The government earned glowing headlines this week for its pledge to create a £250 million National Artificial Intelligence Lab for healthcare, but details on how the eye-popping sum will be spent remain sparse.
A Department of Health and Social Care (DHSC) press release said that the AI Lab would bring together academics, specialists and technology companies to work on healthcare challenges including earlier cancer detection, new dementia treatments and more personalised care. It added that the AI Lab would help clinicians estimate future needs, identify patients who could be better treated in the community and automate routine admin tasks.
The lab will sit inside NHSX, the new digital healthcare unit launched by Hancock earlier this year, which will run it alongside the Accelerated Access Collaborative, a collective of healthcare organisations and industry representatives established to drive innovation in the NHS.
However, none of the money will be spent before the 2020/21 financial year, and where it is eventually invested remains unclear.
Where is the money coming from and going?
Critics have questioned the source of the £250 million figure is in fact new funding, after prime minister Boris Johnson's recent announcement of a £1.8 billion "cash injection" for the NHS was revealed to largely be money that Trusts had already earned as compensation for stretching their efficiency.
NHSX was quick to provide reassurance that it was indeed new funding.
"Yes the AI Lab is new money," the unit confirmed in a tweet. "And yes we need to fix the basics too, but it's not either/or. Staying ahead of where the tech is going is good for the NHS. One reason why the NHS struggles with interoperability now is that we failed to futureproof our technology in the past."
This futureproofing will only succeed if the tech sector’s desire for funding and the health sector’s need for effective products and services can overlap. As the funding will only begin in 2020, it will likely be years before it produces clinically-tested treatments.
Dr Jorge Cardoso, CTO at the King’s London Medical Imaging and AI Centre for Value Based Healthcare, has helped design an AI platform for faster processing of NHS data, and hopes his unit will be able to collaborate with the new AI Lab.
"We could actually have a properly aligned strategy where we can do some of the more crazy advanced research, while NHSX could deliver the national translational version of all of that," he said. "A lot of the algorithms we are going to do will work locally and will work well in other hospitals. But NHSX could then go through the process of translating all of the work that we've been doing and other centres have been doing and translating that nationally to all of the other Trusts in the country."
His hopes may well be realised according to Sam Smith, coordinator at privacy campaign group medConfidential. Smith told Techworld he expects a lot of the work to be conducted by academia and NHS research outfits, and questioned the benefits that corporate giants could bring to the project.
"I wouldn’t expect the lab itself to do much of the work - a lot of it will go to academia and NHS research outfits. Probably. It’s where the expertise is," he said.
"If an AI has to be redone for each individual hospital, that’s not going to go very far. That’s not how national NHS diagnostic/triage norms work. although NHSX can change those norms if they wanted to - that they have so much power in this space is why there are more general concerns about NHSX."
Data protection is another area of concern. The lab will need access to quality healthcare data if it is to develop effective AI systems while complying with GDPR. It will also want to avoid the type of controversies that plagued the partnership between the NHS and DeepMind, which was accused of breaching the Data Protection Act and breaking its promise not to share patient data with parent company Google.
“The NHS will need to find quite a significant amount of expertise to be able to deploy, deliver and develop all of these tools, and NHSX doesn't necessarily have direct access to data,” said Cardoso. “It will be interesting to see it NHSX prefers to work with the trusts as data providers, and how are they going to design the models associated with that.”
There are also fears that some companies will receive preferential treatment when bidding for contracts, which have been exacerbated by NHS’ recent deal with Amazon to provide healthcare advice through Alexa.
Smith questioned whether many multinational companies would have the inclination and expertise to solve the problems the NHS actually rather than focus entirely on what would make them money.
He argued that their work could have results similar to DeepMind's successes in detecting eye disease at Moorfields Eye Hospital NHS Foundation Trust, but may instead be more comparable to company's controversial efforts to use patient data to develop a mobile app to assist Royal Free Hospital clinicians.
“Will it look more like the DeepMind/Moorfields project which advanced science, or the DeepMind/RFH project which was unlawful and which has developed nothing that anyone other than DeepMind can use? This is a policy question that should have been addressed in the number 10 splash yesterday, and wasn’t because no one has thought of the detail," he said.
“The uncertainty about what will happen is not good for startups who need to know whether Hancock and number 10 just punched a hole in their business plan. While the NHS lab won’t start until 2021, the announcement changes plans already.”