Launching ageing healthcare systems and people into the future is not a possibility – it’s a necessity.
With tales of Victorian era conditions on hospital wards, chronic ambulance queues and surgery delays reminiscent of the NHS’s nadir in the early 1990s, it almost seems delusional to look at the potential for artificial intelligence (AI) in healthcare.
But engage we must, as there is no future for healthcare delivery in isolation from increasing technological change. Indeed, the overwhelming need to reduce the burden on health systems here, and in the developing world, makes turning to tech ever more urgent.
Fortunately, the hype around AI guarantees plenty of activity in this space, with lots of new research, such as the excellent contribution from think tank, Reform, whose new report ‘Thinking on its own; AI in the NHS’ makes practical recommendations for its adoption.
The prize on offer is achieving better outcomes in preventing ill health, for example, through wearables and deploying the data they collect, as well as making faster, more accurate diagnoses and utilising public resources more effectively. Getting better diagnosis is not only good for patient outcomes; it means the right clinical approach is found more quickly, reducing waste and inefficiency.
Leap of faith: Transforming the NHS today & tomorrow
The effective use of information and technology is fundamental to the NHS transformation envisioned in the ‘Five Year Forward View’ and in Sustainability and Transformation Plans. This is being reflected in the NHS by, for example, the Driving Digital Maturity Programme which aims to ensure the NHS is paper-free at the point of care.
It is also found in the establishing of Global Digital Exemplars (GDEs), an internationally recognised NHS provider delivering exceptional care through the use of world-class digital technology and information. NHS England is currently supporting seven digitally advanced mental health trusts, and 16 digitally advanced acute trusts, through funding and international partnership opportunities, to become GDEs.
The NHS is also at the forefront of harnessing technology in genomics. With over 41,000 genomes sequenced so far, the 100,000 genome project is well underway. One sequenced genome is equivalent to 200GB of data – the approx memory of a laptop . Multiply that by 100,000 people and that’s a lot of data. This is the issue of fundamental importance to any AI aspirations; the scramble for good, clean data.
Currently, NHS patient data is of poor quality, in the wrong format, not collected consistently and massive concerns remain around privacy and access. According to Norman Lamb MP, whose Science & Technology Committee is currently carrying out an inquiry into decision-making algorithms, NHS digital capability is “a million miles away” from unlocking AI’s potential.
Data day: Future is now
Another major challenge is trust; the whole digitialisation and AI project relies on carefully nurturing and protecting public trust. Trust of course is the essence of all successful health treatment – a basic level of trust in your physician and in the system is required before anyone will take a prescribed medication.
When it comes to use of our personal data, we must tread very carefully. Every scandal plagues progress, from Alder Hey in 1999 to the care.data fiasco a few years back; each one erodes trust and sets us back years.
So far the UK’s approach to AI in healthcare is piecemeal; individual projects run by an array of partners using their own technology to demonstrate effectiveness. It has been estimated that there are 500 separate AI pilots across the NHS at present.
Moorfields Eye Hospital and DeepMind Health launched a medical research partnership in 2016 using AI technology to better analyse scan data with the aim of changing the way professionals carry out eye tests and leading to earlier detection of eye diseases.
A project between Genomics England and Illumina is developing a knowledge base to improve and automate genome interpretation. The tools will operate within the Genomics England secure database to enable researchers and clinicians to access information and reports more readily.
Amazon Web Services has partnered with Cerner to help healthcare providers better use their data to make health predictions about patient populations. Cerner’s ‘healthIntent’ is a cloud-based population health management platform which would give the NHS the ability to aggregate, transform and reconcile data across the whole of healthcare.
This level of data on a population would enable public health leaders to engage citizens to take an active role, and manage each individual to improve their health. If this data could reach beyond electronic health records (EHRs) and capture GPS data and browsing history too, then one can imagine how transformative – and possibly alarming – this could be.
Put aside even the risks around data breaches and the operational, regulatory and legal concerns to navigate, the public are sceptical of any misuse of their data by corporations.
Last year’s autumn budget confirmed the establishment of the world’s first national advisory body for AI, while the Centre for Data Ethics and Innovation will set standards for the use and ethics of AI and data.
What’s required now is a clear regulatory framework that enables the private sector to form strong and transparent partnerships with the holders of the data. While some predict that the future lies with tech firms seeking to buy healthcare providers as a way of ensuring collection of clean data, this trajectory must surely be resisted.
Power lies with those who hold the data and, therefore, it’s advantage NHS. The public sector and academia should be encouraged to engage with the private sector, but where our data is used to develop AI, the current bias towards a profit motive should be removed.
Some good investment from the increased R&D spending outlined in the industrial strategy is invaluable, but it needs to be more ambitious so that we see a real national imperative to drive this technology. The trade-off may be slower progress, but the result will be better constrained AI and more effective healthcare.
Ultimatey, the NHS must benefit financially from AI opportunities, and they must benefit the patient. If financial benefit falls only to commercial organisations, trust among the public could be jeopardised.
To build towards the use of AI across the NHS we need robust data ethics, clever articulation of the rewards at stake and, critically, investment. Only then can we truly arrive in the future of healthcare.