IISc team to collect cough recordings from hospitals

A team of Indian Institute of Science (IISc) researchers that has been working on a tool for diagnosis of Covid-19 based on respiratory, cough and speech sounds are now engaging with hospitals in Karnataka to collect recordings from patients, including Covid-19 positive subjects.

After receiving a nod from the Indian Council of Medical Research (ICMR) to collect data and develop the tool the research team has collected recording from MS Hinduja Hospital in Mumbai, IISc will soon be collecting data hospitals in the state.

“We approached MS Hinduja Hospital in Mumbai and their team was enthusiastic about the study and with the help of their team, we got our proposal approved through the ethical board and now have started collecting and recording from their hospital facility. In a similar vein, we are now working with hospitals in Karnataka also. The hospitals in Bengaluru that have expressed interest are the MS Ramaiah Hospital and the St. Johns facility.

Similarly, KMC Mangaluru has also indicated their willingness to participate in the study,” Sriram Ganapathy, the head of the IISc research team, told Bangalore Mirror.

Sriram and this team of researchers are working on developing Coswara, the diagnostic tool which aims to create a data set composed of voice samples from a mix of healthy individuals as well as from those affected with the virus using a webpage (https://coswara.iisc.ac.in/) and a mobile app.

Sriram said the proposals placed before hospitals in Karnataka are going through their ethical board committees and the approvals are awaited.


We have received over 1,200 participants as of early August. All this data has been listened to and manually curated by our team

–Sriram Ganapathy, head of research

“In the hospitals, we aim to recruit both Covid-19 positive patients as well as subjects with pre-existing respiratory ailments. Many doctors from these hospitals are now actively part of the team as well,” Sriram said.

The team has been collecting data from April but, until the end of May, they did not have any Covid-19 positive subjects in the study.

“In the last month or so, working with hospitals, we are starting to see more and more data coming into the pool. We have received over 1,200 participants as of early August. All this data has been listened to and manually curated by our team. We are also happy to find that most of the data is indeed noise free. Using the data, the machine learning experiments and analysis are underway,” he added.

The team is looking to work with more hospitals and health centers.

“As with AI based systems, the more the data, the better the chance of their success. Further, we would like to collect as many variants of the data as possible so that the algorithms can distinguish between say a Covid-19 induced cough versus general asthma/pneumonia induced cough,” he said.

Source link


Sagar Biswas

Leave a Reply