AI-powered wireless system screens five lung diseases

This is the claim of an international team of engineers and computing scientists based in Scotland and Pakistan who said their findings could lead to new forms of personalised health monitoring in clinical settings and in the ‘smart homes’ of the future. Their work is detailed in Communications Medicine.

The system uses radio signals of multiple frequencies paired with artificial intelligence to recognise the characteristic breathing patterns of five common lung diseases. 

It works by exposing patients to microwave signals emitted by a pair of software-defined radios at 5.23Ghz, which is at the lower end of the bands expected to be used for future 6G and WiFi7 networks. AI-enabled analysis of the signals reflected from the patients’ chests allows the system to identify the breathing patterns that are caused by different lung disorders.

In lab tests using real-world radio data, the system was able to screen for asthma, chronic obstructive pulmonary disease (COPD), interstitial lung disease (ILD), pneumonia and tuberculosis with 98 per cent accuracy.

The team tested their system by gathering microwave reflection data on 190 patients diagnosed with various respiratory diseases at a hospital in Lahore, Pakistan. They chose the data collection window to be during the high-smog season between October 2023 to January 2024, and again in January 2025. They also gathered data from 30 healthy individuals to act as a control group.

The team recorded nearly seven and a half hours of microwave-frequency data from the study’s participants. They analysed the data using five different machine learning models and three deep learning models to determine which performed best at correctly classifying the patients’ conditions. A deep learning model called vanilla CNN delivered the best performance, correctly spotting the signs of specific illnesses 98 per cent of the time, and identifying the healthy individuals with 100 per cent accuracy. 

In a statement, research lead and corresponding author Professor Qammer H. Abbasi, director of Glasgow University’s Centre for Integrated Sensing and Communication Enabling Cognitive Cities, said: “The ultrafast 6G wireless communications networks of the future have the potential to do integrated sensing and communication [ISAC], which will unlock a wide range of benefits for people around the world, with healthcare being one of the key applications. 

“This research showcases the effectiveness of ISAC, which allows a single communications infrastructure to both transmit data and perform sensing tasks at the same time. The sophisticated sensing which underpins our results only took up 12.5 per cent of the system’s available bandwidth. That means that the rest of the system’s bandwidth could be used for data transmission to help enable future generations of integrated, continuous health monitoring devices.”

MORE FROM MEDICAL & HEALTHCARE

Corresponding author, Professor Muhammad Mahboob Ur Rahman, of the Information Technology University in Pakistan, said: “By combining AI with radio sensing in a 6G/WiFi framework, we’ve been able to accurately spot the signs of lung disease without the need for physical contact, stethoscopes, or imaging scans. That could help enable safe, continuous, and contactless screening and early detection of anomalies, which has the potential to reduce healthcare costs due to early medical intervention.

“This work is also anticipated to have a big impact in low-resource settings and during future outbreaks of infectious diseases like COVID-19, where reducing contact with patients could help limit the spread of infections.”

Researchers from the University of Lahore and Heriot-Watt University also contributed to the research and co-authored the paper. The research was supported by funding from the Engineering and Physical Sciences Research Council (EPSRC).