Trained sniffer dogs detect coronavirus with 94-percent success rate
Earlier this year, we looked at a research project in the UK where scientists were exploring how dogs could be trained to sniff out signs of the novel coronavirus. Similar ventures are now being carried out around the world, including one at Germany’s University of Hanover where a team found that with little training, sniffer dogs were capable of identifying positive samples with a high degree of accuracy.
The incredible sensitivity of dogs’ noses has seen scientists look to use them for all kinds of purposes, including detecting cancer, malaria, and explosive devices. By exposing the animals to samples in a room and teaching them to distinguish between those that are infected and those that aren’t, the hope is that dogs can become a powerful screening tool in public spaces to help slow the spread of COVID-19.
The University of Hanover scientists were investigating these possibilities with eight specialized sniffer dogs. The team spent one week training the animals to distinguish between samples infected with SARS-CoV-2, the coronavirus strain that causes the disease COVID-19, and uninfected controls.
Some 1,012 saliva and tracheobronchial samples were collected, with the dogs then tasked with identifying those that were infected with SARS-CoV-2. The samples were randomly distributed so neither the researchers nor the dog handlers knew which were positive.
The dogs correctly identified 157 positive samples and 792 negative samples, while incorrectly identifying 33 negatives and incorrectly rejecting 30 positives. All up, the team notes this makes for an average sensitivity (detection of positives) of 83 percent, an average specificity (detection of negatives) of 96 percent and overall average detection rate of 94 percent.
“The results of the study are incredibly exciting,” says study author Professor Holger Volk. “We have created a solid foundation for future studies to investigate what the dogs smell and whether they can also be used to differentiate between different times of illness or clinical phenotypes. ”
The research was published in the journal BMC Infectious Diseases, and you can hear from those involved in the video below.
Source: University of Hanover (German)