Donating your cough to fight COVID-19

A rapid test for the coronavirus that YOU can help develop

Image source: Mohamed Hassan from Pixabay

As the number of cases of COVID-19 declines in the UK and lockdown measures are eased, it’s becoming increasingly important to identify people with the virus to help contain its spread and prevent a second peak of infections. A common approach in many countries around the world has the development of track and trace apps, that aim to selectively quarantine people who have been in contact with coronavirus-positive individuals. However, these technologies have hit a number of stumbling blocks, from issues with privacy and data protection to a lack of uptake. Moreover, all these approaches rely on someone displaying symptoms to self-report them, get an antibody test, and then wait for the results. In the meantime, the risk of infecting others continues. However, a company in the UK is now proposing an alternative, that could diagnose COVID-19 patients faster than traditional testing, as well as track its spread.

Novoic is the company proposing this new approach. Founded by Oxford and Cambridge graduates with a background in neuroscience and machine learning, Novoic is a digital biotechnology company initially set up to develop algorithms that could detect neurodegenerative diseases from just the voices of patients. The idea behind this approach came from a study of nuns conducted in the 1990s, that was able to detect deterioration in language ability in these elderly women through the letters and diary entries they wrote [1].

Novoic takes this a step further. Their technology is based on experimental data showing that parameters in speech, including sentence structure and word use, as well as the audio components of the sounds you produce, change as you develop a neurological condition [2–4]. They use deep neural networks, computational linguistics and audio processing techniques to train their algorithms to identify key features that differ between the natural speech of people with a condition and healthy, age-matched controls, to help identify patients that may have the early stages of a disease. This technology can then be applied to a host of disorders, including dementia, Parkinson’s disease and motor disorders, as well as mental health conditions such as depression, based on the unique differences in speech production that these conditions create.

So how has this company gone from designing apps to detect Alzheimer’s disease to identifying COVID-19 cases? Well, the principles behind the projects are very similar. Respiratory diseases also change how you speak — although the words and sentences you form may not change, the condition of your airways will alter the airflow through your larynx and hence subtly change the sound you produce. When compared to people without respiratory disorders, a number of different conditions can be diagnosed, all based on the changes they make to your voice. Colds, asthma, influenza and pneumonia have all been shown to have unique auditory signatures that can be identified from speech (see the Novoic COVID-19 page for more details about detecting respiratory diseases from voice samples).

And COVID-19 is no different. CT scans in patients diagnosed with COVID-19 have shown that the infection specifically targets the periphery of the lungs, altering the airflow differently than if the centre of your lungs was affected. Moreover, COVID-19 is thought to thicken the lining of your lungs, and even create scar tissue with bad infections, that will all alter your ability to take in and pass out air, altering your voice [5]. The programmers at Novoic are now training their algorithms to detect the specific auditory outputs from patients with coronavirus, specifically how the virus changes the sound of your cough, in the hope that they can accurately diagnose people from home in the very near future.

And this is where we, the general public, come in. Novoic wants as many people as possible to donate their cough, to help provide enough data to train the algorithm to identify the sound of a COVID-19 cough. Following a questionnaire about whether you have previously had or currently have symptoms of coronavirus, your age and general respiratory health, you are simply asked to provide three coughs and a few simple sounds, to get a recording of airflow through your lungs.

And that’s it! Just five minutes will provide invaluable data to the research team to help perfect their ability to detect the features of a coronavirus cough. They need healthy controls just as much as people diagnosed with COVID-19, so anyone can donate their cough!

If you’d like to help out, here’s the link to learn more about the project, and to sign up to donate your cough:

Although this tool is still in development, and hence cannot yet diagnose COVID-19 without an antibody test, they hope that this technology will soon be able to accurately detect people with coronavirus and hence track the spread of the virus in the local population. Unlike track and trace systems, it doesn’t rely on complex laboratory testing, so could be a cheap but effective way to test huge numbers of people in the UK and beyond.

Donate your cough today to play your part in stopping the spread of COVID-19!

This article was written following the webinar: ‘Rapid test for COVID-19: Screen an entire nation using cough recordings’, presented by Emil Fristed and hosted by Oxford University Pharmacology Society on 18th June 2020. A recording of the talk can be found here.


[1] Snowdon, David A. Aging and Alzheimer’s Disease: Lessons From the Nun Study, The Gerontologist. 1997; 37(2):150- 156,

[2] Blair M, Marczinski CA, Davis-Faroque N, Kertesz A. A longitudinal study of language decline in Alzheimer’s disease and frontotemporal dementia. J Int Neuropsychol Soc. 2007;13(2):237–245. doi:10.1017/S1355617707070269

[3] Ahmed S, Haigh AM, de Jager CA, Garrard P. Connected speech as a marker of disease progression in autopsy-proven Alzheimer’s disease. Brain. 2013;136(Pt 12):3727–3737. doi:10.1093/brain/awt269

[4] Low DM, Bentley KH, Ghosh SS. Automated assessment of psychiatric disorders using speech: A systematic review. Laryngoscope Investig Otolaryngol. 2020;5(1):96–116. Published 2020 Jan 31. doi:10.1002/lio2.354

[5] Pan F, Ye T, Sun P, et al. Time Course of Lung Changes at Chest CT during Recovery from Coronavirus Disease 2019 (COVID-19). Radiology. 2020;295(3):715–721. doi:10.1148/radiol.2020200370