This guidance is intended to provide further information for applicants to the Artificial Intelligence in Health and Care Award on choosing the NHS or social care sites where they will test, evaluate, and implement their technologies.
Aspects to consider when selecting sites
- Applicants should consider whether sites selected demonstrate a) appropriate geographical spread across the devolved nations, b) appropriate spread between urban and rural populations, c) appropriate ethnic diversity in populations, and d) appropriate spread across geographies with high health inequalities.
- The AI Award site selection process should support the NHS Long Term Plan commitment to continue to target a higher share of funding towards geographies with high health inequalities.
- Applicants should consider whether the choice of site exacerbates health inequalities or whether NICE will review the research evidence produced from the populations selected favourably, as detailed in Chapter 6 of Developing NICE Guidelines: the manual.
- There is an expectation that chosen sites would involve local patients and public/communities across the project.
- The application should include a site where the AI solution has not been deployed before, to test the external validity of the AI solution.
Useful resources
- The health characteristics of the population within geographical areas covered by the target deployment sites can be investigated using the resources available at Public Health England Public Health Profiles website.
- A useful framework for developers is the PICO framework, outlined in Chapter 4 of Developing NICE Guidelines: the manual.
- For additional support with site identification, contact your Local Clinical Research Network.
Site engagement
Required engagement from sites at Stage 2
Discussions with sites should be initiated as soon as possible. At Stage 2, there is a minimum level of engagement that would be looked upon favourably, although the AI Award team will take a holistic view of the application if not all of the required engagement has been evidenced.
- Named exec-level site sponsor and clinical sponsor of AI Solution for each site.
- Confirmation letter/email from exec-level site sponsor/clinical sponsor of AI Solution
- Clear plan on how site(s) will involve and engage patients, the public and practitioners (health and care professionals) at all stages of the work and how this will be evidenced and measured.
- Confirmation of support from clinical and managerial team involved in the project deployment, and awareness of the internal resource requirements necessary for the evaluation.
- Identification of the counter-factual that will be used in the site as part of the project.
- Confirmation that site(s) are aware of any data sharing agreements that would be needed.
- Confirmation that relevant commissioning organisation is aware of product
- If a diagnostic/screening product, an ‘in principle’ approval or full approval from relevant PHE Research Advisory Committee.
- Confirmation whether site is in special measures
Expectations of sites
Organisations should be made aware of their commitments as sites for the AI Award.
- Site(s) will involve and engage patients, the public and practitioners (health and care professionals) at all stages of its work, and this will be evidenced and measured.
- Site(s) will agree appropriate data sharing agreements with companies.
- Site(s) are aware that the AI Award granted to a technology company funds the costs of real-world evaluation borne directly by the technology company or by its partners in line with the application’s objectives and work plan. Costs for the AI solution should be covered by the award funding. Additional costs (i.e. additional server space) will also be funded by NHSX, but these costs must be incurred by the supplier on behalf of the trust and will require invoicing. Staff time should be funded via the award if additional time is required to run the project (e.g. training time).
- Given the above, it is not expected that any collaborating trusts will incur unfunded cost pressures as a result of the collaboration, or will be entered into restrictive commercial agreements with the technology partner covering the post-award period.
- Site(s) aware of the resource required to be part of the evaluation
The NHS AI Virtual Hub by NHSX is a platform to allow stakeholders to directly connect and can be used to engage with colleagues across the NHS.