AI App Swaasa Uses 15-Second Cough to Screen TB & Asthma

The AI app has enabled users to record a short cough sample on a smartphone and receive an assessment of potential respiratory risk within minutes.

Swaasa, an Indian AI app, is turning a simple 15-second cough into a rapid screening test for tuberculosis, asthma, COPD and other lung diseases, offering a non-invasive, smartphone-based tool designed to expand early detection across underserved communities.

The AI app has enabled users to record a short cough sample on a smartphone and receive an assessment of potential respiratory risk within minutes, making early screening faster and more accessible.

Developed by Salcit Technologies, a Telangana-based respiratory healthcare company, Swaasa was conceptualised in 2017 to address gaps in respiratory diagnostics, particularly in rural and peripheral health settings where spirometry is often unavailable.

The AI app was trained on a large database of cough sounds collected from both healthy individuals and patients with lung disease. By analysing acoustic signatures in each cough recording and combining them with symptom history and demographic data, Swaasa classifies cases as normal or abnormal and further categorises them into obstructive or restrictive patterns.

“The recorded cough is analysed using an AI algorithm, alongside symptom-based history and basic demographic details. Based on this combined information, the system identifies whether the person is likely to have obstructive or restrictive lung disease, including conditions such as COPD and asthma,” Dr Harshal Ramesh Salve, additional professor at the AIIMS Centre for Community Medicine.

According to its developers, the AI app has demonstrated an overall accuracy of about 90% in identifying respiratory abnormalities. It differentiates between obstructive and restrictive conditions with roughly 85% accuracy, detects asthma or COPD at around 83%, and identifies pulmonary tuberculosis at approximately 79%. Chronic cough patterns carry distinct acoustic markers, which the AI system is trained to recognise and interpret.

“Depending on where these receptors are activated - upper airways in asthma, deeper in the lungs in COPD - the cough exhibits distinct acoustic characteristics. Variations in glottis and respiratory structures across diseases are reflected in sound waves,” Narayana Rao Sripada, founder of Salcit Technologies, said.

The AI app has undergone validation in primary and secondary healthcare settings, including trials at Apollo Hospitals, Christian Medical College Vellore, AIIMS institutions, and JIPMER Puducherry.

It is currently being piloted in states such as Uttarakhand for COPD screening and has been used in community health programmes, including TB detection initiatives.

Stay tuned for more such updates on Digital Health News