Covid-19 ‘Sounds’ app devised for audio detection of coronavirus
Developed by University scientists, the app aims to build predictive models which can contribute to the early diagnosis of Covid-19
Cambridge researchers have designed an app which aims to utilise machine learning algorithms and human sound data to inform diagnosis of the coronavirus.
Those who download the app will be asked to provide some basic demographic and medical history data, as well as voice samples generated by reading text on the devices screen and some breathing and coughing sounds. The app will then ask users if they have tested positive for the virus, before collecting one location sample.
The app does not track the movement of users, and only collects data when actively interacted with. Data is then stored on University servers and is used only for research purposes.
The app will not give medical advice and any reports of symptoms will not be responded to by medical assistance.
So far, there have been about 7,000 unique users, and of these around 100 have declared a positive test result for coronavirus.
Clinicians and epidemiologists internationally are said to have taken an interest in the data, which researchers hope to be able to share after establishing legal agreement between institutions.
“There are other groups around the world that have contacted us because they are looking at the same problem. Most of these have clinicians in their teams and are hoping to see something in these sorts of approaches”, explained Professor Cecilia Mascolo from Cambridge’s Department of Computer Science and Technology, who led the development of the app.
When asked what the place of the Sounds app might be alongside governmental tracing apps such as that launched for trial this week on the Isle of Wight, Professor Mascolo added that “after we analyse the data we might be able to see if our models could be used as pre-screening for covid symptoms”.
This could then “advise someone to have a test for example” or “advise someone they need hospital assistance at that point as they are deteriorating”.
The dataset could also help shed light on the progression of the disease, and Proffessor Mascolo’s team hopes to release the dataset to other researchers after the initial analysis and preprocessing.
The Covid-19 Sounds App is available now on all Android devices, as well as in web app form for Chrome and Firefox browsers. An iOS version has also been submitted and is awaiting approval.
- Features / Noiseless noise: headphone culture at Cambridge28 January 2025
- Arts / Grammar schools, Oxbridge, and The History Boys28 January 2025
- Fashion / Warm and well-dressed: a guide to the winter coat27 January 2025
- Features / The etiquette of inequality at Cambridge: making tradition inclusive24 January 2025
- Comment / Why university rankings don’t add up 24 January 2025