NHS Glasgow turns to AI and wearables to improve treatment of lung disease

Madeline Bennett Profile picture for user Madeline Bennett June 26, 2019
The NHS board is using the Microsoft Azure platform to send patient data from their home directly to medical staff to cut down on emergency hospital visits

Image of a patient in a hospital receiving new AI lung treatment

Chronic Obstructive Pulmonary Disease (COPD) is a serious lung condition affecting 1.2 million people in the UK. As the second most common cause of emergency hospital admissions, the NHS is understandably keen to explore ways to reduce this number, to offer better patient care and save money.

NHS Greater Glasgow and Clyde (NHSGGC) is pioneering the use of artificial intelligence (AI) in the treatment of COPD to reduce the amount of cases where sufferers end up in hospital.

NHSGGC is one of the largest healthcare boards in the UK. Its 38,000 staff serve more than one million people in around 30 hospitals housing 6,000 beds. Overall, the board admits and discharges 400,000 patients per year, including 10 COPD patients every day in the emergency department, with an average cost of £6,000 per admission.

Last year, the board approved a five-year digital strategy, including a two-year AI trial for COPD patients. The pilot involves a wearable mask that patients put on at home, which collects data and sends this back to clinicians.

NHSGGC is using Microsoft’s Azure cloud technology as the foundation for this project, along with machine-learning software from KenSci and support from digital transformation consultancy Storm ID. Once the pilot is up and running, medical staff will be able to see patient-reported information and physiology data produced by the wearable devices and breathing machines, which blow air through a mask at a varying intensity to increase the capacity of the person’s breathing.

Patients and clinicians can communicate through the cloud platform, and consultants can change the ventilation remotely via an online portal. Algorithms are able to track and anticipate when a flare up is likely and alert clinicians, so they can take action and avoid the need for emergency hospital admissions. Denise Brown, head of Strategy and Programmes at NHSGGC, explained:

We want to treat patients as near to their home, like at a community clinic, or in their home if that's possible. Bringing them in, having them travel to take readings is not ideal. This sort of remote monitoring and intervention allows the patient to remain at home and to be monitored on an ongoing basis before they reach crisis point.

NHSGGC has been using technology since the start of the decade to let patients collect basic information like blood pressure readings and send this data to GP systems. The AI system will gather data on a much smarter basis.

If the data is generated by the wearable and it's moved through our environment securely, we can decide what data is relevant and valuable to the clinician and the patient, rather than the clinician being bombarded. It allows us to configure the data so it's very targeted and clinically relevant.

In the last 18 months to two years, things really have moved on in terms of the technology that is available to support these sorts of initiatives, whereas previously with the analogue kind of processes, not a lot changed for an awfully long time.

Maximising time

The ability to gather data at home rather than having to attend appointments at hospital means less travelling for the patient, and getting the treatment they need sooner. It also reduces the number of cancelled hospital appointments and frees up clinician time. Brown said:

The more we can maximise clinician time and provide them with the data that they need rather than having to go look for it or run tests, or spend time traveling to patients' homes unnecessarily, that's more efficient and convenient.

If we can treat patients in their home or closer to home, if it’s clinically appropriate, that's our aim. To identify when a patient needs care in an acute setting sooner rather than having them reach a crisis point where they have to come in themselves, and they've become quite unwell.

The technology will also improve self-management for COPD sufferers, helping them with their breathing, escalating existing treatment or reaching out to the community respiratory team without the need for hospital-based clinical intervention.

NHSGGC is already a year into the pilot programme, which involved the design and app development work. The 12-month implementation phase will kick off in July, with 400 COPD sufferers testing the masks and AI system. Brown said a key factor in the pilot getting approval and set up quickly was having really supportive clinicians. Indeed, it was the medical rather than digital team that initiated the project.

We’ve got really proactive clinicians, who are leaders. Ten years ago, you would probably find e-health trying to drive agendas around technology and looking for clinicians to support it. We've got clinical e-health leads as well, who work directly with us, they have dedicated time to work with us. My experience of that is that they are very much leading on this and technology is enabling it.

They will be front and central in terms of leading these sorts of initiatives. And it's enabled by the technology and not necessarily driven by the technology. It goes back to positive patient outcomes and the patient being the priority, it’s not the technology.

Looking to the future 

All the evaluation work will be done during the 12-month phase, which will involve observing the patients, analysing the data and looking at how to potentially scale up the project afterwards. What comes after the pilot will be subject to the evaluations and an in-depth study based on the evidence – but Brown sees good things ahead for AI in healthcare, as long as projects are chosen wisely.

We're looking at a number of AI initiatives. AI for me has two aspects to it: there's the patient care aspect, how do we improve outcomes for patients. This is a really good example of that on the wearable side, patient-generated data, how do we use that and how can it support patient care. And then there's the other aspect around AI, in terms of efficiency, automating processes, using machine learning, there are significant opportunities. Some Trusts will be really in the early stages and others would be just watching, waiting to see what others are doing.

You could consume yourself with pilots in this space. You have to be quite focused in terms of if you've got a strategy, how can the outcomes in that strategy be supported through AI and focusing in-depth rather than spreading ourselves very thinly.

A grey colored placeholder image