8 August 2025
A newly published study led by researchers at Royal Brompton and Harefield hospitals has found that artificial intelligence (AI) software can significantly improve diagnostic accuracy for lung conditions when used by clinicians in primary care settings.
Led by Professor Will Man, respiratory consultant, and funded by the National Institute for Health and Care Research (NIHR), the SIPRO-AID study is the first randomised controlled trial to test the impact of AI in the diagnosis of lung disease in GP practice and community settings.
Why diagnosing lung conditions can be difficult
Spirometry is a common breathing test that is used to help diagnose long-term lung problems like chronic obstructive pulmonary disease (COPD) and asthma.
However, the way results are interpreted can vary widely in primary care, leading to missed or delayed diagnoses.
In fact, Asthma Lung UK estimates that one in four people with COPD wait five years or more for a diagnosis, and one in eight people wait more than a decade.
Testing AI to help diagnose lung conditions
The study aimed to evaluate whether the AI tool, ArtiQ.Spiro, could support clinicians in making more accurate assessments by analysing spirometry data and generating reports that assess test quality and suggest the most likely diagnosis.
The research team enlisted over 100 GPs and nurses in primary care to take part in the study. Each reviewed 50 real patient spirometry tests. Half of the clinicians used AI software to assist with diagnosis (intervention group), while the other half worked without it (control group).
Key results from the research include:
- Clinicians using the AI software were 9% more accurate overall in diagnosing lung conditions.
- Accuracy improved by 16% for COPD diagnoses with AI support.
- AI software alone was more accurate than the combination of the clinician and AI or clinician alone.
Why this study matters for patients and GPs
Professor Man highlighted that the findings show how AI can play a valuable role in supporting clinical decision-making and help identify lung disease earlier so that patients can get the right treatment sooner. He said:
“This study is an important step in showing how digital tools like AI can directly support busy GPs and practice nurses. By improving the accuracy of diagnoses, we can reduce delays for diagnosis, and get the right treatments to the right people. This could potentially reduce the burden on hospitals.”
The study aligns with the NHS 10-Year Health Plan for England, supporting key goals like moving diagnostic services into the community, empowering clinicians with digital tools, promoting early diagnosis, and preventing illness before it worsens.
You can read the full paper here.
To find out more about our research, please contact us.
Read more research stories or sign up to the research newsletter.