AI has the potential to transform the NHS
We must address all of the challenges if Artificial Intelligence is to save time, money, and lives, writes Lord Holmes of Richmond
These are exciting times. Artificial intelligence (AI), already affecting us in ways we may not currently even be aware of, will completely transform our lives in the next 25 years. One of the areas in which this transformation has the potential to be most significant is in healthcare.
In the UK, our crowning glory is the National Health Service. The principle of universal healthcare free at the point of need is, rightly, a point of national pride. I’m still moved by the memory of Danny Boyle’s tribute to the NHS in the 2012 Olympic opening ceremony.
To best understand AI, it is helpful to think of it as a new type of cognition that can be added to the symphony of human intelligence generally. Intelligence is not merely a single scale of low to high as measured by IQ tests; there are many different types of intelligence. AI can be added to this symphony and we should be genuinely excited about what this will mean for our future.
AI can also be conceived as a utility to drive efficiency and productivity. Human intelligence can be freed from certain tasks and applied to other disciplines. In this we see the real potential of the strength of a team of human and artificial intelligence. The augmented human or, in the NHS, the enhanced consultant – freed to do what they do best and spend more time with patients.
To steer the specifics we must embrace the trend, the idea, the potential – which is why I was pleased that one of the grand challenges set out in the government’s industrial strategy is the mission to use data, AI and innovation to transform the prevention, early diagnosis and treatment of chronic diseases by 2030.
The message is important but practical steps must also be taken to promote UK-based R&D and innovation. This change is inevitable and if we don’t develop the technology we’ll be licensing it from others and the NHS will be paying for it.
'We need to improve data storage, availability and the flow through the NHS at pretty much every level'
AI is already being used in the NHS in diagnostics and non-clinical or administrative applications. UK hospitals are facing an acute shortage of human radiologists. Medical scans provide relatively structured data that is easier to analyse in the development of AI diagnostic solutions. Several companies are working on AI algorithms to diagnose breast cancer. Google’s DeepMind has also worked with Moorfields eye hospital to develop automated analysis of retinal imaging. The development of AI in diagnostics may not even require large or expensive datasets; there is some fascinating research from a team at MIT which developed a diagnostic tool using datasets from just 50 mobile phone images rather than 10,000 expensive medical images.
Data is the essential part of this and it is crucial that we address all the issues and challenges. The medical data of over 65 million people is, collectively, a dataset with incredible potential value but must be handled appropriately, not least as a result of the implications for public confidence if we get it wrong.
If people are not convinced that their privacy rights are safe then public trust, not yet secured, will be eroded, as demonstrated when the Royal Free hospital failed to comply with the Data Protection Act when handing over patient data to DeepMind. We need a mechanism for demonstrating societal benefit from the data as it is being used.
We also need to improve data storage, availability and flow through the NHS at pretty much every level. Data quality, sharing and interoperability is absolutely key to realising the potential of AI. Our House of Lords Select Committee report on AI in the UK called for a framework for the sharing of NHS data and stressed that the quality of NHS data be ensured through careful (and consistently formatted) digitisation of current practices and records.
Another challenge – how to prepare the healthcare workforce – has been considered by the Topol review. Within 20 years, 90% of all jobs in the NHS will require some element of digital skills. It is essential that we invest in people as well as technology. NHS organisations will need to enable a learning environment and create a culture of innovation, prioritising people, developing an agile and empowered workforce, as well as digitally capable leadership, with effective governance and long-term investment.
If we get it right, the NHS can continue to be the greatest healthcare system in the world, and in addition can become the greatest example of the deployment of AI for the public good.
Lord Holmes of Richmond is a Conservative peer and vice-chair of the APPG AI