We use some essential cookies to make this website work. We’d like to set some additional cookies to understand how you use the website and to improve it. We also use cookies set by other sites to help us deliver content from their services. You can read more about our cookies before you choose.
This is a new website – your feedback will help us to improve it.
4.1 Artificial intelligence (AI) multi-disciplinary teams (MDTs)
Chapter 4: Workforce Transformation
Artificial intelligence (AI) multi-disciplinary teams (MDTs) could be a significant factor in the successful adoption of AI technologies in healthcare.
Interviewees for this research noted that specialised artificial intelligence (AI) multi-disciplinary teams (MDTs), which could exist across or within NHS organisations, are a significant factor in the successful adoption of AI technologies. This aligns with HEE’s broader strategic goal to support the expansion and development of MDTs.17
AI-specific MDTs can comprise of a mix of individuals like specialist clinicians, information technology and governance specialists, clinical domain experts, data scientists, and software engineers. These teams can be further strengthened by data security experts, and regulatory and clinical safety specialists with advanced AI knowledge and skills.
It is critical that AI MDTs are sufficiently diverse and consider inputs and feedback from other experts and patients. Processes to involve stakeholder and the public in the development and implementation of AI technologies, can include participatory assessments of their potential economic, social and environmental impacts conducted as part of algorithmic impact assessments.41
Diversity in teams who develop and deploy AI has been shown to mitigate the risk of bias and unfairness, leading to more ethical and safer AI products and implementations.18
Interviewees highlighted that successful AI MDTs involve sharing of knowledge across disciplines and development of common objectives to address challenges when implementing and monitoring AI technologies. This concept is supported by findings of the Topol Review, which noted that the ‘teams required to adopt technology-enabled change are likely to be non-hierarchical, self-organising, multi-disciplinary teams in which colleagues have equal status and responsibility.’3
Interviewees also noted thatAI MDTs can lead the evaluation, implementation and ongoing monitoring of AI technologies, as well as advise and assist individuals and project teams responsible for AI procurement decisions (primarily within the Driver archetype).
Box 4 provides a list of existing and potential responsibilities for AI MDTs, as suggested by this research’s interviewees, while Box 5 details how these teams can operate in a fictional example.
Box 4: AI MDT responsibilities
The responsibilities of AI MDTs could include the below.
Drive development of novel AI solutions or co-create AI technologies with industry innovators.
Consider the needs and views of stakeholders and patients across diverse backgrounds, and seek to involve other experts and members of the public when developing AI technologies.
Co-ordinate engagement with industry innovators and act as a conduit between the 2 sectors.
Facilitate the introduction of AI technologies including technical integration, systems change management and education of users.
Lead the ethical governance and implementation of AI technologies at their settings.
Evaluate potential AI technologies to assess their suitability for local deployment, including the need for local validation, prospective clinical studies, and assessing technical requirements.
Ensure clinical safety, manage the reporting process for possible safety incidents and establish appropriate back up processes.
Undertake ongoing post-deployment monitoring, surveillance and audits of AI technologies to ensure they continue to perform safely and effectively.
Evaluate the impact of AI technologies on clinical pathways, including review process efficiencies and patient outcomes.
Design, deliver and continually update product-specific user education.
AI MDTs will likely be required at different levels within the NHS. For example, project-level MDTs might manage the deployment of specific AI technologies within individual healthcare settings. These teams could be supported by organisational-level AI MDTs acting as a shared resource and source of advice and guidance across multiple sites (such as an Integrated Care System, primary care network or large hospital trust).
The Goldacre Review suggests identifying ‘Data Pioneer’ analytics teams in Integrated Care Systems and trusts with strong existing skills in analytics, informatics, and software engineering. These can be an example of organisational-level AI MDTs, and could provide a framework for other team structures by making their methods visible to the wider community and providing open documentation of their work for learning.4
Interviewees noted that the development of organisational-level MDTs will need to coincide with structural changes in the NHS to allow for sharing of resources and available capacities at each healthcare site.
Box 5: AI MDT - fictional case study
This fictional case study is based on an integrated care system (ICS) team working to co-create an imaging-based AI technology with an industry innovator within a healthcare setting. It complements the fictional case study presented in Box 2.
The aim of this case study is to outline the types of roles and responsibilities of individuals within the AI MDT team, and their relation to the archetypes discussed in this report. It should only be considered as an illustrative example rather than a fixed structure or blueprint as AI MDT size, structure and roles will vary depending on needs, products and local factors.
