The media is awash with reports of the dangers of AI and the Internet of Things (IoT), some of it warranted, and some of it fear-mongering to sell papers, clicks, or agendas. However, these concerns have largely arisen from the growing influence of automation and AI across all aspects of our personal and working lives – fears that remind tabloid commentators of popular dystopian fantasies.
But the naysayers perform a useful function: they mitigate against the negative effects of digital transformation and prompt governments into reform. Regulation both actively supports growth around technology, and protects the interests of citizens. GDPR exists for a reason.
However, one area where advances in AI stand to benefit all is healthcare – not least in the UK’s National Health Service (NHS), where regardless of a person’s background or wealth, they can expect the same treatment as the next person.
So can AI be the great enabler of connected healthcare?
While much popular media coverage focuses on the health service’s funding problems, there is much to celebrate when it comes to AI and the IoT in healthcare.
A recent report from independent think tank Reform, ‘Thinking on its own: AI in the NHS’, has taken a deep dive into current uses of AI, algorithms, and data in the health service, and as such is a useful jumping-off point for this analysis.
• Editor’s note: Last year, Reform courted controversy by suggesting in an automation and AI report that the public sector could save millions of pounds by stripping 250,000 workers out of public service and automating their roles, leaving workers – doctors, nurses, and teachers among them – to compete via reverse auction (to work for less money) in the gig economy.
For suggesting that automation was primarily a route to forcing down public service workers’ wages, Reform was censured by many commentators – rightly, in our view. We believe that new technology should be a strategic enabler in the provision of better services, and not a blunt instrument for operational cost-reduction, regardless of social consequences. – IoB.
This new Reform report explores the obstacles to AI’s further adoption, and what the future might hold, so Internet of Business has seized the opportunity it presents to take a wide-angle look at the state of play in AI- and IoT-enabled healthcare, and to share some takeaways from these and other findings.
Noman Lamb MP (LibDem), chair of the Science and Technology Committee, summarises the situation in his forward to the research:
We are on the brink of a major transformation in the way we diagnose, treat, and even prevent ill health. Whether it is wearable devices, AI surgical robots, or AI algorithms that can detect certain conditions with unprecedented speed and accuracy, these advances have the potential to propel the health and social care system into the 21st century – improving care both in the hospital and at home, and making much more efficient use of resources.
How AI could transform the NHS
AI can augment healthcare professionals’ skills across the whole spectrum of services. In the first instance, AI can help predict which individuals or groups are at risk of illness, so that they can be targeted with early treatment or preventative measures. The adage ‘prevention is better than cure’ is key to keeping hospital beds free and expenditure down.
Care in the community plays an essential role here, too, and this can be supported by connected technologies to keep healthcare professionals informed about their patients. Round-the-clock data, provided by non-invasive wearables, are already opening the door to valuable new information streams.
There’s even scope for consumer devices to be adopted by healthcare providers to better monitor patients and gain access to vital health data from outside the hospital environment.
Clearly, there is huge potential for the NHS to employ AI to analyse data from the likes of Apple Watch, WearOS, and Fitbit devices, particularly when one in seven people in the UK owns a health tracker.
Similarly, AI tools can help reduce the ‘care and quality’ and the ‘efficiency and funding’ gaps outlined in the NHS’s Five Year Forward View, with cutting-edge diagnostic treatment and automation – helping to spot issues earlier and speed up recovery times.
With the support of Moorfields Eye Hospital in London, Google’s DeepMind has already demonstrated AI’s ability to diagnose eye diseases. Elsewhere, examples abound of AI diagnostics for cancer detection and heart disease, potentially saving the NHS millions of pounds.
It must be acknowledged, of course, that the Royal Free NHS Foundation Trust was was censured in 2017 by the UK’s Information Commission for failing to protect patients’ data when it shared records with DeepMind.
However, many of these examples are from outside the UK. By comparison, adoption of AI and related technologies within the NHS has been sparse to date.
This may be because 2.5 million scientific articles are published in English language journals each year, so it’s an impossible task for health professionals to keep up with all the latest research, without a central programme to guide them.
Nuance Communications issued a Freedom of Information request to 45 Trusts asking about their use of AI; 30 responded. Of these, 43 percent said they were investing in what they considered to be AI, though many of these are relatively simple patient services platforms that provide things like AI-enabled self-care advice.
A better example, from the NHS Innovation Accelerator, is the National Institute for Health and Care Excellence’s approval of AliveCor’s mobile heart monitor, which uses AI to detect, monitor, and manage atrial fibrillation.
