How can data help in the pandemic?
In a recent virtual panel hosted by CogX, director of AI at NHSX, Indra Joshi and special advisor for AI and digital health at NHSX, Dominic Cushnan, convened to discuss how data has helped during the pandemic. Interesting insights were shared by both speakers outlining how the COVID-19 Data Store has assisted in showing how evidence is produced for the safe deployment of artificial intelligence (AI).
During the discussion, Joshi highlighted the work being carried out in setting up the NHSX AI Laboratory.
“The AI Lab is there to set up a way of ensuring that any AI technology that is being either designed, built or deployed by the NHS, is safe and effective and essentially doing as it says it would. It is a complex task and we’re moving onwards with the regulations with very good partners like the Office of AI.”
Opportunities for the broader system
In recent times, NHSX has been tasked with making sure the UK is prepared to design these technologies but also embed and deploy them into the health and care setting. NHSX has subsequently written several reports and guides such as, 'Artificial Intelligence: How to get it right', in order to help solve problems such as image recognition and the triage of images.
“Imaging is definitely a space where we know we have good data. This is where we can see some good impacts coming pretty much now. Diagnoses is still some time down the line. As much as you train the model, you’ve got to have a very vast dataset to do that,” said Joshi.
In fact, the National COVID-19 Chest Imaging Database (NCCID) was consequently set up in response to the start of the pandemic to support algorithms in images.
“What we were able to do quite quickly was start convening a conversation about how we collect images so that we can understand it from a disease-specific route. How do we help those organisations that have those technologies that can be used during COVID-19 for chest images and detection of the disease? How can we get them to be validated on independent data?” explained Cushnan.
They have since worked with the Hennepin Healthcare Research Institute (HRA) and received ethical approval for the programme to make sure that the database has the ability for research to happen on top of it.
Ensuring the ethical deployment of AI
NHSX AI Lab’s mission has been to accelerate the safe adoption of AI and provide effective tools to make informed decisions. As an extension of this, the recently published 'Buyers Guide and Checklist for AI in Health and Care' was mentioned in the discussion for its aims in providing policy context and an overview of the opportunities presented by AI.
“It helps buyers think through what they need to do to buy safe and effective AI and also how developers and vendors are aware of the buyers’ expectations,” said Cushnan.
Furthermore, the importance of using AI to inform decisions was emphasised by Joshi, adding that a key aim was 'to ensure that we educate people and bring a community along with us so that people understand what they’re doing and feel confident in the technologies they’re using, and also how they train people for how they use the data in a slightly different way to the way they might be using it.”
NHS Futures has been utilised to bring together data analysts into an era where big data and big analytics is critical and has produced several webinars, programmes of work and content.
“We’re doing this both at a national, but also at a regional perspective and showing people how you get involved with that and change the dynamic and conversation, not necessarily a top-down or a central-down response but actually a community of proctors who are moving forward,” added Indra Joshi.
What does this technology currently mean for the health system?
Another area of progression mentioned was predictive analytics, particularly in the context of the NHS Data Store. The COVID-19 NHS Data Store was set up to look at the operational response of action taken by the NHS.
“We’ve covered certain datasets that allow us to look at how the NHS is managing some of its operational response such as bed capacity, oxygen and ventilators.”
“We were able to look across at the numbers at a national perspective to see the regions we needed to focus our attention on or give more aid to. That different thinking in how we use data and numbers has really helped make sure that we’ve given the frontline a positive response. This is something we’re helping with and using the numbers to look across regions, versus quite traditionally where you look at localities and don’t see the national picture,” concluded Joshi.