The state of AI and UK healthcare

The COVID-19 pandemic has intensified the pace of global change and the pressure on companies to respond. But, despite rapidly building their digital capabilities during the pandemic many  organisations are yet to fully embrace transformation.

There has been a rapid uptake in the adoption of digital resources and data in UK healthcare since the onset of Covid-19 – from the government’s new NHS data strategy announced in June, to the launch of the NHS Covid-19 app, which had been downloaded 10 million times by July 2021. The result of this surge in activity, has raised the question, amongst members of the public, of how this data will be used, while for healthcare professionals and pharmaceutical companies, it has paved the way for innovation possibilities stemming from this digital transformation.

Continuing this momentum could enable the UK healthcare industry to make significant strides in fields such as diagnostics, drug delivery, health, and social care. By also using this data to apply Artificial Intelligence (AI), it can provide healthcare professionals with the critical insights which can fuel innovation. This could drive new outcomes from improving patient care and reducing clinician burnout to up levelling the overall care experience and lowering costs.   

Accenture figures estimate that AI applications in healthcare could save up to $150bn annually by 2026. Given the ongoing challenges faced by the NHS to provide universal healthcare to all UK citizens, cost savings on this scale could help keep the doors open.

This article explores the current state of AI in the UK’s healthcare system, its potential and what steps are needed to realise this.


AI will support rather than replace health workers

There’s a lot of confusion around AI, so let’s start with the current state of play in UK healthcare. Essentially, there’s two types to be aware of: Narrow AI and General AI. 

General AI is where a machine can start solving a problem as a human would, or perhaps even better than we would, however instances of machines actually doing so are rare. Indeed, this is usually shown only through gimmicky applications of AI such as playing chess or completing puzzles. For healthcare, the focus is on Narrow AI – more specific use cases – where rapid organisation and analysis of data can support healthcare professionals to make better decisions more quickly. 

This is not a way to replace healthcare roles in the future, but to support individuals to achieve a more intellectual layer of capability through the support of AI. Here, we’re talking about levelling up healthcare professionals to support in the more menial, time-consuming tasks. This means, at an educational level, our approach to teaching in healthcare must adapt to ensure human expertise is being used to its fullest potential. 

Radiology is one such area. AI is already impacting radiology more quickly than other medical fields. We should view any uneasiness among radiologists to embrace AI as analogous to how airline pilots were reluctant to embrace autopilot technology in the early days of automated aircraft aviation. Today there’s a certain comfort in knowing that pilots are supported by cutting edge navigational systems and aerospace engineering. AI is not going to replace radiologists. Instead, it will simply allow radiologists to increase the number of patients they are able to process, accelerating diagnosis and treatment.

To succeed, radiologists must now be aware of the basic principles of how AI observations are obtained and how they should be interpreted. Datasets used to train AI models do have limitations and can potentially include bias. In this new paradigm, radiologists will need to know how to interact with AI solutions, how to flag studies that provided incorrect results or failed AI processing, and how to interact with the data scientists and IT supporting these solutions.


It’s not an AI project, it’s patient outcomes

Too often we see time spent discussing AI as the end goal – the sole focus of a project – when really, it’s simply a means to an end. The purpose of any healthcare project should have patient outcomes at its heart. 

When discussing projects involving AI in healthcare, it’s vital to focus on the data – the availability of ‘clean’, labelled data to train deep learning models and the need for datasets that represent diverse patient populations. Additionally, focussing on the importance of an appropriate infrastructure to support moving large amounts of training data quickly and efficiently to graphics processing unit (GPU) compute environments.

These areas are crucially important to the success of a project, but the narrow focus on technology can reduce the efficiency of a project and prevent those involved from getting to the end goal. 

Shifting the focus from what AI can do, to how it can help achieve a set of more specific objectives to deliver a beneficial patient outcome, will be critical to AI maturing as a healthcare technology service, rather than remaining the preserve of technologists. It is also imperative that ease of use is a focus for any AI-incorporated project to ensure healthcare professionals are actively using AI rather than feeling that it is being imposed on them.


Implementing AI and cloud in the UK’s healthcare system

There are many considerations here and a key one is data – data outcomes, biases, and influences. These need to be understood and addressed from the outset to understand influences that have the potential to impact outcomes for years to come. 

But the issue that today’s industry faces is multifaceted. Remaining agile and keeping options open is of paramount importance, as is setting realistic and achievable objectives. Another is processing systems, with the need for compute power to process data at pace. Here, understanding storage as a key element in achieving data access and agility is key.

Data and resources already exist across complex ecosystems of in-house datacentres, private and public clouds, and trusted delivery partners. Embracing new cloud technologies and attempting to place data at the heart of everything requires alignment of policy, governance and risk management through software development and architectural disciplines, right down to the underlying IT infrastructure. Realising this data-led vision and shaping its future pathway ultimately comes down to how data is stored and processed at the very foundation of the IT stack. 

The government recognises this need, stating that it “must expand work to treat data as a strategic asset, and create a whole-government, collectively responsible approach to investing in data foundations, so that everyone can benefit from the improved outcomes data can offer.” 

We recently found that digital wastage – paying for services not used or failing to maximise the potential of these services – could be reduced by up to 60 percent using hybrid multi-clouds. When it comes to the UK’s healthcare sector, it can sometimes be easier to secure a large sum in one go as opposed to a smaller sum on a recurring basis – but whilst cloud migration may seem like a destination, it’s actually a journey that needs to focus on patient outcomes and take clear steps to ensure maximum roll-out flexibility and minimal impact on service delivery.

All of this is underpinned by the need for collaboration – bringing disparate groups and sets of data together to deliver against an outcome-focussed project. Again, here there needs to be a clear and defined focus on the project, understanding how technology can provide support but not allowing it to become the end goal. A great example here would be the innovations in AI to support glaucoma pixel examinations. AI is used in imaging analysis to find the white specks that are the root cause of glaucoma and spot its development early, with the overall goal to identify the problem early and ensure patient treatment can happen quickly. 


So, what’s next?

The recent announcement of the government’s new NHS data strategy highlighted the fundamental importance of data in modern healthcare decision making. If harnessed correctly, the data retained by the NHS has the potential to provide huge value to research and enable health and care staff to make quicker, more informed decisions with up-to-date information and support from AI. 

However, such a drive does not negate the fact that, within a wider data strategy framework, cloud adoption and migration is an intimidating task. Much of the industry’s discussion seems to jump from A to Z in the migration process; cloud and legacy systems are positioned as almost opposing forces, with resources often organised in the same, uncoordinated way. The reality we face is a complex array of interdependent systems that cannot simply move from A to Z; some of them may never move. And if we are to harness the information stored in these systems, it needs to be made accessible as well as secured with due regard to patient confidentiality.

With disparate and legacy systems across the UK’s healthcare system still in place, a data-first mindset will be essential to making information readily available at the point of care. Data was a vital resource in dealing with the pandemic and now is the time to press forward and invest in the infrastructure for the needs of tomorrow as we face post-pandemic challenges.  


Tim Skinner is public sector director at NetApp