Development of AI/ML REB review tools at BC Cancer: A collaborative multi-disciplinary approach

DESCRIPTION OF SESSION

As Artificial Intelligence (AI) and Machine Learning (ML) techniques become rapidly prominent in the research landscape, Research Ethics Boards (REBs) nationally are reshaping their review processes and considerations to address the growing breadth of ethical considerations in applying AI/ML to human data. In a rapidly evolving research landscape, AI/ML techniques are currently being integrated into research and applied to human data at an unprecedented rate. The REB must remain empowered to target critical concerns in order to ensure a comprehensive understanding and identification of the unique ethical implications that AI/ML can pose to our patient population and healthcare system. BC Cancer's REB has adopted a proactive strategy to address the recent rise in the number of AI/ML-based research ethics applications. We will discuss the development of BC Cancer's AI/ML review tools used by both researchers and REB members which streamline and standardize the process of evaluating AI/ML-based research. We will identify and explore the challenges and strategies involved in managing these reviews, including re-evaluating the REBs role in AI/ML reviews and the review model adapted to collaborate with AI/ML experts in overseeing the review of these research applications. This session will highlight how an adaptable and informed REB can navigate an evolving research landscape to address ethical oversight with the use of collaborative AI/ML review tools. Please come prepared with thoughts or questions regarding AI/ML-based research ethics during the collaborative portion of this session.

‘We’re not making it as easy as we might’: issues and opportunities Open Science mandates present to Canadian ethics officers and librarians supporting human participatory research

DESCRIPTION OF SESSION

Background: Open Science (OS) offers a grand challenge by demanding both cultural and structural change across the ecosystem to realize global benefit and research accountability. The regulatory function of ethics review for human participatory research (HPR) ensures protection of the participant throughout the entire research project. With OS mandates, the review process and its stakeholders are located at the centre of this challenge: upholding the values of OS while also defending participant autonomy and privacy, community stewardship and situational policy obligations, as well as evaluating and managing other risks. Data management plans (DMPs) reflect all these aspects in a project. Purpose: To explore the experiences of supporting research data management (RDM) and/or DMPs as part of ethics review process by ethics officers, chairs and librarians and/or staff. Methodology: Eight semi-structured, individual interviews were conducted and audio recorded by the author. Transcripts were analyzed using a critical realist approach to identify patterns of support considerations. Analysis/Findings: An inter-related grouping of themes arose from participant reflections in their organizations: time, capacity, personal v. system-constructed relationships. Time, and its consumption, was the determining factor in both providing access to support and demonstrating its value. Definitions of capacity extend beyond infrastructure limitations and availability to require a collective knowledge and awareness across all organizational levels. Relationships are the ‘engine’ of capacity, and in the absence of system-identified networks, the efficacy of support is dependent on the provider’s agency, leadership mobilization, and ‘known quantity’ within the organization. These themes, informed by the participant’s function and positionality within their organization, together contributed to an overall thematic perception of paralysis v. engagement with respect to data management plan adoption and utility, and an ‘us against the system’ dynamic. Data management topics of data sharing, deposit and sovereignty were issues of specific concern. Conclusions: To better meet the complex, interdisciplinary challenges of OS-related data management-ethics concerns, a movement from solo or isolated units of support, defined by process or discipline, to new matrixed system structures which facilitate role awareness, knowledge exchange, and common language should be encouraged.