Our overarching aim is to support people with undetected respiratory disease to seek appropriate healthcare earlier.
The Artificial Intelligence assisted earlier Respiratory Evaluation (AIRE) prototype will enable people concerned about their breathing to access an ‘AI respiratory specialist’ in their own environment. End-users will call a UK telephone number for an AI-powered respiratory specialist assessment. AIRE will replicate conversational clinical history-taking for respiratory symptoms, and analyse audio and speech cues to deliver a clear outcome of lung health via a traffic light indicator to encourage people to seek healthcare where indicated.
Stakeholder involvement is at the heart of the project; expert researchers will work with end-users to co-create a prototype that is ready for real-world evaluation. By leveraging interdisciplinary approaches combining advanced engineering and physical sciences (EPS), conversational AI, and respiratory medicine, AIRE will deliver an intelligent, inclusive and accessible low-cost solution that uses the familiarity of the telephone to accelerate the early detection of respiratory conditions.
Multiple interacting factors negatively affect health-seeking behaviour for respiratory symptoms such as breathlessness including stigma, health literacy, digital exclusion and challenges accessing primary. Reluctance to present to a healthcare professional, need for information, lack of acknowledgement and consultation constraints all contribute to delays, but an identified key solution, and role for AIRE, is “selecting a potential helper” (Hutchinson A 2018). Significant delays to diagnosis remain even after healthcare help is sought with over 1 in 4 people with chronic respiratory disease (CRD) waiting over five years (ALUK 2025, Karsanji 2024). This diagnostic delay is compounded by health inequalities with CRDs having the largest gap in mortality rate between rich and poor than any other condition (ALUK 2024, NHS England 2024).
'A careful history will lead to the diagnosis 80% of the time' is a widely used adage (now evidence-based) in medical teaching and practice (Hampton 1975). A respiratory specialist clinic appointment will start with taking a detailed clinical history which is a more complex interaction than simply having the required knowledge and asking a set of questions. Such diagnostic reasoning can be modelled as following an algorithm of sequential symptoms with individual probabilities of differential diagnoses which increase in certainty as the consultation progresses. Over many years clinicians develop the skills of extracting pertinent clinical details from patients who have different personalities, educational backgrounds and language skills, and combine them to make a rapid and accurate diagnosis. It involves how the questions are asked, how they are answered, and being sensitive to many conversational markers and exchanges that help a clinician establish the patients’ condition. For respiratory illnesses this may also extend to subtle clues in the audio, including how they breathe and when they pause in conversation. It is not just about what a patient says, but how they say it.
The AIRE prototype will use state-of-the-art language engineering techniques and Artificial Intelligence (AI) to replicate the respiratory specialist clinical history and diagnostic reasoning process. In addition, patients’ audio speech patterns and conversational markers will be analysed to further support an earlier diagnosis of respiratory disease. AIRE will use cutting-edge artificial conversational interaction to enable a fluid human-like conversation, and audio and content analysis using Large Language Models (LLMs) to generate the traffic light indicator of ‘lung health’ which gives agency to seek further medical help thus reducing delays to diagnosis. Empathy and motivational interviewing techniques will be embedded to promote health activation and empower the caller to take appropriate action after AIRE: for example, accessing a recommended diagnostic intervention potentially in future linked to NHS community diagnostic centres (CDCs).
In practice, AIRE has the potential to be cheap, scalable, equitable and accessible at home or a person's usual environment. The AIRE prototype will be based on a telephone call as this is more inclusive for those with low digital literacy. Once developed, AIRE will be easy to maintain, as updates can be made incrementally without retraining models from scratch. It will also be highly replicable, scalable and adaptable across different environments due to its modular, open-source foundations.
-ALUK 2024. Asthma and Lung UK mission report. Mission-lung-health-report-2024. Accessed 05/07/2025
-ALUK 2025. Asthma and Lung UK survey. Transforming Respiratory Diagnostics: The Way Forward. Asthma + Lung UK. Accessed 05/07/2025
-Hampton 1975. Relative contributions of history-taking, physical examination, and laboratory investigation to diagnosis and management of medical outpatients. BMJ 1975;2:486–9.
-Hutchinson 2018. Living with breathlessness: a systematic literature review and qualitative synthesis. Eur Respir J. 2018 Feb 21;51(2):1701477.
-Karsanji 2024. Time to diagnosis and long-term outcomes for adults presenting with breathlessness. medRxiv 2024.02.19.24302618
-NHS England 2024. NHS England » Core20PLUS5 (adults) – an approach to reducing healthcare inequalities. Accessed 05/07/2025.
Funding for AIRE is provided by UK Research and Innovation under grant number: UKRI3605