Dogs can spy on coronavirus with impressive accuracy

Dogs can spy on coronavirus with impressive accuracy

Many long for a return to a “normal” post-pandemic that can bring about some concerts, trips and large gatherings. But how can one be sure, given these potential public health risks?

According to a new study, dogs are one possibility. A proof-of-concept study published today in the journal PLOS ONE suggests that specially trained detection dogs can spy out COVID-19 positive samples with an accuracy of 96%.

“This is not a simple thing we ask of the dogs,” says Cynthia Otto, senior paper writer and director of the Working Dog Center at the University of Pennsylvania’s School of Veterinary Medicine. “Dogs need to be specific to recognize the smell of the infection, but they also need to generalize the background smells of different people: men and women, adults and children, people of different races and regions.”

In this first study, the researchers found that the dogs can, but training needs to be done with great care and, ideally, with lots of samples. The results feed into a further study, which Otto and colleagues have called the “T-Shirt Study”, in which dogs are trained to use the volatile organic compounds between the smells of COVID-positive, negative and vaccinated people to make a difference you go on a t-shirt that is worn overnight.

Get the BARK NEWSLETTER in your inbox!

Sign up and get answers to your questions.

“We collect a lot more samples in this study – hundreds or more – than in the first and hope that the dogs get closer to what they might encounter in a community,” says Otto.

Through the Working Dog Center, she and her colleagues have years of experience training dogs with medical detection, including those that can identify ovarian cancer. When the pandemic hit, they used that expertise to design a coronavirus detection study.

Staff Ian Frank of the Perelman School of Medicine and Audrey Odom John of the Children’s Hospital in Philadelphia provided SARS-CoV-2 positive samples from adult and pediatric patients, as well as samples from patients who tested negative, to serve as experimental controls . Otto worked closely with Penn Medicine’s coronavirus expert Susan Weiss to process some of the samples in Penn’s Biosafety Level 2+ lab to inactivate the virus so the dogs can sniff safely.

Due to job stoppages due to the pandemic, researchers did not work with dogs at Penn Vet, but instead worked with Pat Nolan, a trainer with a facility in Maryland.

The study used eight Labrador Retrievers and one Belgian Malinois that had not previously performed any medical identification work. First, the researchers trained them to recognize a distinctive scent, a synthetic substance known as a Universal Detection Compound (UDC). They used a “scent wheel” in which each of the 12 ports is loaded with a different sample, and rewarded the dog for responding to the port with UDC.

When the dogs responded consistently to the UDC odor, the team began training them to respond to urine samples from SARS-CoV-2 positive patients and differentiate positive from negative samples. The negative samples were subjected to the same inactivation treatment – either heat inactivation or detergent inactivation – as the positive samples.

The team processed the results, with the assistance of Penn criminologist and statistician Richard Berk, and found that all three dogs could easily identify SARS-CoV-2 positive samples with an average accuracy of 96% after three weeks of training. However, their sensitivity or ability to avoid false negative results was sometimes lower, the researchers believe because of the strict criteria of the study: If the dogs passed a port with a positive sample even once without reaction, this was marked a “miss” .

The researchers encountered many complicating factors in their study, such as the dogs’ tendency to differentiate between actual patients rather than between their SARS-CoV-2 infection status. The dogs were also shed from a sample from a patient who tested negative for SARS-CoV-2 but had recently recovered from COVID-19.

“The dogs reacted again and again to this test and we kept telling them no,” says Otto. “But obviously there was something else in the patient’s sample that the dogs typed on.”

The key lessons learned from the study, along with confirming that there is a SARS-CoV-2 odor that dogs can detect, that future training should involve a large number of different samples, and that dogs will not repeatedly train on samples from a single person should be.

“We can drive this forward not only in our COVID training, but also in our cancer work and all other medical detection efforts that we undertake,” says Otto. “We want to ensure that all steps are in place to ensure quality, reproducibility, validity and safety for operationalizing our dogs and to begin screening them in community settings.”