Artificial intelligence (AI) is increasingly making great advances in health and care. But, explains Guy Boersma, managing director of the Kent Surrey Sussex AHSN Network, there are obstacles to be overcome.
Watching the TV adverts for the myriad of virtual assistants that use AI to help run your home, it might feel like the future is here. Interestingly Sally Davis, the Chief Medical Officer, used that same phrase as the title of her summary of her 2018 annual report. She wrote: “Big data and the computing power to make predictive analytics everyone’s business is already here. Artificial intelligence that can diagnose disease earlier and improve prognosis is already here. We need to embed and build upon these innovations to accelerate and normalise implementation of what works across England.”
To help that happen more quickly, the AHSN Network, the Department of Health and Social Care and NHS England published Accelerating artificial intelligence in health and care in autumn 2018. We summarily defined AI as “advanced technologies that enable machines to effectively carry out complex tasks that would require intelligence if carried out by a human”. The report is based on ‘a state of the nation’ survey with respondents including chief executives and senior managers across England. It details an increasing number of examples of how AI is being used in the NHS.
AI in the NHS
The Technology Improved Health Management (TIHM) test bed, based in Surrey Heartlands, is testing the ‘Internet of Things’ in the homes of 400 citizens to support people living with mild to moderate dementia. Data comes 24/7 from a variety of sensors, monitors and wearable devices into a server in the University of Surrey, where they collate the data, cleanse it and provide it out to a clinical monitoring team. It is able to detect early signs of things like urinary tract infections that often lead to falls. This triggers an intervention by a GP to help keep people safe and healthy in their own homes and avoid the need for an emergency hospital admission.
Then there is Oli the chatbot in Alder Hey Children’s Hospital. The lead paediatrician is unequivocal in stating that the chatbot has improved the care he provides, as the children and families ask different questions of the chatbot than the ones they are prepared to ask him directly. In all his years of clinical practice he'd never had a child ask what one child asked the chatbot: "Will I wake up from my operation?" This has changed how he provides care for his patients.
And as for diagnosis, a recent study in Annals of oncology asked dermatologists to make a diagnosis of malignant melanoma or benign mole just from dermoscopic images. The dermatologists detected an average of 86.6 per cent of melanomas. However, a form of AI detected 95 per cent of melanomas. The researchers saw this as useful as ‘decision-support’ to dermatologists rather than ‘decision-taking’ instead of dermatologists. The more I discuss AI advances, the more I am convinced that it can also help meet the workforce challenges within the NHS. I know it's hard to see how the 100,000 vacancies in the NHS are going to be filled tomorrow, let alone how we will cope in 2040. The demographic reality is that as time passes there won’t be enough people of working age to look after a growing and ageing population unless our models of care change.
At the moment around 1 in 12 radiologist posts are vacant and recruitment is difficult. As AI can help ease workloads through automated and reliable image recognition technology for certain X-rays, MRI and CT scans, it can help free-up time for decision-making and the more complex and difficult analysis that needs human intelligence.
Supporting AI in the NHS
In spite of these examples, when compared to the consumer market (with biometric passports and barcode scanners in supermarkets), the market for AI in healthcare is relatively immature. We give away data about our shopping habits by using our store cards, telling all on social media and allowing our phones to track our location, but people seem to be wary about allowing access to personal data for the NHS to use to help us be healthier.
Hopefully this will change when people see the benefits of allowing access to their data and have greater trust in the NHS. So we have set up a new AHSN Network AI initiative to develop an ecosystem of leaders and innovators committed to reducing obstacles to adoption of AI, so we can demonstrate how it can transform health and social care.
The first AHSN state of the nation survey set a baseline against which progress can be monitored and we have been recommissioned to produce another report for 2019.
To make sure our 2019 report captures a fuller story of how AI is transforming healthcare for patients, please tell us about your examples by emailing firstname.lastname@example.org
Guy Boersma is managing director of the Kent Surrey Sussex Academic Health Science Network and leads for the AHSN Network on AI. Follow them on Twitter @KSSAHSN and @AHSNNetwork.