AI in healthcare: Smarter delivery of care

Mike Wright, innovation business partner, CW Innovation, Chelsea and Westminster Hospital NHS Foundation Trust, talks about the use of AI at the Trust, and highlights a need for a step-by-step approach.

Most of us are already obliviously interacting with AI on a daily basis; most commonly on our email and social media platforms, and with chatbots and mobile banking. AI is increasingly being used to support shortlisting of applicants for job interviews, for fraud detection in the payment of claims on insurance policies and determining eligibility for financial products such as loans. In short, AI is a factor in many aspects of our everyday lives.

In November 2018, then health secretary Matt Hancock said AI "will play a crucial role in the future of the NHS." I think it will, over time, but not in the imminent future because despite its already far-reaching usage in other parts of society, AI is still in its infancy in healthcare. At Chelsea and Westminster Hospital NHS Foundation Trust, a Global Digital Exemplar and emerging leader in innovation and discovery, we are – carefully – harnessing the benefits of AI. 

AI often conjures up ethical concerns about privacy and surveillance, bias, and discrimination, and perhaps the deepest, most difficult area is the role of human judgment. With this front and centre, our step-by-step approach is to identify areas that we feel AI will make the biggest impact for patient outcomes, and then work with the right partners to codevelop and rigorously test the technology against existing practices to ensure favourable outcomes without bias.

We are currently exploring the use of AI in both clinical and back-office functions. In a clinical setting, we are ‘testing and scaling’ the innovative AI algorithm DERM (Deep Ensemble for Recognition of Malignancy) from Skin Analytics. This can identify skin cancers from dermoscopic images taken using a smartphone and attachment and has been shown to be more accurate than dermatologists in identifying some skin cancers. We have completed the recruitment of 700 patients to an exciting prospective study to evaluate the impact of DERM on healthcare resource allocation for the diagnosis of skin cancers whilst also assessing patient acceptability. Pending these exciting results and a successful grant application we hope to implement DERM as a decision support tool into our standard care pathway.

In the back-office, we are investigating how AI can shoulder the burden of dealing with routine enquiries to HR and Procurement functions releasing staff to focus on more value adding tasks such as training and development of employees at the Trust. One of the benefits of this type of machine learning is that we can provide an ‘always on’ service, enabling both our staff and patients to access the information they need, regardless of shift patterns or the time of the day.

There’s no doubt that the COVID-19 pandemic has rapidly accelerated the use of digital solutions and AI. One example is the algorithms we’ve developed, combining our clinical expertise with Sensyne Health’s AI and machine learning know-how, to crunch patient data incredibly quickly and give clinicians insights and predications they can act on immediately.

The first algorithm called SYNE-COV recently achieved UKCA mark status. SYNE-COV predicts what is likely to happen to a COVID positive patient; specifically, their risk of needing treatment in ICU or invasive mechanical ventilation, based on vital signs, test results and other useful data. This helps clinicians intervene sooner and a patient that might have ended up in ICU can be treated effectively before they deteriorate further.

The second algorithm, SYNE-OPS-1, predicts resource requirements overall, helping maximise available resource by identifying problems, such as a shortage of beds or ventilators, before they became an issue.

At the start of this year, the UK Parliamentary Office of Science and Technology published its AI and Healthcare Research Briefing about the use of AI in the UK healthcare system, acknowledging that whilst AI is not yet widely used within the NHS, there are multiple AI products at various stages of trial and evaluation. We are excited to be working on some of these pioneering applications and play a leading role in establishing the UK as a champion of digital technology in the healthcare sector.

The opportunities are endless but what’s critical is the approach. Our data warehouse and our increasing use of data-mining tools and data science techniques are absolutely delivering smarter management and delivery of care. They are enabling clinicians and managers to get diagnoses right first time to identifying patients with deteriorating health earlier.

Through constantly evaluating and exploring new solutions in the use of AI with partners and staff, we will continue to steer our own path, being cautiously optimistic, whilst ensuring the best outcomes for patients both now and in the future.