Improving AI development in health and social care
Elizabeth Bohm established the Data and AI Policy team at the Health Research Authority (HRA) and oversees their programmes to future proof the way that research studies using artificial intelligence (AI) and data-driven technologies are approved. We asked her to outline the HRA’s work in this area and what it will mean for researchers and developers.
The UK is positioning itself to be world leading in AI. The government has set out a number of Grand Challenges in its Industrial Strategy, one of which is around data and AI. Its mission is to use data, artificial intelligence and innovation to transform the prevention, early diagnosis and treatment of chronic diseases by 2030. In order to realise this goal, the government has invested £210 million into the NHS AI lab, which is run by NHSX.
The NHS is an incredible resource, often envied by other countries with its focus on healthcare that is free at point of care and the vast volumes of data it generates. This data is estimated to be worth nearly £10 billion annually. Releasing this potential is no mean feat. There are challenges to being able access high quality data and use it well. This can be further complicated by the number of different regulators and regulations in this space.
Research and AI
The HRA has a role to play and we established the Data and AI Policy team to make sure we could do our part to ensure that the public benefit could be realised from using health and care data to develop products to improve health and social care.
The focus of our work is in three areas:
- streamlining our processes to ensure research involving AI and data-driven technology receives the appropriate review as quickly as possible
- working with partners to demystify the regulatory landscape for AI and data-driven products and assessing where we can work together improve the regulatory pathway
- increasing the transparency of our processes and championing transparency in data use to evidence our trustworthiness in using health and care data.
To inform the development of our programmes, we conducted user research which highlighted a few areas that we need to focus on.
Firstly, those outside the health and care sectors, for example computer scientists, mathematicians, bio-informaticians and technologists, who are doing exciting work in this area, may not be familiar with the HRA or our role. So a quick introduction.
The HRA is funded by the Department of Health and Social Care and we protect and promote patients (and their data) in health and social care research. We ensure that the research is safe, legal and ethical and that those with the appropriate lived experience have contributed to the research. We also have a role in accessing a particular kind of health and social care data – confidential identifiable patient data without consent.
Data and proof of concept
This brings me onto the second thing we heard from our user research – to develop a product developers need to access data quickly, particularly at the proof of concept stage. This is something that we will be exploring with our partners, however, there also seemed to be low awareness of the range of datasets available.
We understand that there may be a need to use large identifiable patient datasets and that it may be impracticable to gain consent from all of the participants. But this data represents the highest risks to the participants, and we have to make sure that only the data that is actually needed is included and it is being used in the public interest. So although it is possible to apply to the Confidentiality Advisory Group for approval to access this data, it is worth considering if this the data you actually need to develop a product. Could anonymous, pseudonymous, synthetic or simulated data be more appropriate in the early stages of development?
Improving the AI ecosystem
We are working with partners to clarify regulation and streamline the process in order to help you access health and social care data. We’ve heard that it needs to be easier at an earlier stage to identify which AI and data-driven activities are research, and exactly what approvals will be required.
For the rest of 2021, and throughout 2022, we will be scoping solutions to these issues. We will be working with partners like NHS Digital, Health Data Research UK and the Medicines and Healthcare products Regulatory Agency (MHRA) to see how we can make improvements across the system to help you get access to data quickly.
Being trustworthy
In order to build public trust, we need to demonstrate trustworthiness. This is why our priorities are legal and ethical research, public involvement and transparency.
We know that people are happy for their data to be used in research provided there is a public benefit but that they have concerns about data being accessed by companies or used by those outside of their care. The recent reaction to the data extraction proposed for the General Practice Data for Planning and Research and resulting significant increase in national opt-outs, shows the importance of public involvement in ensuring that health and social care data is used appropriately and transparently.
As part of our work we will be increasing the transparency of our processes and working with others to ensure that patients and the public know how their data is being used and the safeguards in place. We have seen good practice in patient and public involvement and hope that you will join us in our mission to build trust amongst the public in how health and social care data is used through transparency, public involvement and showing the public benefits of using it.
Help us improve
We would really value your input in our work. If you are at an early stage of conceptualisation and would like to be involved in helping us change the regulatory landscape for the better, please contact will.navaie@hra.nhs.uk We can help you to understand how to best navigate the regulation and the most efficient way to get access to data.