The potential for artificial intelligence to revolutionise the healthcare industry is incredible. There have been some AI technologies which have already been deployed, and others that show there is still potential for more.
A number of UK tech companies are working to transform healthcare using AI, such as DeepMind Health, Babylon Health and BenevolentAI to name a few.
Google’s DeepMind Health in particular has leveraged machine learning technology in order to boost medical research. Using the technology, DeepMind is able to analyse swathes of medical data to improve the way in which diseases are diagnosed and treated.
Focusing on the NHS specifically, AI deployments still remain at a minimum due to fear of the risks and consequences that could disrupt traditional healthcare service.
“Because the NHS is not configured to deal with these technologies it will change the way doctors work. It will change the productivity and have some large consequences for the NHS which needs to be thought through,” said Julian Huppert, Chair of independent review for DeepMind Health at the House of Lords AI committee.
“Some of these challenges aren’t new within healthcare, especially the prejudice and transparency issue but I think the NHS doesn’t have the capacity to control the risks, in some areas it can be seen as a threat to employment but it is gradually changing with the NHS now realising that they need to be more current.
“For now we need to do much more to help the NHS modernise,” Huppert added.
Nicola Perrin, head of understanding of patient data at the Wellcome Trust, argued it's crucial the NHS knows the value of data, as companies’ access to data may be a major challenge for the NHS and it is unlikely that patients want their personal data to be shared with third parties.
“One of the suspicions of DeepMind Health is the fact they haven’t made clear what their data rules are. I think there will be some parallels in data and I think it is absolutely essential for the NHS to get it clear,” she said.
Data preparation and AI adoption
“Getting it right is quite tricky to do but DeepMind Health has been looking at not pushing other companies out but instead actually welcome them in. If we make it hard for companies to apply AI here, what will happen is that other countries will bring them in.
“I don’t think there is any heavy use of training for AI yet, but people can learn algorithms and data sets in the US and elsewhere, so what we need to consider is how the NHS can do this,” Huppert said.
The NHS has begun to adopt AI, with its partnership with DeepMind Health being a prime example. Further deployments of any healthcare AI will need vast amounts of data. Witnesses at the AI committee warn the NHS lacks an understanding of this issue.
“There is still a huge amount of work that needs to be done to make the NHS more digitally savvy. We would get much more value from NHS data if it was in a much more secure format,” Huppert added.
“DeepMind is using a blockchain audit for data to look at who accesses the data. I agree that there is a problem with digital capacity within the NHS, it is clear there is a lot more work needed to re-enable the hospitals and there is a lot further to go across the whole country,” he said.
According to Google’s DeepMind, AI can transform known healthcare difficulties such as delays and important information being missed. The company argues that the use of AI tools will help tackle traditionally human-managed tasks like analysing test results or identifying whether a patient may be at risk.
However, witnesses explained to the committee that the current state of the NHS data may not be up to standard for the application of AI. Instead, it would be better if the NHS took time to understand how AI and algorithms are used.
“I think the public tend to believe that the NHS has all the data put in one place but it is absolutely not like that, it is very much in silos at the moment but that may change. However, it is about whether they can change that to make it better or worse,” Huppert said.