase in median fused score from healthy to heavily infested colonies, with the mild category falling neatly between, indicates the system tracks infestation severity rather than behaving as a binary switch. A separate 50-hive, 14-day simulation estimated that temporal smoothing detected more than 80 percent of developing infestations before their peak, with a median lead time of 4.6 days — enough to preserve the beekeeper’s full range of treatment options. The system also survived robustness testing under low light, high noise, extreme temperature, and partial sensor failure, averaging 0.875 accuracy across all stress conditions. The full pipeline sustained 28 to 29 frames per second on the Raspberry Pi 5 with power draw under 5 watts and no thermal throttling.

The study is candid about its limitations. All nine hives sat at a single apiary over a single season, so generalization across regions, bee subspecies, and climates awaits multi-site validation. The environmental classifier was trained on a synthetic dataset built from published biological parameter ranges, a workaround made necessary by a striking methodological discovery: the only publicly available labeled real-sensor hive dataset contained a leaky column — cumulative bee weight — that carried 45.7 percent feature importance but cannot be measured at inference time. Any model trained on it without removing that column reports a suspicious 100 percent accuracy and would fail in the field. No controlled infestation experiments were conducted, for ethical reasons; all infestation levels reflect naturally occurring conditions.

What distinguishes this work is less any single algorithm than the demonstration that deployment-grade reliability is achievable at a price individual beekeepers can afford. Most of the world’s beekeepers manage fewer than 25 colonies, and no commercially available system previously offered continuous, non-destructive varroa monitoring below $200 per hive. The complete inference pipeline, trained models, and processed field datasets have been released publicly, and the modular architecture — camera, microphone, environmental sensor, late fusion — extends naturally beyond varroa to livestock respiratory disease monitoring, crop pest surveillance, and coral reef health. For a parasite that has spread to every inhabited continent since jumping from its original Asian honey bee host in the 1970s, and against which the western honey bee has evolved no effective defenses, an affordable early-warning system may be the most practical lifeline apiculture has seen.

Subject of Research: Non-invasive multimodal machine learning detection of Varroa destructor infestations in honey bee colonies

Article Title: A multimodal machine learning system for non-invasive detection of varroa destructor infestations in honey bee colonies

Article References: Gaba, P. (2026). A multimodal machine learning system for non-invasive detection of varroa destructor infestations in honey bee colonies. Smart Agricultural Technology, 15, Article 102354. https://doi.org/10.1016/j.atech.2026.102354

Image Credits: AI Generated

DOI: 10.1016/j.atech.2026.102354

Keywords: varroa destructor, honey bees, machine learning, sensor fusion, precision beekeeping, edge computing, colony collapse, acoustic monitoring, computer vision, Raspberry Pi, apiculture, early detection

The central obstacle the team faced was the sheer complexity of the honey bee’s sense of smell. Bees rely on more than 200 odor receptors to navigate their world, detecting and responding to a vast array of volatile compounds in their environment. That sensitivity is what makes bees such effective foragers, but it also makes their olfactory behavior extraordinarily difficult to predict. Finding a scent that reliably repels a bee, rather than attracting it or doing nothing at all, has long been considered a major challenge. Anandasankar Ray, a professor of molecular, cell and systems biology at UC Riverside and an expert on insect olfactory behavior, led the interdisciplinary effort to overcome it.

“Bees rely heavily on their sense of smell to forage, but that sensitivity makes it tough to find odors that push them away instead of drawing them in,” Ray said. “Our goal was to flip that script and find a way to use scent as a deterrent — safely and effectively.” To pursue that goal, Ray’s team collaborated with honey bee researchers in the laboratory of Boris Baer, a professor of entomology at UC Riverside, combining expertise in computational modeling with hands-on knowledge of bee behavior.

The approach the researchers developed was a machine-learning model trained on two kinds of information: the chemical structures of odorant molecules and previously recorded behavioral responses of bees to those odorants. By learning the relationship between molecular features and behavioral outcomes, the model could begin to predict how a bee would respond to a compound it had never encountered. Crucially, the team did not stop with the initial training data. They refined the model with new behavioral data generated in their own laboratory, gathered from both honey bees and Drosophila, the fruit fly that serves as a workhorse of insect olfaction research. That iterative improvement step allowed the system to make increasingly accurate predictions of insect olfactory responses.

