Ensuring the benefits of AI in healthcare for all: Designing a Sustainable Platform for Public and Professional Stakeholder Engagement

Lead Research Organisation: University of Oxford
Department Name: Population Health

Abstract

Japan and the UK are both investing heavily in national programmes to accelerate the implementation of AI in healthcare delivery (1,2,3). The perceived benefits of this technology include increased efficiacy and cutting costs by using software and algorithms to 'identify patterns too subtle to be detected by human observation, and to use those patterns to generate accurate insights and inform better decision making (4,5). While the use of AI in healthcare promises to be transformative, its ability to approximate conclusions without direct human input using large datasets of personal information, raises a number of concerns about responsibility, transparency and accountability, and public acceptance (6,7).

The use of AI in clinical settings challenges the principles that underpin healthcare delivery: the patient's trust in a system governed by professional codes of conduct; the belief that the primary concern of healthcare is to safeguard well-being, and the subsequent legal and regulatory safety nets that have evolved (8). Expert reports have identified a number of issues surrounding the use of AI, such as: privacy concerns associated with access to clinical data; how it might influence people's legal right to know how decisions about them are made; and the possibility of discrimination of vulnerable populations due to database and algorithm implicit biases (8,9). In the UK, concerns have already been expressed about whether access to publicly-generated data by AI companies might erode public trust in the National Health Service (NHS), potentially jeopardising wide-spread adoption of AI and undermining the basic tenets of healthcare practice (1).

For these reasons, a number of reports have recommended that patients and wider public, who will be most impacted by these technologies, are central to the future development of AI and involved in the design, implementation and governance of software for healthcare, as part of a co-design process (8), that will improve success rates and enhance patient-centred care. To date, very few studies have focused on public concerns regarding AI implementation, or the best strategies for engaging not just with patients, but healthcare professionals and wider publics. What is needed is an engagement platform to support a sustained dialogue with a broad range of stakeholders, so the implementation of AI is in accordance with changing public concerns and for the benefit of humanity.

This research will be situated in research hospitals in the UK and Japan that are pioneering the use of AI systems, to allow us to identify effective models for sustained dialogue and engagement that can inform policy recommendations for future use of AI in healthcare. We will co-design an innovative programme of research to elicit the views of stakeholders, including patients and the public regarding: 1) the current and anticipated use of AI in treatment, diagnostic decision-making and precision medicine 2) the issues that stakeholders perceive will influence the adoption and implementation of AI in healthcare; 3) the types of engagement mechanisms, safeguards and regulatory controls they would like to see in place; and 4) how to develop a platform for engagement that can address issues of trust, responsibility, accountability and transparency, and influence normative practice. This will provide insights for both countries, leading to better sustained public dialogue, as well as informing global policy-making.

Planned Impact

Our research will have considerable impact as it will identify effective models for sustained dialogue and engagement that will benefit a range of stakeholders by bringing different constituents together to stimulate the creation of an interdisciplinary, multi-sectoral ecosystem.

Innovators and industry - beneficiaries include non-academic researchers, technologists and firms (SMEs and large companies) with an interest in developing, translating and commercialising AI to infer health-relevant information and products and services derived from them. Research findings will have impact by i) informing early stage development of AI technologies by providing insight into the needs and valuation practices of a variety of stakeholders and ii) enabling emergent business models to take into account public views, that will be directly relevant to product design. The mapping of the implementation landscape in the UK and Japan will also provide valuable guidance on the legal and ethical requirements and responsibilities of service providers.

Policy and regulation - beneficiaries will include agencies and regulatory actors responsible for the governance of AI and healthcare systems; lawyers and data protection officers involved in the regulation of personal information online; policy makers promoting digital innovation in healthcare and the potential social and economic benefits of data-driven health research, and politicians and activists involved in public debates about personal health data. Research findings will have an impact by i) providing an evidence-base on the social and organisational implications of current research trajectories, translational pathways and challenges to implementation of AI in health, including the valuation practices that are applicable for different applications of AI, ii) Highlighting significant dimensions of the development and implementation pipeline of AI systems for health, particularly relating to data needed, to inform policy and regulatory decision-making, iii) Informing policy and legislative consideration of the governance of online health-relevant information and its use, iv) Sharing findings and best practice between the countries involved (UK and Japan), highlighting priorities across each country and outlining transnational differences in public preference and legal frameworks, v) Providing an evidence base about the public, legal and ethical underpinnings of a trustworthy and robust governance system for digital sharing of sensitive information.

HCPs and managers - HCPs and health service managers will be represented in expert workshops and data generation phases of the project. Research findings will have an impact by i) ensuring that development of AI systems for healthcare reflect the needs and values of HCPs, ii) encouraging HCP participation in debate about the value and challenges of integrating AI systems to healthcare, iii) informing HCPs about the values at stake when health data is shared with or received from external developers of AI, and the challenges associated with implementation of AI systems. The project will also have impact on HCPs indirectly, through increased public awareness and discussion of the benefits and challenges of utilising data collected outside of healthcare treatment and diagnosis.

Patients, citizens and publics - beneficiaries include patients, charities and patient/citizen organisations. Recruited patients and citizens of both countries will share their perspectives on the use of health-related data in the empirical research. Research findings will impact by i) ensuring that development of AI systems for healthcare reflect the needs and values of patients and citizens variously located in society, ii) informing and encouraging public participation in debate about the value and challenges of integrating AI systems to healthcare and more broadly on permitting data collected outside of healthcare to be used in treatment and diagnosis.

Publications

10 25 50
 
Description AIDE Project Oxford Public and Patient Involvement Panel 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Patients, carers and/or patient groups
Results and Impact A core aspect of the research project is that it is co-designed with patients and members of the public, as well as key industry and academic partners. In 2020, we recruited a panel of expert patients to assist the research team in understanding key issues relating to artificial intelligence in healthcare that would potentially impact patients and the public. This Public and Patient Involvement Panel (PPIP) is made up of 6 members located in the London and Oxfordshire regions. The panel's role in the project is to help shape the research agenda from a patient's and citizen perspective, review study materials, and identify and connect the researchers with stakeholders to recruit to the project's empirical studies. The panel has met with the researchers 4 times in 2020, the first to induct them to the project, and the last three meetings consisted of virtual knowledge building workshops, where academics and industry experts in the filed of AI discussed with the panel members use cases of AI in healthcare, provided examples of how AI may shape healthcare delivery, overview of regulation of AI, and understanding social and ethical issues relating to privacy, agency, and bias and discrimination. Feedback from these workshops was generally positive in terms of the content and knowledge gained, however, the virtual nature of the delivery was sometimes tiring for participants despite scheduling breaks, and sometimes technical difficulties stifled group work, which would have been better in face-to-face meetings. The next steps will be to hold two co-design workshops which will discuss what conditions need to be in place for trustworthy AI in healthcare, key issues the research needs to investigate from a patients/public perspective, and ideas for recruiting participants to the focus groups and interviews study.
Year(s) Of Engagement Activity 2020