Stories from the University of Cambridge

COVID-19 Sounds App

  • Cecilia Mascolo[1], Pietro Cicuta[2], Andres Floto[3], Chloë Brown[1], Jagmohan Chauhan[1],[4], Jing Han[1], Andreas Grammenos[1], Apinan Hasthanasombat[1], Dimitris Spathis[1], Tong Xia[1], Ting Dang[1], Erika Bondareva[1]

    1 Department of Computer Science and Technology, University of Cambridge, Cambridge, UK 2 Department of Physics, University of Cambridge, Cambridge, UK 3 Department of Medicine, University of Cambridge, Cambridge, UK 4 ECS, University of Southampton, Southampton, UK

  • 2020

  • Cecilia Mascolo:
    cm542@cam.ac.uk

    Tong Xia:
    tx229@cam.ac.uk

  • Han J, Xia T, Spathis D, Bondareva E, Brown C, Chauhan J, Dang T, Grammenos A, Hasthanasombat A, Floto A, Cicuta P, Mascolo C. 2022. Sound of COVID-19: exploring realistic performance of audio-based digital testing. npj Digital Medicine 5, 16.

  • https://covid-19-sounds.org/en/

  • Cystic Fibrosis Trust (UK), European Research Council (European Union)

ABOUT THE OPEN-RESOURCE

Background 

Affordable, easily accessible, and accurate screening tools to detect human physiological signals (such as cardiac or respiratory signals) is an ultimate goal in medical care. Prof Cecilia Mascolo, head of the Mobile Systems Research Laboratory, at the Department of Computer Sciences and Technology, has been demonstrating how audio collected by portable devices widely used by most parts of the world population can aid the diagnostics of some diseases. Prof Mascolo’s research group was applying these concepts especially on heart disease diagnostics, when the COVID-19 pandemic happened. They took it as an opportunity to explore if it would be possible to use cough, breathing and voice to detect Coronavirus disease. COVID-19 Sounds App was then developed for audio-based digital testing of COVID-19.

Function

COVID-19 Sounds App is used to collect the sounds of voice, breathing and cough from the participants, to inform the diagnosis and disease stage of COVID-19 by developing algorithms; and to understand the difference between the sounds of COVID-19 and other respiratory illnesses.

Development process

The app development started during lockdown in 2020. Prof Mascolo points out that the app development per se was not too complicated as her group had backbone apps to work with. The difficulty was to release an app with COVID-19 in its title during a pandemic, where Apple and Google were very careful on what they allowed to be released. “It took us weeks or months to go through the approval of Google and Apple for the app. Some colleagues inside the University of Cambridge and the Department of Public Health helped with letters, supporting our research initiative,” says Prof Mascolo. 

IMPACT

Current use

Although the code of the COVID-19 Sounds App is shared, the metadata related to the collection is the most successful point about it. The data collected through the COVID-19 Sounds App has already been shared with more than 300 institutions. Data analysis shows the potential of machine learning for respiratory disease detection as well and its progression and evolution within individuals.

Successful stories 

The numbers that COVID-19 Sounds App has already generated are impressive. It is a dataset crowd-sourced from 36,116 participants, consisting of 53,449 audio samples (over 552 hours in total). This dataset is comprehensive in terms of demographics and spectrum of health conditions and also provides participants’ self-reported COVID-19 testing status with 2,106 samples tested positive.

Open source choice

“We are a university and I think that open source should be our model, and in particular for COVID-19, as it was an international endeavour. It was almost a no brainer that it had to be done in an open way,” says Prof Mascolo. 

GOING FORWARD - WHERE TO IN THE NEXT 3-5 YEARS?

At the moment, the Mobile Systems Research group is working towards a data collection for disease progression. Some of the app users are providing more than one sound sample, so it is possible to understand how they are getting out of COVID-19 infection. Also, there are other institutions using the dataset to work on different diseases.

COVID-19 Sounds App sound collection page. © 2020, Mobile Systems Research Laboratory, licensed under CC-BY 4.0. Reproduced from https://www.covid-19-sounds.org/en/.

COVID-19 Sounds App homepage. © 2020, Mobile Systems Research Laboratory, licensed under CC-BY 4.0. Reproduced from https://www.covid-19-sounds.org/en/ .