arrow-right cart chevron-down chevron-left chevron-right chevron-up close menu minus play plus search share user email pinterest facebook instagram snapchat tumblr twitter vimeo youtube subscribe dogecoin dwolla forbrugsforeningen litecoin amazon_payments american_express bitcoin cirrus discover fancy interac jcb master paypal stripe visa diners_club dankort maestro trash

Shopping Cart

Moonshot for Life

Making health data work for all

Making health data work for all

“We set out to answer the question: what could be deemed a fair value exchange between patients, the NHS and industry, when there’s access to health data to create a product or a service. It was actually an area where literature and thought in the public discourse was lacking a bit. It’s an area that can be quite contentious in the eyes of some people, as it can easily be tagged as ‘privatisation through the backdoor of the NHS’, which is a very sensitive topic in the UK. But, I thought, in order to make the most of healthcare technologies, you need to have that uncomfortable conversation around data exchange – it’s a necessary one to really understand what the public deems as fair, what kind of avenues could they use to have their voices heard beyond binary ‘opt in or opt out’ data-sharing mechanisms, and what really are their thoughts about commercial models within the NHS.”

Those are the words of Eleonora Harwich, Director of Research and Head of Digital and Tech Innovation at Reform – the leading public service reform think tank in the UK – who recently published the report ‘Making NHS Data Work For Everyone’. It looked into the core question at the root of the AI dreams in healthcare – what do we do with the data? 

Realising The Potential of Health Data

At the moment, data is currently used in the NHS – the UK’s National Health Service – primarily in direct-to-patient care. There’s a secondary usage by research institutions to produce new findings, and by commercial organisations to build products and services for the health system as a whole. In some sense, the bulk of the usage being direct-to-patient seems right, as general discourse in 2019 would suggest that people aren’t happy for their data to be used without their knowledge and for means that aren’t clear, and, of course, it means better healthcare for those individuals. Trust, and understanding, is low.

But in our era of AI starting to show potential for several health use cases – from managing digital health records to reading scans and augmenting decisions around diagnosis – the current skew towards direct care means the data isn’t well set up for new innovations to be built around it, due to the type of data collected and the means by which it’s recorded. Advocates for faster implementation of machine learning might argue that the NHS data – years of information on an entire nation’s health – can and should be used better to develop technologies, find unexpected patterns and bring the health system into the 21st century, even if it means going to a private company. Simply put, there’s an argument to be made that the value of the health data simply isn’t being realised. 

It’s not quite as straightforward as a push towards better data management. As Harwich explains, you have to work out who gets to use the data, in what format, and at what price: “What does the value of data mean? That’s a very difficult question. How you value data depends on who is looking at it – there are lots of things that might affect the value each party deems fair.”

Are we Sitting on a Gold Mine?

If you consider the three main stakeholders when it comes to health data, you have the patient, the health system, and the technology provider. If the health system is to sell data to a technology provider, of course the provider gets more, better data to build a better product with; the health system gets money in the bank, a good product – eventually, you would hope – and possibly some technical expertise to implement into its work; the patients – indirectly, eventually – get better care.

But is the data ‘good’ enough to warrant a price? Will the technology provider have to spend a lot of time cleaning it before it’s usable? Should the NHS be paying the provider – as they would a supplier of linen, for example – to create for them a superior product? Should the patient benefit monetarily from the sale?

For Harwich and the report, the answer is not immediately clear: “Some people say that the NHS is sitting on an absolute gold mine, and that we really need to protect it; and then you have the other extreme which is people saying the data is of such bad quality anyway that it’s completely worthless. I guess the truth lies somewhere in the middle.” That doesn’t mean that thinking about it isn’t important, however: “If the healthcare system is overprotective of what they consider a very valuable asset, it might actually lose out on some of the innovation that could have been created, had that protection of assets been a little bit more relaxed.”

A Framework for Fairness

“There’s no national index of data sharing agreements between NHS and industry, which could give an indication at least of how much data is currently being shared, and because of that there’s not a registry describing the commercial models happening on the ground. So we tried to a bit of detective work in working out what kind of commercial arrangements are happening,” Harwich explains. “We explored models of proportionate governance that have been used in Scotland to allow for data access requests for secondary uses such as academic research and R&D by private companies, and we thought that kind of model could be quite interesting and potentially used as a medium to have that conversation about commercial models more broadly.”

Harwich and her team’s paper wasn’t just about exploration, though: “What we called for in this paper is a clearer national strategy, something that goes a bit further than a code of conduct. There needs to be a very clear strategic decision as to what are the different types of commercial models that are fair, can develop on the ground, and can lead to a good national outcome.”

Fairness is a key facet of the discussions around commercialisation of NHS data. For example, if a local hospital uses a data sharing model where they get an equity stake in the technology provider, that could mean revenue stream back into said hospital – a great way of bringing income into that local health system. But on a national scale, allowing localised sales and incentivising the system as such may mean running the risk of some trusts becoming much richer than others, which can lead to healthcare inequality from one area to the next. 

Making Health Data Work For All

The conversation around data sharing when there’s a national healthcare service at play is an interesting one – not only because it’s a balancing act between the parties involved, but because it brings up more philosophical and ethical questions about whether the benefit should go to the individual versus the whole.

In terms of benefitting the whole, there’s some interesting ideas out there in terms of monetising the data across the board, and using the revenue from data sales to create a sovereign wealth fund with profits funnelling back into the healthcare system. “To me it’s an interesting conversation to have with the public as we are facing a really difficult situation financially with the NHS, and even if more money is poured in in 2019, it’s unlikely to be enough. At the moment, the only way to do that is by raising taxes, but many are not keen for that to happen.” This is a very different conversation to be had with the public – instead of just talking privacy, data sharing and ethics around commercial profit, the discussion could also be about how best to fund the UK healthcare system; a decision on how best to improve healthcare outcomes for the entire nation.

On the more individual side, there’s been some innovation in the world of blockchain, with some companies building products that would allow the individual – instead of the healthcare system – to sell their data on a case by case basis to those requesting it. “I think it’s an interesting model, at least in the sense of devolving the decision-making and potentially giving a bit more of a meaningful choice around data sharing as opposed to a simple opt-in-opt-out framework,” Harwich says. “I just really wonder how practical the dynamics are of dynamic consent. I wonder about the ethics and the whole framing of the requests you receive – if you have these little pop ups on your personal healthcare records asking if you want to share your data for this specific piece of research, depending on how you phrase the question you might get very different outcomes – and I don’t know if there’s anyone looking at that. And then, if you devolve the decision-making to the individual, I think you can end up in a situation of the tragedy of the commons, where no one will have the incentive at the individual level to donate back to the collective. Actually, the amount that you could get for the collective is way more than that of the individual – I highly doubt that, given a decision, you would forgo this £5 to donate it to the NHS instead of keeping it for yourself.”

The discussion around data sharing is far from over – but if the ‘brave new world’ of AI-powered healthcare is to be truly realised, more decisions need to be made, quicker. There is a sweet spot that can be found which brings the most benefit to each of the parties, but the only way to get there is to have difficult conversations and bring the unanswered questions into the public eye. Harwich and team, with their report, have made a start.