An AI-powered app that can ‘hear’ disease

The machine learning platform screens coughing sounds for signs of illness

Spotted: Spirometry, a simple test that measures how much air you can expel, is an important tool in spotting conditions like cystic fibrosis and asthma. But this kind of test is only readily available to those with easy access to hospitals and doctors’ surgeries, or those who can afford to buy one for home use. Now, Indian company Salcit Technologies is revolutionising this kind of testing with the help of AI.

The startup has created a patented machine learning platform, called Swasaa, that can ‘hear’ illness in the sound of your cough. The innovative bioacoustic technology uses audiometric analysis, taking into account any other symptoms flagged by the user, to accurately measure the health of someone’s lungs and spot potential diseases – just by using a smartphone.

Wearing a mask, users hold the phone horizontally just a few centimetres away from their face and cough for 10 seconds. Detailed insights into lung health are then available quickly and users can use this information to take further action, including consulting with their doctor for treatment. All of this can be done with the easy-to-use Swasaa app.

Being able to conduct these tests at home could transform the way those with ongoing conditions, like chronic obstructive pulmonary disease (COPD), manage their health. Instead of needing to regularly visit their doctor, sufferers can instead test their lung health where they are, and this information is shared with the relevant medical teams. This means that even if it is too difficult or expensive to see a medical professional in person, a user’s health is still being carefully monitored.

Earlier this year, Google researchers developed their own bioacoustic foundation model called HeAR (Health Acoustic Representations), which was trained on 300 million pieces of audio to recognise health patterns in different breaths and coughs. Salcit is now exploring the potential of HeAR to expand the capabilities of its own technology, specifically to enable the early detection of tuberculosis (TB) from coughing sounds.

Written By: Matilda Cox