BENGALURU: A new tool could soon be added to the list of preliminary checks used to detect Covid symptoms: A mobile app that can listen to your cough and tell you whether you may be at risk of developing Covid-19. Developed by HCG Hospital and Wipro, the Covigilent app uses deep learning – a field within machine learning that uses algorithms inspired by the structure and function of the brain – to process up to 1,000 samples per second and produce results within five seconds.
The developers claim the app – which can be downloaded free – has an accuracy of 84%. They have approached the state government seeking permission to use the app in schools and offices so that cases can be detected early. “A smart phone would serve as a recording device and the risk categorisation handed out to the user in a matter of seconds.
The app can idetify essential cough signatures, quickly, inexpensively and reliably detect high-risk patients. It can detect if you are high risk, and one can undergo Covid testing and isolation. With a second wave anticipated, the app can help with a more directed and focused approach. It would help categorize high-risk contacts, and reduce the financial burden and address limited healthcare manpower,” said Dr Vishal Rao, associate dean for academics at HCG Hospital.
There are four key steps to the Covigilent app – cough detect, cough segment, Covid detect and explanation of inference. “This can also create a database that can be leveraged to continually improve model predictions,” said Dr Rao.
How to use the app
Enter your information, keep the phone 6cm away, either to your left or right side, and cough three times in a completely silent environment. Wearing a respirator or mask could give false results. The app can be downloaded at https://covidassist.wipro.com