This week, Health Secretary Matt Hancock launched a report with the Taxpayers’ Alliance (TPA), calling for increased automation of NHS services through partnerships with private sector technology companies. The pairing is highly questionable in itself – the (misnamed) Taxpayers’ Alliance is a right-wing lobby group backed by big business, which seeks to drastically reduce the role of the state and has previously called for replacing the NHS with an “insurance-based model” of healthcare.
But that’s not all that should concern us. The report is just the latest in a string of moves by the Health Secretary to open up hugely valuable, publicly-held NHS datasets to the private sector. Vast amounts of private NHS data have been transferred to private firms, under the auspices of recent public-private partnerships to develop artificial intelligence and other data-driven technologies for the NHS. Some of them are highlighted in the TPA report itself. One such example is the partnership between the Royal Free NHS Trust in London and DeepMind, a wholly-owned subsidiary of Google’s parent company Alphabet. Announced in 2016, the agreement provided DeepMind with access to 1.6 million patients’ medical records, which it would use to develop and launch its new healthcare app, Streams. Although the way DeepMind handles NHS data has changed since the Information Commissioner’s Office deemed the agreement to have breached the Data Protection Act, the company has continued to develop and scale the app at further NHS Trusts across the country.
In the case of DeepMind, the company has stated that the app could help save lives lost each year by enabling doctors to identify life-threatening illnesses like sepsis more quickly. These sorts of big claims made at a relatively early stage in product development are fairly common in the private medtech sector, partly due to its reliance on venture capital and the associated demands for promises of high returns on investment.
While occasionally these technologies do yield good results in the end, the sector has largely not been held to the same regulatory or clinical standards that we have for the biopharma or biotech industry, for example. There have been damaging consequences for individuals, investors, and public infrastructure, several of which emerged last year. Users of the app Natural Cycles were misled by adverts claiming it was a “highly accurate” method of birth control, the Advertising Standards Agency ruled. The founder of Silicon Valley-based bloodtesting technology company Theranos was charged with criminal fraud for making false claims about the technology’s effectiveness. And a public-private partnership between a Danish public healthcare provider and IBM folded at the end of last year, with a former official involved in the negotiation saying: “It was very oversold, what IBM Watson could do. There is something of the emperor’s new clothes about it.”
It’s a complaint echoed by Martin Kohn, IBM Research’s former chief medical scientist, quoted in a recent article in engineering magazine IEEE entitled “How IBM Watson overpromised and underdelivered on AI health care”. “Merely proving that you have powerful technology is not sufficient,” Kohn commented. “Prove to me that it will actually do something useful—that it will make my life better, and my patients’ lives better.” In terms of peer-reviewed papers in the medical journals demonstrating the benefits of AI to patient outcomes and public health systems, Kohn added “to date there’s very little in the way of such publications, and none of consequence for Watson.”
The political economy of privatisation
The idea that the merely rolling out new technologies in the NHS will save public money is misguided. In fact, a key driver of healthcare cost growth in publicly-funded health systems is the procurement of new technologies. This is partly because they require new or adapted skills from the healthcare workforce, partly because so many of them fail to ever improve patient outcomes.
It’s just one of a number of misguided ideas at the heart of the market approach that has dominated health policy for decades. An approach that sees patients as primarily economic units to be managed more “efficiently”, that sees privatisation and outsourcing as the solutions, and that wrongly assumes it’s the private sector that generates innovation. This political economy of healthcare is crucial for understanding public-private partnerships that involve the transfer of publicly-held datasets to private hands.
Creating public value
Perhaps most damningly, the public sector has not even succeeded in recouping a fair share of the immediate financial value of the data that is being used to develop new healthcare technologies. Providing access to data is seen as a secondary element of such partnerships, rather than recognised as what enables private profit and company growth. There is no doubt that this access is contributing to the bloated capitalisations of the companies producing the technologies and the bank balances of their investors. Google’s parent company Alphabet earned $39.27bn in revenue in the final quarter of 2018 alone.
The data held by public sector bodies is so valuable because it is unlike anything the private sector can collect by itself. Patient data held within NHS trusts and other public bodies may be decentralised and messy in parts, but it is often far more structured and comprehensive than health data the private sector can collect by itself. The failure of Google Flu Trends proves just how difficult it is to make worthwhile use of the unstructured, non-probabilistic big data captured by digital platforms. Ultimately, we have to assume that if Google could develop these technologies without accessing NHS data through a public-private partnership, it would. Google markets many other services and products to public sector organisations around the world.
Beyond its transaction value, a further risk of assuming the private sector will develop groundbreaking innovations in healthcare is that public sector research and development will remain underfunded, despite evidence that this is where many of the lifesaving technologies that we take for granted today were first developed. Research has also shown that far from driving innovation in healthcare, the majority of new medicines patented by the pharmaceutical industry in recent years in fact offer little therapeutic advantage over existing treatments.
New asymmetries and dependencies
Perhaps the most significant danger of not only providing private actors like Google with access to these valuable datasets, but also investing in them to develop new technologies instead of our existing public sector research institutions, is that we are creating today new institutional dependencies on these companies to deliver services in the future. Once the partnership with a technology company ends, and its technologies have become established infrastructure or tools in the NHS, what political leverage will we have to negotiate new agreements that are beneficial to the public purse?
Ultimately, by relying on these unaccountable, shareholder-driven private actors to develop these technologies, deliver innovation and thus also build new knowledge bases and expertise, we are fostering new knowledge asymmetries. Public-private partnerships that involve big data transfer will not only weaken the power of the public sector in the future, but also potentially risk more instances where public actors are simply unable to assess the claims made by tech companies about what their products can do.
We should be optimistic about the potential for new technologies driven by big data to enable us to live happier, healthier and more equitable lives. But the advent of artificial intelligence is no panacea. And we should also be concerned about how private tech companies business models often rely on the exploitation of so many workers, including at the level of the extraction of raw materials relied on to build them.
Once we understand how valuable NHS and other publicly-held datasets are, we can only see them for what they are: public assets. And like all public assets – whether council housing or hospitals – the use and misuse of our data deserves scrutiny. In the short-term, we need our politicians to put a stop to Matt Hancock’s ambitions to open up our healthcare data to private interests and see it for what it really is: the latest stage in the privatisation of the NHS. If we want to ensure the rewards of innovation are not only properly assessed but also accessible to everyone, we must likewise strengthen the knowledge and innovation base of the public sector through investing in its research institutions and universities.
Which brings us to the issue at the heart of all of this: that the only real way we can capture the opportunities these new technologies present is by truly democratising their development. That doesn’t mean we never work with the private sector, though it does mean rethinking what the private sector should look like in a democracy, what kind of ownership models work for all of us, and how our accountable public sector can drive this change. Above all, it means bringing discussions about value in healthcare – and elsewhere in society – back into politics.