APG researchers, UPenn training dogs to detect asymptomatic coronavirus carriers; further testing coming

A trainer observes as a dog investigates a Training Aid Delivery Device. U.S. Army researchers from Aberdeen Proving Ground and the University of Pennsylvania’s Working Dog Center are training dogs to detect the coronavirus in humans.
A trainer observes as a dog investigates a Training Aid Delivery Device. U.S. Army researchers from Aberdeen Proving Ground and the University of Pennsylvania’s Working Dog Center are training dogs to detect the coronavirus in humans. (Courtesy photo U.S. Army)

U.S. Army researchers at Aberdeen Proving Ground and University of Pennsylvania’s Working Dog Center have set out to prove whether the nose really knows, training nine dogs to detect the coronavirus in humans — even asymptomatic patients.

The dogs began training just over the Maryland border in Greencastle, Pennsylvania, on May 26 as part of the proof of concept study. Researchers set out to see if the dogs could detect odors and byproducts the human body produces when infected with COVID-19, said Michele Maughan, a contract senior scientist with Excet Inc and researcher at APG’s Combat Capabilities Development Command Chemical Biological Center in the Edgewood area.


Then, they are exploring the feasibility of putting COVID-sniffing dogs out in the real world.

“This is absolutely a proof of concept study, but it is a proof of concept study we have a lot of confidence in,” Maughan said.


The study was funded through donations and completed last week. The program will take a short break before resuming in mid-September.

Researchers do not yet have the full results of the tests, but Maughan said they will be published soon. The next phase of the project will test if the dogs can detect odors from COVID-carriers’ clothing, Maughan said.

When the virus infects someone, their immune system responds before they begin to show symptoms, Maughan explained. That produces an odor and other biological reactions the dogs have picked up on, making them particularly useful for identifying incipient infections or asymptomatic carriers.

The study tasked nine dogs — eight Labrador retrievers and one Malinois — with identifying samples of saliva and urine from COVID-positive patients loaded in specialized containers and fitted to spokes of a rimless wheel. Some held neutral control substances; others held things like an open magic marker, food or a tennis ball. Originally, there were 10 dogs, but one dropped out of the study.

The dogs could get a good whiff of the samples without risk of contracting the virus because of the containers’ design. A specialized membrane allows the odor to seep out without risking infection. At no point were they exposed to the live virus.

Cynthia Otto, the director of the Penn Vet Working Dog Center, said the dogs were highly accurate depending on the metric used to evaluate them. The dogs were about 68% accurate in finding the odor at first pass on the wheel, but 96% accurate if they got another pass at the material. They had only a 1% false-alert rate, corresponding to only two instances of falsely identifying the neutral control substances, she said.

Otto said the prospect of using the dogs in a wider, real-world screening situation was exciting, but their noses’ insight also gives a point of reference to electrical and chemical engineers who are working to develop detection tools within their own disciplines.

“That has really always been our goal — to have the dogs give the information to refine the electronic and chemical sensing,” she said. “This is the quickest way to say ’Yes, that is an option.’”

Maughan said the dogs could be useful as a complement to other precautions like fever checks and screening questions in a real world scenario. Their real-world application can already be seen in the The United Arab Emirates, where they are used to identify COVID samples at Dubai International Airport.

For the dogs’ use in the real world, securing funding has been challenging, the scientists said; the effort does not clearly fall into traditional medical or technological funding silos. Though the work shows promise, some have been skeptical of whether it is worth the investment — even though bomb- and drug-detecting dogs are used around the world.

“For me there is a little bit of a disconnect there in, ’[I] wonder why one and not the other,” Maughan said.

Otto said that she has a long list of organizations that want a COVID-detecting dog. Still, finding funding for the study has been an issue. The Department of Defense will provide seed money for the next stage of the project, but the effort will need more to see through to completion.


“Everybody wants a dog, and nobody wants to pay for the science that go into it,” Otto said. “The dogs have some hope ... but that hope is empty if we do not have the data behind it.”

The scientists still need samples for next phase of the project. People 18 or older in the U.S. who have tested positive or negative for the virus and received their results back in 24 hours are eligible. After a quick health survey on the university’s working dog center’s website, eligible participants will be sent a cotton t-shirt to wear while sleeping and send back to the lab to be used in the test.

Recommended on Baltimore Sun