Despite the significant amount of work focusing on it, Bovine respiratory disease (BRD) remains one of the biggest issues on both beef and dairy operations. Part of the challenge is that BRD isn’t a single disease with a single cause. Instead, it develops through a complex interaction of viruses, bacteria, stress, immunity and environmental factors.
“It’s the leading cause of disease in post-weaned calves. We know that it’s also the leading killer of post-weaned calves,” says Melissa Cantor, assistant professor of precision dairy science at Penn State University and one of the most recent guests on “The Bovine Vet Podcast.”
“It’s the second-leading cause of disease in our pre-weaned animals, and the leading reason for antibiotic use. We’re talking over 90% of cases identified by producers that are reported to the USDA being treated with some sort of antimicrobial or antibiotic. We know this is a disease we really need to manage for judicious antibiotic use and also for economic management.”
There is no shortage of tools to diagnose BRD once disease becomes apparent. Clinical scoring systems and thoracic ultrasound have improved detection, but both rely on calves reaching the point where illness can be recognized. Identifying those animals also depends heavily on the experience of the people observing them every day, making early detection as much an art as a science.
That challenge led Cantor to ask a deceptively simple question: Could calves be telling us they were getting sick long before anyone realized it?
One calf left just a little milk in her bottle. Another spent a little more time lying down than usual. A third seemed just a bit slower getting up to eat. None of those changes would make most calf managers diagnose BRD, but after spending countless hours caring for dairy calves during her PhD research, Cantor couldn’t shake the feeling that those tiny behavioral shifts were trying to tell her something.
That curiosity became the foundation of the CalfHealth system, an interdisciplinary research project at Penn State University that combines animal behavior, precision livestock technologies and artificial intelligence to identify calves at risk for BRD before they appear sick. By recognizing subtle deviations from normal behavior, researchers hope the system will eventually give veterinarians and producers an earlier opportunity to intervene while supporting more judicious antimicrobial use.
Why Early BRD Detection Remains So Difficult
CalfHealth is built around the idea that calves often exhibit behavioral changes before clinical disease becomes apparent. The challenge is recognizing those changes consistently enough to act on them.
The idea behind the project didn’t begin with artificial intelligence, but with an observation.
While managing calves during her doctoral research at the University of Kentucky, Cantor found herself repeatedly noticing calves that simply seemed “off.” They weren’t clinically sick, but something about their behavior suggested they were heading in that direction. Rather than dismissing those observations as intuition, she decided to test them.
“We were seeing that these calves were oftentimes declining in behavior before they were clinically sick,” Cantor says. “We knew that we could capture sick transition cows with rumination collars and activity changes, so we wondered if we could do this for calves. It seemed like a calf would be just a little bit off. She might start coughing, but coughing by itself isn’t diagnostic of respiratory disease. She might be off feed a little bit or a little more lethargic. As a manager, I was thinking, ‘That calf is getting respiratory disease. Something’s wrong,’ but she still wasn’t positive by clinical signs.”
Rather than relying on instinct alone, Cantor and her colleagues launched a longitudinal study following calves through their first 90 days of life. They continuously monitored calf behavior while using thoracic ultrasound to identify the onset of clinical pneumonia, allowing them to compare subtle behavioral changes with what was happening inside the lungs.
The findings opened the door to something Cantor hadn’t originally set out to build.
“What we saw was that those calves all had these associations with changes in their behavior prior to pneumonia. What was even more fascinating was when we worked with a computer scientist and looked at those patterns retrospectively, we found they were actually useful for an AI tool,” she says.
What started as a question about calf behavior quickly became an interdisciplinary collaboration. The subtle changes Cantor had been noticing weren’t simply observations anymore. They were measurable patterns artificial intelligence could recognize, creating the possibility of identifying at-risk calves days before traditional diagnosis.
How the CalfHealth System Uses AI to Detect BRD Earlier
Today, CalfHealth has evolved into a multidisciplinary effort involving veterinarians, animal scientists, computer scientists and sociologists. Each discipline contributes a different perspective, reflecting Cantor’s belief that solving complex health challenges requires more than advances in veterinary medicine alone.
Rather than searching for one perfect technology, the team is evaluating how multiple streams of information can work together to build a more complete picture of calf health:
- Accelerometers measure activity, movement and lying behavior. Feeding technologies monitor milk intake, drinking speed and visits to automated feeders.
