How AI can help revolutionise NHS respiratory disease care

Dr Ameera Patel, CEO of TidalSense, explains how AI could completely change diagnostics in respiratory care.

Respiratory diseases affect one in five people. Already the third biggest cause of death in the UK, the number of people impacted by these conditions is rising. The latest NHS figures show that hospital admissions for respiratory illnesses are very close to pre-pandemic levels. Furthermore, analysis by Asthma and Lung UK also highlights a direct link between admissions and deprivation, due to factors like increased exposure to air pollution, dampness and mould.

Against a backdrop of NHS pressures – most notably, rising numbers of patients with long-term health conditions, and widespread staff shortages – diagnosis of respiratory diseases is not keeping pace with the growing prevalence of respiratory conditions.

If we take chronic obstructive pulmonary disease (COPD) as an example, around two thirds of people with the disease in the UK are undiagnosed, with one third only identified once they are admitted to hospital, when it is likely their disease is already significantly advanced and their symptoms severe. This goes a long way to explaining why the UK has the second highest death rate from lung disease in Europe, second only to Turkey.

For lung conditions, starting treatment early is critical. For asthma and COPD effective treatment reduces symptoms and exacerbations, reducing healthcare visits – including emergency hospital admissions. But the current clinical pathway for respiratory conditions is ineffective, inefficient and expensive – many patients are misdiagnosed and aren’t escalated into appropriate treatment quickly enough. COPD alone is the second most common reason for an emergency hospital admission and total admissions for COPD are estimated to cost the NHS £491 million annually.

This contributes significantly to the NHS’ financial burden – all lung conditions (including lung cancer) cost the health service around £11 billion annually. COPD and asthma, the two biggest chronic respiratory conditions which affect one in five people in England, cost the NHS around £5 billion each year.

Doing away with misdiagnosis

Early and accurate diagnosis is critical to easing the mounting pressure on our health service, eliminating unnecessary patient appointments while enabling earlier interventions for those who urgently need them.

But current diagnostic methods present a significant barrier to this goal. For example, the current test for COPD and asthma is spirometry, an early-Victorian technology that can be unpleasant for patients and requires specialist training to operate. Not only is this 180-year-old approach complex to perform, but it is also dependent on patient technique. What’s more, abnormal results can be challenging to interpret, meaning that misdiagnosis is rife.

Access to spirometry tests is patchy at best and diagnostic testing completely shut down during the pandemic. Conservative estimates predict there are around 27,000-34,000 people currently awaiting a diagnostic test.

Integrating new technologies – such as AI – is needed to get to grips with the backlog, and open the possibility of accurate, fast diagnoses.

It’s perhaps not surprising, therefore, that The NHS Long Term Plan prioritises accurate early diagnosis and access to testing for chronic respiratory diseases as a way to create efficiencies for the NHS, and improve the quality of treatment and care for patients.

More than the human eye can see

Thanks to its ability to analyse and understand large quantities of clinical information, AI has huge potential to pave the way to highly accurate diagnoses. AI-led technology is already being applied to the assessment of everything from stroke detection through to retinal screening, using trained algorithms and deep learning to quickly detect signs of disease that may not have been visible to clinicians.

There have already been successful demonstrations of identifying respiratory conditions using existing clinical data. For example, AI has been applied to aid the diagnosis of lung cancer and pulmonary fibrosis to help clinicians identify at-risk patients, speed up decision-making and reduce unnecessary procedures.

If applied to respiratory diagnostics, AI could mean that patients with chronic respiratory diseases would be spared the ordeal of spending weeks or months moving between clinicians to secure a diagnosis, instead giving them access to the right treatment, medication, and dosage at the right time. Better disease management could also deliver significant savings to the NHS.

Going beyond diagnostics

AI-led technologies are also opening powerful predictive and forecasting capabilities. For example, these technologies could be used to predict a patient’s future disease development, helping guide clinical decision making and opening access to early medical or lifestyle interventions. AI can even be used to predict the people within populations who are most at risk of developing chronic respiratory disease, ensuring they are prioritised for diagnosis or screening programmes.

At the same time, AI has considerable potential for improving the patient experience – empowering patients to self-monitor and manage their condition outside of the healthcare environment, resulting in a better quality of life for the patient, and further efficiencies for the health service.

In Greater Glasgow and Clyde, 500 COPD patients are being monitored at home to enable earlier interventions while also relieving pressure on the NHS. The scheme combines patient records with real-time data from fitness trackers and at-home breathing equipment, and users can directly message doctors with any health concerns via a smartphone app. A new trial later this year will also apply AI to this data, immediately flagging up patients who might be experiencing more severe symptoms. Early results are positive, suggesting that the scheme has already reduced hospital admissions by over half.

Saving time, saving lives

As the number of patients with chronic respiratory conditions continues to grow, it will be impossible for the NHS to meet its objectives to improve the quality of life and health outcomes of people with respiratory disease, unless the hurdle of diagnosis is overcome first.

Technology will be critical in bridging the gap between patient demand and clinical supply, with AI enabling faster, more accurate diagnoses and opening access to diagnosis outside of the traditional clinical setting. The increased capabilities of digital technologies are paving the way for more effective treatment plans, reducing the likelihood of frequent hospitalisations, and generally contributing to a better quality of life.