AI-driven app in India to listen for sounds of TB in coughs

The sound of a patient’s cough could soon be enough for doctors to diagnose cases of tuberculosis, a senior medical official in India’s Jammu and Kashmir region said.

The use of artificial intelligence has given a major boost to efforts to eradicate the disease, which the World Health Organisation said killed about 1.3 million people last year.

The group lists TB as the second most deadly infectious disease after Covid-19.

India aims to reduce the fatality rate using a mobile app designed to screen patients for the disease by comparing their coughs to a data set of about 15,000 TB-positive and TB-negative cough sounds.

Medics in the Jammu and

Kashmir region are being trained to use the platform as part of a programme led by Dr Adfar Yaseen, head of Kashmir’s TB treatment division.

She has been tackling cases of the disease for the past three decades, an effort that has often been frustrating.

It previously could take up to a month for test result to arrive in the territory, India’s northernmost region, where the climate and terrain are harsh.

“As a screening tool, it is going to be helpful because we have hard-to-reach places and at many remote places we do not have X-rays. So it is going to help us and refine our screening,” Dr Yaseen told The National.

The app is “very efficient” and will be able to give medics as much information as an X-ray, she said. The platform works by converting the sound of a patient’s cough into spectrograms used in a network that learns to predict the likelihood of someone having the disease, the Wadhwani Institute for Artificial Intelligence in Mumbai said.

The institute developed the technology and is an AI partner of the Indian government’s Central TB Division.

“For every 2.8 TB cases prevalent in the community, one case gets notified and 1.8 cases get missed,” said Wadhwani, which is also supported by the US Agency for International Development.

The app is only one part of India’s increasing efforts to improve surveillance and testing for TB.

The fatality rate for untreated TB is about 50 per cent and treatment requires patients to take medication regularly and for a long period of time.

India aims to eliminate TB by 2025, a formidable task considering 23 per cent of the 10.6 million global cases last year originated in the country.

Rouf Ahmad Tramboo, supervisor of the TB Centre in the Budgam district of Kashmir, said the app was an effective early detection system.

“This app is excellent for a place like Kashmir, where some places are so remote that there is not even electricity available,” he said.

“The app generates the result whether the person is presumptive positive or not.

“There is also an option that lets the healthcare worker decide, on his own analysis and diagnosis, whether a further test is needed or not.”

In Kashmir, where temperatures often plummet below freezing during the winter, respiratory illnesses and coughs are common, making it difficult for people to identify symptoms of TB.

The region has had success in its fight against TB, with several districts declared free from the disease.

“We have other respiratory ailments here, like chronic obstructive pulmonary disease or asthma, but we do not have tuberculosis to the extent as it is in other parts of the country,” Dr Yaseen said.

Effective screening is necessary to reduce the number of cases in the region further, because there is always a threat of a resurgence of infection, she said.

“We just cannot sit back. We have to be vigilant,” Dr Yaseen said.