Screening for COVID-19 infection needs no introduction at the time of writing. It is a process of importance during the current pandemic, both in healthcare and state policy. Confirmation of COVID-19 infection is currently based on the reverse transcription-polymerase chain reaction (RT-PCR) tests. These tests are fairly rapid and accessible. As a result, they are widely used. Research evidence indicates that the sensitivity of the test is about 80% after symptoms appear, but 62% on the first symptomatic day, and just 33% on the fourth day after exposure. Better outcomes might follow from supplemental screening and detection, as well better approaches that could allow for large scale screening with only minimal input. Two emerging approaches are the cough audio file, captured from a phone voice recording, and the use of sniffer dogs.
COVID-19 screening using audio samples
In the UK, Han et al. (2021) developed a prototype of a voice-based approach. Their app was evaluated using 343 participants. The combination of voice signals and symptom reporting resulted in an estimated accuracy of 79%. Recent research in Argentina by Pizzo, Esteban, and Scetta (2021) went even further. Pizzo et al. described the use of audio recordings of coughing and machine learning based analysis to develop a cough classifier. The accuracy rating of about 88% was assessed using RT-PCR tests, of more than 2,000 multi-site subjects r. Research is ongoing across the globe, including competitive challenges such as the INTERSPEECH 2021 Computational Paralinguistics Challenge.
COVID-19 screening using sniffer dogs
There have been various suggestions in relation to the use of sniffer dogs for the identification of individuals infected with COVID-19, with ongoing research worldwide, including Italy, India, Iran, Finland, France, the UK, and elsewhere. Sniffer dogs have already been trained to identify various cancers and states such as hypoglycemia. A pilot project using sniffer dogs to screen passengers for COVID-19 began in airports in Finland in the fall of 2020. In Iran, Eskandari et al. (2021)conducted a study where they provided training to sniffer dogs, and then tested the accuracy of the canine identification of COVID-19 infection. This study used two types of experiments, one using pharyngeal secretions, and the other using masks and clothing of the human sample. The dogs were more accurate in relation to the use of masks and clothes. In both tests, the dogs had a higher accuracy rate in relation to true negatives, with sensitivity between 65% to 86%, and sensitivity of 89% to 92.9% across experimental conditions. The dogs, in other words, were much more accurate than the RT-PCR.
It remains to be seen whether high levels of accuracy will lead to large scale dissemination of such tools. A particular concern are the regulatory approvals required in many countries. Sniffer dogs, with a high rate of accuracy, could easily become features in airports as part of primary outbreak prevention. It is also plausible that they might be used as a security tool for screening at large scale events. These are issues that are not entirely clear. This reflects the potential for fast-tracked or interim policy level approval. After all, sniffer dogs provide more accurate identification of COVID-19 than the current diagnostic approach, the RT-PCR. The way forward is to use many and different modes of screening for the earlier identification of COVID-19 infection.
Dive in to the research
Else, H. (2020). Can dogs smell COVID? Here’s what the science says. Nature, 587(7835), 530-531. https://www.icpcovid.com/sites/default/files/2021-03/Ep%20117-11%20Can%20dogs%20smell%20COVID_%20Here%E2%80%99s%20what%20the%20science%20says.pdf
Eskandari, E., Marzaleh, M. A., Roudgari, H., Farahani, R. H., Nezami-Asl, A., Laripour, R., … & Shiri, M. (2021). Sniffer dogs as a screening/diagnostic tool for COVID-19: a proof of concept study. BMC infectious diseases, 21(1), 1-8. doi: 10.1186/s12879-021-05939-6 https://link.springer.com/article/10.1186/s12879-021-05939-6
Grandjean, D., Sarkis, R., Tourtier, J. P., Julien, C., & Desquilbet, L. (2020). Detection dogs as a help in the detection of COVID-19: Can the dog alert on COVID-19 positive persons by sniffing axillary sweat samples? Proof-of-concept study. PLOS ONE. doi: 10.1371/journal.pone.0243122 https://www.biorxiv.org/content/10.1101/2020.06.03.132134v1.abstract
Han, J., Brown, C., Chauhan, J., Grammenos, A., Hasthanasombat, A., Spathis, D., … & Mascolo, C. (2021). Exploring Automatic COVID-19 Diagnosis via voice and symptoms from Crowdsourced Data. arXiv preprint arXiv:2102.05225. https://arxiv.org/abs/2102.05225
Kmietowicz, Z. (2020). Sixty seconds on… covid-19 sniffer dogs. British medical journal 2020(370). doi: 10.1136/bmj.m3758 https://www.bmj.com/content/370/bmj.m3758.short
Pizzo, D. T., Esteban, S., & Scetta, M. (2021). IATos: AI-powered pre-screening tool for COVID-19 from cough audio samples. arXiv preprint arXiv:2104.13247. https://link.springer.com/article/10.1007%2Fs11017-017-9402-3
Schuller, B. W., Batliner, A., Bergler, C., Mascolo, C., Han, J., Lefter, I., … & Kaandorp, C. (2021). The INTERSPEECH 2021 Computational Paralinguistics Challenge: COVID-19 cough, COVID-19 speech, escalation & primates. arXiv preprint arXiv:2102.13468. https://arxiv.org/abs/2102.13468