Can cough sound help COVID-19 diagnosis? Yes, says IISc, Bengaluru scientists

Similar to the dataset creation stage, the application requests for recording the voice samples, and preferably provides a score indicating the probability of Covid infection.

NEW DELHI: Can respiratory, cough and speech sounds diagnose COVID-19 carriers? A group of scientists from the IISc, Bengaluru experimenting on sound-based diagnostics for coronavirus, are confident that it can help in faster and economical detection as limited sample study show that cough of a COVID patient is different from those suffering from other respiratory ailments such as bronchitis or upper respiratory infection.

For large data collection to test the hypothesis, they got a nod from the Indian Council for Medical Research (ICMR).

The project, “Coswara”, attempts to provide a simple tool for COVID diagnostics based on respiratory, cough and speech sounds.

The team aims to release the diagnosis tool as a web/mobile application. Similar to the dataset creation stage, the application requests for recording the voice samples, and preferably provides a score indicating the probability of COVID infection.

Sriram Ganapathy, Assistant Professor, Electrical Engineering, IISc, along with a seven-member team is working on the project since March-end, when COVID-19 cases in the country picked up.  The team started collecting data (voice, cough sample on smartphone) of people affected with Covid-19 and other respiratory infections.

“Cough has lot of variants and we found some interesting differences in the cough of the healthy subject to those suffering from respiratory ailments. But the sample was very small to make any conclusion and we decided to reach out to hospitals for data sharing. But due to internal regulations, there were issues related to data sharing and that is when we approached the ICMR for data collection approval,” said Ganapathy.  

During the research for the project, the team found an interesting US-based research paper showing that cough of COVID-19 patient is very different from other respiratory ailments like bronchitis or upper respiratory infection.

"It showed 95 per cent accuracy using voice and cough based data for Covid-19 detection. I shared the paper published early May with the ICMR during the virtual presentation and they were like this can help public based health system and will be a significant boost,” added Ganapathy.