Exploring Public Perceptions on Health Data Sharing for AI Research in the UK
Artificial intelligence (AI) poses a transformative potential in healthcare by enhancing diagnosis, patient care, and system efficiency. However, the development, evaluation, and monitoring of AI systems post-implementation require substantial healthcare data. Public trust is imperative to facilitate the sharing of such data, especially when it involves sensitive health information. A recent study aimed to delve into the public’s perceptions regarding the sharing of UK health data for AI research, identifying key factors that influence individuals’ willingness to participate.
Methods and Analysis
The study employed eight 90-minute online focus groups conducted between May and July 2024, involving 41 adult participants from the UK. These individuals were recruited through a national register and departmental social media platforms, ensuring a diverse mix in terms of age, ethnicity, household income, education, health status, and geographic location. The researchers utilized thematic analysis to iteratively and inductively develop themes and subthemes from the discussions.
Key Findings
The study revealed three pivotal themes:
1. General Risks of Health Data Sharing
Participants expressed concerns about the risks associated with health data sharing. These included the potential limitations of anonymization, especially concerning rare diseases or large linked datasets, and the sensitivity and scope of data being requested. Issues of data governance, security, and trust in various data custodians also emerged as significant concerns. Anonymization was viewed as essential yet fallible, highlighting a pragmatic view on data sharing with commercial entities.
2. Individual Risk-Benefit Assessment
Participants assessed personal risks versus potential benefits when considering data sharing. Concerns about discrimination, data misuse, and risks to children were weighed against perceived benefits such as altruism, improved care, and the clinical value of AI advancements.
3. Importance of Informed Consent
Informed consent emerged as a crucial element for establishing trust. Participants expressed a preference for clear, study-specific, and tailored information about data usage and AI objectives. They emphasized the importance of consent processes that offer choice, avoid emotional pressure, and allow time for reflection or withdrawal if necessary.
Conclusions and Implications
The study underscores that confidence in sharing health data for AI hinges on participant-specific and study-specific risk-benefit assessments. The findings illuminate public expectations for transparent governance, clear justification for data use, and demonstrable public benefits, especially where commercial involvement is concerned. As the UK and European Union expand their healthcare digital infrastructures and develop regulatory frameworks for AI governance, understanding these expectations will be crucial. Building and maintaining the social license necessary for large-scale data usage will support equitable participation and representation in AI research.
For further details, the full study can be accessed Here.
“`