One of the most striking aspects of the work is that it succeeded without the enormous datasets usually assumed to be necessary for machine learning. “It is generally thought you need abundant data to do any kind of machine learning, but that’s not true for olfaction,” Ray said. “You simply need good-quality data and iterative improvement steps.” That insight has implications well beyond bee research, suggesting that carefully curated behavioral measurements can substitute for sheer data volume when the underlying structure of the problem is learnable. Once the model was optimized, the researchers put it to work on a truly industrial scale, screening a library of more than 50 million compounds. From that vast chemical space, the system identified roughly 130 candidates predicted to have strong potential as bee repellents.

Predictions, of course, mean little until they are tested against living insects. The team reports in eLife how they put the top-performing candidates through exactly that gauntlet. In laboratory assays, honey bees exposed to the candidate compounds exhibited clear avoidance behaviors, and those responses aligned closely with what the model had predicted. That agreement between computation and behavior is the heart of the study’s significance: it demonstrates that a model trained on limited but high-quality data can generalize to new molecules and correctly forecast how bees will react to them in the real world.

The validation did not stop in the laboratory. The researchers then moved to field experiments with freely foraging bees, a far more demanding test because wild conditions introduce wind, competing odors, and the full complexity of natural foraging behavior. All seven of the compounds tested in the field reliably repelled bees from honey combs, and importantly, they did so without harming the insects. A repellent that injured the very pollinators it was meant to protect would defeat its own purpose, so the demonstration that these odorants deter bees safely is a critical part of the result.

“This is a powerful demonstration of how machine learning can help solve real-world ecological problems,” Ray said. “By keeping bees away from harmful pesticides, we can potentially reduce their risk of exposure without compromising the protection of crops.” The logic of the approach is straightforward: instead of banning or restricting every pesticide that poses a risk to bees — a regulatory path that has already removed some of the most harmful products from use — farmers could blend repellent compounds into their pesticide applications. The crop would retain its protection against pests, while bees, sensing the deterrent odor, would simply choose to forage elsewhere. Unintended exposure to pesticides is widely considered a contributing factor to bee population decline and colony collapse, and minimizing bee contact with treated crops addresses that exposure directly.

The researchers also point to applications that extend well beyond agricultural spraying. “In certain public environments — hospitals, office buildings, and residential areas, for example — reducing the formation of beehives can help avoid human-bee conflicts,” Ray said. In those settings, a safe repellent could discourage bees from establishing colonies where they are unwelcome, without resorting to lethal control. There are agricultural uses as well: some farming practices, particularly with seedless fruit varieties, aim to avoid pollination altogether, and a reliable repellent could help keep bees away from crops where pollination is undesirable rather than essential. According to Ray, the research is a step toward developing bee-friendly pesticide formulations — products that safeguard pollinators while still meeting the needs of modern agriculture.

“Protecting pollinators doesn’t have to come at the expense of food security,” Ray said. “With the right tools, we can strike a balance — and this model helps us get there. We believe our work will help guide further exploration of machine learning-guided solutions in environmental protection and sustainable farming practices.” Ray and Baer were joined in the research by Joel Kowalewski, Barbara Baer-Imhoof, Tom Guda, Matthew Luy, and Payton DePalma. The work was funded by a grant from the California Research Alliance by BASF. The research paper, titled “Machine learning of honey bee olfactory behavior identifies repellent odorants in free flying bees in the field,” was published in eLife on 23 September 2026, with the DOI 10.7554/eLife.104831.3. The authors note that Ray is founder and president of Sensorygen and Remote Epigenetics and holds equity in both companies, Kowalewski holds equity in Sensorygen, and several of the authors are inventors in a patent application covering the compounds discussed in the article. If the approach scales from honey combs to whole orchards, the humble trick of making treated flowers smell unappealing could become one of the more elegant tools in the effort to keep the world’s most important pollinators alive.

Subject of Research: Machine learning-guided discovery of odorant compounds that repel honey bees from pesticide-treated crops

Article Title: Machine learning helps identify chemicals that repel honey bees from pesticides

Article References: Machine learning helps identify chemicals that repel honey bees from pesticides. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: honey bees, machine learning, pollinator protection, pesticides, olfaction, repellents, eLife, UC Riverside, colony collapse, sustainable agriculture, chemical ecology, Drosophila