An estimated two million people across the United Kingdom have chronic obstructive pulmonary disease (COPD) without knowing it. Without a diagnosis, patients receive no treatment and face avoidable deterioration. The cost falls on emergency services as well, because COPD is the second leading cause of unscheduled hospital admissions.

A new initiative, the COSMIC (COPD Spirometry Management Supported by AI in a Community Setting) programme, aims to narrow that gap. It will supply National Health Service (NHS) primary and community care providers with diagnostic decision-support software, targeting the testing backlogs that have persisted since the COVID-19 pandemic.

The burden of the disease is not evenly spread. COPD prevalence, adverse outcomes and mortality are all higher in socioeconomically disadvantaged communities. Late diagnosis in those areas means more patients reach hospital in an acute state, when earlier assessment in a GP surgery or community clinic might have allowed treatment to begin years sooner.

Spirometry is the principal method of assessing lung function, and it is central to confirming COPD. Routine testing was significantly reduced during the pandemic, and services have yet to fully recover. Many patients who would ordinarily have undergone testing during that period remain on waiting lists.

Staffing compounds the problem. Primary care units report a shortage of accredited clinicians able to carry out spirometry to the required standard. Clinicians who are not specialists often have less confidence in performing the test correctly and in interpreting the results, which can lead to repeat appointments or missed diagnoses.

The COSMIC programme addresses this through collaborative projects and software donations that place an AI-driven spirometry platform, ArtiQ.Spiro, into everyday primary and community care workflows. The software assesses the quality of each test as it is carried out. If a reading is poor, it prompts the clinician to repeat the test while the patient is still in the room, which removes the need for a second visit.

The platform also interprets results within seconds. It produces guideline-aligned readings together with probability metrics, giving the clinician a structured basis for deciding whether a patient has COPD. The aim is to support judgement rather than replace it, particularly for staff who perform spirometry only occasionally.

The software operates within an established regulatory framework. The National Institute for Health and Care Excellence (NICE) has issued guidance supporting its use, which allows NHS deployment under a three-year evidence-generation period. During that time, data on performance and patient outcomes will be gathered to inform any longer-term adoption decisions.

Organisers describe several objectives. The first is to widen the pool of clinicians who can carry out reliable respiratory assessments, so that testing no longer depends on a small number of accredited specialists. Success would be measured in part by how many patients are tested in community settings who would otherwise wait for a hospital referral.

A further aim is to change where COPD care begins. At present, a large proportion of patients first come to attention during an emergency admission. Programme leads want to move that point of contact into primary care, where early treatment, smoking cessation support and monitoring can slow the progression of the disease and improve long-term outcomes.

Deployment is also intended to follow need. The programme directs diagnostic tools towards underserved regions, where undiagnosed disease is most likely to be concentrated. If testing capacity rises in those areas first, the initiative could reduce some of the wider inequalities in respiratory health that have been recorded for years.

The evidence gathered over the next three years will determine whether the approach is extended across the NHS. The scale of undiagnosed disease means that even modest gains in testing capacity could identify large numbers of patients who are currently untreated.

‍