Recently established digital transformation unit, NHSX, has published a major new report - Artificial Intelligence: How to get it right. This dovetails with a £250 million investment in a new NHS AI lab to be run in collaboration between the Accelerated Access Collaborative and NHSX.
The report explores the roll-out of AI technology across the spectrum of healthcare, in:
- knowledge generation (drug discovery, pattern recognition, etc)
- public health (screening programmes and epidemiology)
- system efficiency
- precision medicine (also called P4 - predictive, preventive and personalised and participatory)
Organisations like Genomics England, with its resource of over 100,000 genomes and over 2.5 billion clinical data points offer an unparalleled opportunity to make advances in cancer diagnosis and treatment, and rare disease analysis. Currently, diagnosis and screening are the major areas of application for AI, with over 130 products targeting over 70 different conditions under development. Achieving the benefits of AI throughout all potential areas of application is not guaranteed. The report identifies the major challenges and the work being done to address them.
An evidence-based approach
Importantly, the report focuses on real-world evaluation and evidence collection, as the best way to promote acceptance among both clinicians and patients.
Demonstrating the effectiveness of AI-based tools will be an important part of achieving widespread trust and adoption. Collaborative effort between NHSX, Public Health England, MedCity and health technology assessor NICE has produced an evidence standards framework for digital health technologies. This work will be taken forward by NICE with a pilot evaluation programme.
The report highlights a confusing jungle of different regulators that have some involvement in the development of a healthtech innovation from concept to clinic. It is often not clear which body has oversight of which area of regulation – and indeed, some areas are not overseen at all (the quality of data used to train algorithms, for example). It is unsurprising that developers find the pathway difficult to navigate. The new NHS AI Lab will take on the task of helping to forge a clear pathway for innovators.
Many developers lack awareness of the NHS Code of Conduct for Data-Driven Health and Care Technology and the regulatory approval pathways they may need to follow. For example, half of developers surveyed in a State of the Nation analysis had no intention to obtain certification of their technology as a medical device when this is often likely to be required.
Internationally, development is outpacing the ethical and regulatory framework. A survey of members of the Global Digital Health Partnership highlights the way policy and regulation is trailing behind the development of AI applications in health and care. Regulators are working to catch up, but there is much more to do. Development of law and regulation in a piecemeal way, country by country, is likely to hold back development. The report highlights the efforts to achieve international coordination, such as Artificial Intelligence for Health (FG-AI4H) – a focus group set up by the World Health Organisation and the International Telecommunications Union, to work towards a standardised assessment framework for the evaluation of AI-based digital health technologies.
Embracing the potential of AI to build and strengthen the NHS’s healthtech offering will be an important next phase for patient care, but there are risks and challenges to be overcome. Addressing these head-on is welcome. The future of the NHS AI Lab will depend on political developments in the UK, but a continuation of the current policy direction promises real opportunities for both healthcare providers and technology developers.