- Clinical observations, body weight and thoracic ultrasound provide additional information about each calf’s health status.
- Wi-Fi sensing technology that could detect coughing within a group of calves by analyzing subtle changes in wireless signals around a water station, potentially offering a low-cost method for monitoring respiratory disease in group-housed calves.
Modern dairy farms already generate enormous amounts of information. The challenge isn’t collecting more data; it’s understanding which pieces matter and how they fit together. Artificial intelligence can analyze those combined data streams, recognizing patterns that would be difficult for even experienced calf managers to detect consistently across hundreds or thousands of animals.
Cantor says collaborating with computer scientists has changed the way she thinks about calf health. While they focus on algorithms and model performance, she keeps bringing the discussion back to the animal. That balance has become one of the project’s greatest strengths, ensuring every technological advance answers a practical question: Can it help someone recognize a sick calf sooner?
“The calf might be a little off. With my trained eye, I see that,” Cantor says. “But the technology can quantify it. It’s like when a calf rejects its bottle and there’s that last little 0.2 liters inside. No one’s going to record that. But the algorithms can pick up on that. The technology can quantify those small changes in that individual calf.”
Cantor laughs as she talks about the project.
“I’m very passionate about this topic, in case you can’t tell,” she says.
For her, the excitement isn’t really about artificial intelligence. It’s about giving veterinarians and calf managers another opportunity to notice the animal that’s quietly beginning to struggle.
How AI Supports Veterinary BRD Diagnosis
Despite the project’s focus on artificial intelligence, Cantor is quick to emphasize that CalfHealth is designed to support veterinary decision-making, not replace it.
“The goal of this technology is not to replace any veterinarian,” she says. “We’re really understaffed, especially in large animal practice. The idea here is to provide more information to the day-to-day management of farms because a veterinarian can’t be there every day. It’s another sentinel, another tool that helps us monitor calves.”
Instead of generating automatic treatment decisions, the system is intended to identify calves that deserve a closer look, giving another layer of decision support alongside clinical examinations and herd observations. As labor shortages continue to challenge dairy operations, tools that help prioritize which animals require attention may become increasingly valuable.
Could Earlier BRD Detection Reduce Antibiotic Use?
For Cantor, finding sick calves sooner isn’t the finish line. It’s the first step toward giving veterinarians more options.
“We need to be very careful about using the antimicrobials we have because we don’t have a lot of new classes coming,” she says. “If there’s some way to save calves from entering the pipeline of needing antibiotics, I think that’s awesome. There are a lot of ideas about what we could do with those animals before they reach that point.”
Those ideas remain the subject of future research, and Cantor is careful not to suggest earlier detection automatically means earlier antibiotic treatment. Instead, she believes improved disease recognition could eventually create opportunities to better understand disease progression and evaluate supportive interventions before bacterial pneumonia becomes firmly established.
CalfHealth is now being evaluated on commercial dairy farms, where researchers are determining which combination of technologies provides the greatest value under real-world conditions. Although considerable work remains before the system could become widely available, Cantor already has a clear vision of what success would look like.
“In my perfect world, you have a sensor on a calf when it’s born, or a system in the barn, and it’s helping you manage those animals throughout their lives,” she says. “You’re still going to have your veterinarian changing protocols and making treatment decisions, but you’re using that as a monitoring tool to help make better farm decisions. My biggest goal is for us to be more judicious about antimicrobial use because we don’t have a lot of them in our arsenal.”
Artificial intelligence, accelerometers and Wi-Fi sensing are simply the tools. The real goal is giving veterinarians and producers more time: more time to recognize a calf beginning to decline, more time to intervene before disease progresses and more opportunities to preserve the effectiveness of the antimicrobials already available.
If the CalfHealth system succeeds, it won’t be because it replaces experienced cattle people. It will be because it helps them recognize the same subtle behavioral changes that first caught one curious researcher’s attention years ago, turning those nearly imperceptible signals into practical information that helps veterinarians and producers intervene earlier.
To hear more about CalfHealth and how Cantor hopes to improve BRD detection in dairy calves, check out the latest episode of “The Bovine Vet Podcast.” This episode also features Dr. Paige Schmidt-Rios discussing the current state of BRD detection in beef systems.