The AI MDT should represent all teams working for commissioning, developing, and implementing an AI technology, as well as the full range of its users. The team should be diverse in terms of demographics, levels of seniority, and familiarity with digital health to ensure fairness for both staff and patients.
All product and organisation names used in this case study are fictional.
Summary
A chest X-ray triaging AI product (tri-X) is co-designed between the Linchester ICS and an industry innovator, Triadix. The technology is designed to help manage the radiology workflow by prioritising chest X-rays for reporting by radiologists. The project AI MDT comprises key individuals involved with or affected by the development and implementation of tri-X.
ICS Digital Transformation Lead – Driver
Oversee the work of the AI MDT, championing their role and advocating for appropriate funding and resourcing of the team.
Ensure the appropriateness of the AI solution to ICS needs and clinical workflows.
Commission Triadix as a commercial partner to co-create tri-X and manage their contractual relationship, including intellectual property, data sharing agreement and commercial commissioning agreements for the wider roll out of tri-X.
Ensure that appropriate regulatory standards are being adhered to when creating tri-X and the product development aligns with appropriate guidelines.
Radiology Business Manager – Driver
Consider impact of tri-X on finance, human resources, training and sustainability.
Help to define business case and outcome measures that will address real world system benefits to the ICS radiology service.
Conduct service level risk assessment, and oversee service continuity and decommissioning planning.
Define clinical scope of the product and advise on workflow integration plan at the design stage, including representation of a diverse range of users.
Conduct clinical risk assessment and develop mitigation plan.
Review validation performance and prospective clinical studies of tri-X.
Advise on systems change processes involved in the introduction of tri-X.
Support creation and delivery of product-specific user education for tri-X to ICS radiologists alongside Embedder colleagues.
Use tri-X in clinical work.
Clinical Product Manager – Creator
Undertake problem discovery sessions with stakeholders, map out workflows and explore the clinical issue to be solved by tri-X.
Define the scope of use and requirements for tri-X.
Liaise with clinical experts including ICS radiologists to ensure that the product is clinically robust and is integrated strategically into clinical workflows.
Co-create tri-X with users. Conduct regular user research sessions to get feedback on how to improve the product and add value to users.
Ensure tri-X aligns with safety and regulatory standards, working closely with Linchester ICS safety and regulatory team from the outset.
Coordinate with ICS data science teams and clinical specialists to conduct prospective clinical studies of tri-X, in line with Linchester ICS requirements and national evidence for efficacy.
Lead Project Data Scientist – Creator/Embedder
Oversee the ICS data and clinical science team working on technical aspects of creation, validation and embedding of tri-X, including:
analysis of the data set available to address the problem, evaluating the data for representativeness, potential bias and generalisability and ensuring it is of appropriate quality
working alongside Triadix data scientists and software engineers to create, test and improve the tri-X AI model, iterating until accepted accuracy and safety levels are reached using internal test data
overseeing validation of the final algorithm using internal and external validation data sets, as well as conducting prospective clinical evaluation studies in collaboration with users and Triadix team
assessing the regulatory compliance of tri-X
conducting post-market follow up to assess and monitor performance, and act upon any model drift
creating and delivering product-specific user education for tri-X alongside clinical MDT colleagues
Clinical Safety Officer – Embedder
Conduct a clinical safety review and risk assessment of tri-X.
Ensure there are appropriate protocols in place for systems failure, fall-back workflows, reporting errors and on-going monitoring.
Consider mitigation plan for potential risks of deskilling, if relevant.
Information Technology (IT) Officer – Embedder
Lead the technical implementation of tri-X including systems integration, data and cyber security, data storage, system robustness and availability.
Provide ongoing advice and support for users of tri-X.
Information Governance (IG) Officer – Embedder
Lead integration with existing information governance processes or development of new processes as needed for the tri-X project, providing guidance from early in the project ideation stage through to deployment.
18 J Zou, L Schiebinger. Ensuring that biomedical AI benefits diverse populations. EBioMedicine. 67, 103358. 2021. https://doi.org/10.1016/j.ebiom.2021.103358 Accessed May 24, 2022.
3 Topol E. The Topol Review: Preparing the Healthcare Workforce to Deliver the Digital Future. 2019. https://topol.hee.nhs.uk/the-topol-review/ Accessed February 28, 2022.
4 Goldacre B, Morley J. Better, Broader, Safer: Using health data for research and analysis. A review commissioned by the Secretary of State for Health and Social Care. Department of Health and Social Care. 2022. https://www.goldacrereview.org/ Accessed May 24, 2022.
Page last reviewed: 20 April 2023
Next review due: 20 April 2024