At present in the NHS, the introduction of new technologies usually comes down to individual providers. This is why there needs to be a systematic, NHS-wide approach to adopting AI within its transformation plans. Properly supported, AI stands to play a key role in meeting the targets outlined in the Five Year Forward View.
Ethics and transparency
Beyond this, the move towards AI-augmented healthcare needs to be well-informed. AI systems, particularly those that depend on machine learning models, can be opaque in their methods. ‘Black box’ systems should be used with caution, as both transparency and interpretability of AI processes are vital to healthcare. And this is assuming the data is of sufficient quality and lacks bias in the first place.
The Reform report highlights IBM’s efforts to create a healthcare AI system that interacts with clinicians in a more transparent way. Known as WatsonPaths, the platform explains its decisions and qualifies its recommendations with percentages that communicate its level of confidence in them.
Arguably, AI systems could play a crucial role in standardising high-quality care – assuming that data ownership is clear. IBM is one vendor that is on record as saying that it believes that cognitive data belongs to the user or subject, and not to the vendor, and that this principle stands at the heart of Watson.
The ‘AI in the NHS’ report says:
AI can be deployed in healthcare to help clinicians keep abreast of advances. IBM’s Watson deploys natural language processing, which allows computers to process written information. Watson could process existing literature alongside patient data to aid diagnosis and then recommend treatment options to clinicians.
There are also ethical questions around how data is handled. Google’s deal to purchase patient data from the NHS caused huge controversy, highlighting important questions about who owns this data and what they should be permitted to do with it.
There are huge gains to be had from sharing patient data, in the training of AI, but it must be authorised, regulated, and transparent – and in most cases anonymised.
As the next steps are tied up with regulatory and budgetary concerns, the onus rests on the government to show that it is serious about the priorities laid out by the Science and Technology Committee, ensuring that society benefits from the immense opportunities presented by new technology.
The new Office for AI in Whitehall is likely to be critical to the government’s plans, as it seeks to spur AI adoption across every part of government and the public sector. However, its presence within the unwieldy Department for Digital, Culture, Media, and Sport suggests muddy thinking in Whitehall, despite the government’s clearly stated ambitions in the field.
Additional reporting Chris Middleton.
Internet of Business says
AI can be the great healthcare enabler. It can be brought to bear throughout the whole healthcare ‘journey’, from risk detection and diagnosis, to aftercare. IoT even has biotech R&D covered.
The potential gains from AI adoption are clear, therefore, but there are obstacles to overcome along the way. The first of these isn’t technological or regulatory – it’s a question of mindset.
For widespread adoption to occur there needs to be ubiquitous trust and confidence in AI, both from healthcare decision makers and the wider public. This is dependant on demonstrating safety and patient outcomes at the same, or lower, cost.
The trust issue is inseparable from the question of how data is handled. The NHS still runs on piles of paperwork and legacy software systems, with staff often forced to share computers. Many of these are outdated, with unsupported operating systems, as was demonstrated by the unfortunate impact of the WannaCry ransomware last year.
Given this, there needs to be widespread modernisation of computer systems and workflows, and the adoption of open standards that can be applied across departments and organisations (including public health, clinical science, social care, local government, and public representatives). This needs to happen long before AI can become part of the fabric of the NHS.
And citizens’ informed consent, in line with GDPR, will be critical too.
As the NHS’s Five Year Forward View itself admits:
Part of why progress has not been as fast as it should have been is that the NHS has oscillated between two opposite approaches to information technology adoption – neither of which now makes sense.
“At times we have tried highly centralised national procurements and implementations. When they have failed due to lack of local engagement and lack of sensitivity to local circumstances, we have veered to the opposite extreme of ‘letting a thousand flowers bloom’. The result has been systems that don’t talk to each other, and a failure to harness the shared benefits that come from interoperable systems.”
No doubt rectifying this will require huge central investment – funds that won’t be easily extracted from the Treasury.
One problem is that the NHS National Programme for IT, instituted under the Blair Labour government, was the most recent attempt at creating a centralised system to unite a disparate, locally run health service around core technology platforms.
Its ambition, combined with poor detail and management, led to the NPfIT being an unmitigated, over-budget, under-specified, overdue disaster that descended into vendor litigation, censure by the Public Accounts Committee, and a deeper fragmentation of NHS systems.
However, taking a longer-sighted look at the NHS of the future, it’s clear that across-the-board tech investment can’t come soon enough. Greater implementation of AI stands to not only make processes more efficient and cost-effective but, putting a human face on it, improve the quality of care that the NHS provides.