Dr Claire Bloomfield presented at the British Institute of Radiology event “AI in Practice” in January which was hosted by NCIMI partner Dr Amrita Kumar and Dr Hugh Harvey. She presented about the value of health data, and there were also presentations from our partners Brainomix about their experiences using AI in healthcare.
This article is a summary of Claire’s presentation which covers some of NCIMI’s experiences and the practical steps we’re taking to harness the value arising from health data. This is based on our ongoing work with Future Care Capital assessing how to support our partners in working together and generating benefit for all.
Health data- complexity at all levels
The global, national and local landscapes for use and re-use of health data is a complex and evolving one. Although no one is arguing if health data has value – of what amount (as much as £9.6 billion per year), what kinds and who should be a beneficiary of that value are all up for debate.
In the same way, one could have thought we would have reached consensus on a useful metaphor for health data itself, as oil, wine (suggesting economic value), as infrastructure (and therefore as an enabler but not with direct economic value), or possibly as an asset (to be ‘sweated’).
So in the absence of consensus on how we describe health data or how we approach valuation, what do those trying to structure and execute health data deals for research do?
As NCIMI, we are trying to support the harnessing of health data for new insight in medical imaging and AI. We do this as a broker for collection and use by NHS, academic and commercial parties and therefore need to structure deals that reflect the different parties to our collaborations. To support this we have been working with FCC to develop a framework for ensuring all stakeholders receive benefit- patients, NHS, academic and industry and practical support for deal negotiations.
Here are some of my observations through the partnerships and deals we are navigating through NCIMI.
Why does it matter?
The latest Royal College of Radiologists Census data shows that there are workforce shortfalls of over 34% nationally, rising to 44% in the north of the country for radiologists.
This shortfall in staff has been compounded by the backlog generated as a consequence of the Covid-19 pandemic. These alone represent significant challenges for radiologists, and AI could help in addressing these- by harnessing health data to tackle the backlogs, triaging cases and enhancing workflow efficiency.
In addition, the NHS strategy to not only keep current but deliver excellent clinical care coupled to new innovation and insights, and the explosive growth in digital health care investment (which doubled from Oct 2019 to Oct 2020) mean there is a significant opportunity for better use of health data and generation of new tools and insights.
Why isn’t there more AI in practice?
There are significant complexities of both the potential use and valuation of health data, which must be considered when aiming to support the research needed for AI development and deployment into the NHS.
Health data use for research is rightly tightly regulated, with the legal landscape well documented. We have added to this canon of information to help provide support for the ecosystem in assessing what they need to consider from a legal perspective in engaging in healthcare data deal (ref FCC report).
There is also increasing national guidance from the Government to consider data’s appropriate use (e.g. “A guide to good practice for digital and data-driven health technologies”; “Creating the right framework to realise the benefits for patients and the NHS where data underpins innovation”), and also on how it is valued (including the establishment of The Centre for Improving Data Collaboration within NHSx). This is alongside independent reports such as Imperial “NHS Data: Maximising its impact on the health and wealth of the UK” report and
The Reform Report “Making NHS data work for everyone”
The Reform report articulated the concept of the data value chain- with the creation and realisation of ‘value’ at various points along the journey of data. The data has value for new innovation as it travels along this chain with “value increases as data is transformed into information, knowledge and ultimately action”. This data journey speaks to how AI development is also a journey from early R&D, through internal training, testing, external validation and real-world evaluation. Health data plays critical roles along this, and we would recommend that greater focus is placed on these latter stages of data value creation when considering data deals.
How different stakeholders consider the value of the data can depend not only on this stage but their given perspectives- as the report notes, the value of this data is “always lie in the eye of the beholder.”
The national picture
We’ve seen the Code of Conduct emerging, evolving and shifting landscape with the pandemic COPI notices, the implications of Brexit and new Trade deals all providing new aspects to consider. This is the backdrop against which NCIMI has been aiming to broker data deals for NHS, academic, and industry collaboration, considering all the stakeholder needs and priorities.
To support us with this work, we partnered with FCC to help us develop a framework for delivering data deals. We undertook a range of stakeholder interviews across the different types of partners and perspectives including patient, NHS, academic, legal, financial and commercial considerations.
Our work considered and explored some of the factors informing valuation and structured it into real-world deal negotiation. In our stakeholder interviews, we considered both data value attributes and deal value attributes.
The difference between data value and deal value
We used real-world examples of retrospective and prospective studies in our portfolio and considering data acquired as ‘research data’ versus real-world clinical service data. This conversation identified a range of data and deal attributes that could inform how health data might be valued.
We also explored the different kinds of value that were perceived as important, above and beyond any specific commercial financial return.
Our discussions highlighted the view from all stakeholders to assess deals across a range of what we described as value levers to find the right ‘balance’.
We used the analogy of a sound mixing desk where different values can be dialled up or down depending on benefits to: patients, the NHS as a whole, financial benefit, academic, economic reputation, knowledge exchange and so on.
Of course, different stakeholders dial-up and down the different levers both within and between deals. To ensure value is realised for all, we believe that bringing all stakeholders to the negotiating table is critical.
Our role at NCIMI is to support and broker the partnerships to deliver impact and balance the needs of patients, NHS, academic and industry stakeholders will deliver the real value from data.
Meet the team: Dr Mackenzie Graham, NCIMI’s Senior Research Fellow: Imaging & Data Ethics
June 15, 2021
Mackenzie is a Senior Researcher with the Wellcome Centre for Ethics and Humanities, and the Junior Research Fellow in the Humanities at…
PathLAKE webinar: Research Ethics Committee Review of AI projects
June 2, 2021
NCIMI backed study confirms longer-term lung damage after COVID-19
May 25, 2021
Brainomix’s e-Stroke software goes live across Sussex
May 4, 2021