Scientists unveil a new AI technology that can detect pneumonia based on the sound of a cough, potentially reducing healthcare costs and empowering self-diagnosis.

A new artificial intelligence-driven technology has been developed that can diagnose pneumonia by analysing the sound of a person’s cough. This breakthrough, if implemented on a large scale, could allow individuals to identify the illness without visiting a doctor, with the added benefit of potentially reducing NHS costs.

The AI system works by interpreting the acoustic characteristics of a cough, assessing whether the sound indicates the presence of the deadly respiratory condition. To accommodate variations in rooms and recording devices, the technology is equipped with room impulse responses, which measure how sound frequencies interact with different environments.

By combining these impulse responses with recorded cough sounds, the algorithm can function effectively in diverse settings. There are already plans to roll out this technology, enabling patients to be monitored in their homes. The development of an app for home-based social care is also underway.

Jin Yong Jeon, a researcher from Hanyang University in South Korea, explained: “The automatic diagnosis of health conditions through everyday cough sounds will facilitate remote treatment, reducing the need for in-person visits. It will also lower medical costs.”

Pneumonia is a significant health concern, affecting approximately eight in every 100,000 people in the UK each year, with particularly severe consequences for the elderly and young children. Early detection is crucial for the best chance of recovery.

Current diagnostic methods involve blood tests and chest scans, with doctors needing to suspect pneumonia before treatment can begin. These findings were shared at the 183rd meeting of the Acoustical Society of America in Nashville, Tennessee.