Last week, the government set out plans for a new code of conduct to ensure the NHS and taxpayers “get a good deal” on partnerships with technology companies.
The move comes at a critical time for the health service. Despite recent assurances over government funding, many hospital trusts are still struggling under the pressure of austerity.
As technology buyers across the public and private sector know, investment in technology is one of the first things to go when budgets are squeezed, and it’s the whole organisation that suffers as a result.
If this was ever forgotten, last year’s WannaCry attack was a stark reminder. The computer virus swept through dozens of hospital trusts across the UK after IT teams failed to put in place basic protections. Systems were taken offline, forcing doctors to cancel thousands of appointments and operations.
In this climate, it’s not difficult to understand why some NHS executives are eager to hand over patient data to technology companies for free, especially when they are promised lower operating costs and dramatically improved patient care.
The government has previously lent its support to this kind of arrangement. So too has Google DeepMind, which has been credited with some of the best and worst patient data deals in the UK. But not everyone shares their perspective. In a column for NS Tech in May, the tech lawyer turned MP Darren Jones wrote:
The artificial intelligence drive in the NHS is welcome, but we must recognise that this intelligence comes from the machine learning capabilities of algorithms processing huge amounts of data. Data that belongs to NHS patients. Without their data there is no intelligence. So when patient data in the NHS is handed over to commercial companies, it should be made very clear that we expect to share in any profits made off the back of using an NHS data-derived algorithm anywhere in the world.
The code of conduct suggests that the government may have changed its position a little. In a press release posted on gov.uk on Wednesday (5 September), the government promises that above all else, the code will secure a good deal on partnerships for both the NHS and the public. And the code itself specifically addresses Jones’s point about algorithms.
Here are the two relevant passages:
Compliance with laws and rules around data: is there clarity on what will happen to the data (including for patients)? Interesting points here are around the use by AI of data as ‘training sets’ from which an algorithm can learn (increasing its value), but not retain.
Ownership of intellectual property. There are numerous models in use across this area, broader medical and pharmaceutical research, and across academia. Which is most applicable will depend on a host of factors. This is important where the same algorithm is used in multiple applications – the participant (or data source) in each application cannot hope to own the underlying algorithm but might hope to share in the increase in value.
Many in the open data community have been pushing for the government to produce guidance on this issue, and will welcome the code. But some might be disappointed to see that despite public concern about data privacy issues and NHS funding, these principles are non-binding and in some cases loosely worded; they will be impossible to enforce unless the government legislates in the area.
Ultimately, however, this latest document is just a conversation starter – a public consultation will follow – and while it may raise more questions than it answers, at least the government is finally starting to listen.