Diagnostic Accuracy of Integrated Sensing and Communication in Lung Disorder Classification

Article Impact Level: HIGH
Data Quality: STRONG
Summary of  Communications Medicine https://doi.org/10.1038/s43856-025-01181-2 
Dr. Hasan Mujtaba Buttar  et al.

Points

Researchers developed a wireless screening system that uses artificial intelligence and microwave signals to recognize breathing patterns associated with five common lung diseases including asthma and tuberculosis in clinical settings.
The diagnostic platform operates at a frequency of five point twenty-three gigahertz which allows it to perform integrated sensing and communication tasks using only twelve point five percent of bandwidth.
Clinical testing involving one hundred ninety patients and thirty healthy individuals demonstrated that the deep learning model correctly identified respiratory conditions with an impressive ninety-eight percent overall accuracy during trials.
This contactless technology successfully distinguished healthy participants from those with respiratory illnesses with one hundred percent specificity which eliminates the need for invasive tests like spirometry or expensive imaging scans.
The proposed wireless method offers a scalable and low cost solution for real time mass respiratory health monitoring particularly in resource constrained environments or during future outbreaks of infectious diseases.

Summary

This research presents a non-contact respiratory screening system that utilizes Integrated Sensing and Communication (ISAC) at 5.23 GHz to identify five common lung diseases. By exposing patients to low-power microwave signals and analyzing the reflected data via artificial intelligence, the platform distinguishes the characteristic breathing patterns of asthma, chronic obstructive pulmonary disease (COPD), interstitial lung disease, pneumonia, and tuberculosis. The framework operates at the lower end of the bands intended for future 6G and WiFi7 networks, demonstrating the feasibility of performing sophisticated medical sensing using only 12.5% of the total available bandwidth.

The system was validated through a study involving 190 patients diagnosed with respiratory conditions and 30 healthy control subjects in Lahore, Pakistan. Seven and a half hours of microwave-frequency reflection data were recorded and processed using five machine learning and three deep learning architectures. This methodology prioritized the classification of disease-specific respiratory signatures without the need for physical contact, ionizing radiation, or invasive procedures like spirometry. The results indicate that RF-based sensing can provide high-fidelity diagnostic data suitable for both clinical integration and remote health monitoring in smart home environments.

A vanilla Convolutional Neural Network (CNN) model delivered the highest diagnostic performance, achieving an overall classification accuracy of 98%. Notably, the system identified healthy individuals with 100% specificity. These findings suggest that 6G-enabled ISAC platforms can facilitate rapid, low-cost mass screening for respiratory health, particularly in resource-constrained settings or during infectious disease outbreaks. By enabling continuous, contactless monitoring, this technology provides a scalable solution for early medical intervention and reduced healthcare costs

Link to the article: https://www.nature.com/articles/s43856-025-01181-2

References

Buttar, H. M., Rahman, M. M. U., Nawaz, M. W., Mian, A. N., Zahid, A., & Abbasi, Q. H. (2026). Non-contact lung disease classification via orthogonal frequency division multiplexing-based passive 6G integrated sensing and communication. Communications Medicine, 6(1), 9. https://doi.org/10.1038/s43856-025-01181-2