The National Institutes of Health’s Office of Data Science Strategy (ODSS) sought to build a multi-disciplinary community of stakeholders interested in the social implications of technology to collaboratively envision the integration of artificial intelligence (AI) and ethics in biomedicine to advance the NIH mission. The overarching goal was to bring together a diverse cross-section of scientists, social scientists, ethicists, patient advocates, legal scholars, communicators, and artists to identify important areas of consideration and problem-solving strategies at the intersection of AI, machine learning (ML), biomedical and behavioral sciences, and ethics. By forging new collaborations among these cross-disciplinary groups, the NIH sought to identify the benefits, risks, and future directions in biomedical AI that align with the public interest and ensure equitable health benefits for all communities.
ODSS Micro Lab #1: Collaboratively Envisioning AI and Ethics in Biomedical Research, Part 1
During this Micro Lab, ODSS sought to identify the relevant stakeholders and engage in interactive discussions to help shape the conversation about AI/ML ethics in the biomedical and behavioral sciences at the NIH. View Part 1 and Part 2 of the Micro Lab webinar recording.
ODSS Micro Lab #2: Collaboratively Envisioning AI and Ethics in Biomedical Research, Part 2
During this Micro Lab, ODSS expanded the community of participants from cross-disciplinary backgrounds to address questions that were exciting, innovative, and far-reaching to imagine the future opportunities and challenges in biomedical AI ethics. View the Micro Lab webinar recording.
The Innovation Lab: A Data Ecosystems Approach to Ethical AI for Biomedical and Behavioral Research
Developing social and technical approaches to defining and implementing ethics across the AI data ecosystem
The National Institutes of Health’s Office of Data Science Strategy (ODSS) sought to build a multi-disciplinary community of stakeholders interested in the social implications of technology that helped bridge the gap between artificial intelligence (AI) and ethics in biomedicine to advance the NIH mission. By forging new collaborations among these cross-disciplinary groups, the NIH identified the benefits, risks, and future directions in biomedical AI that align with the public interest and ensure equitable health benefits for all communities.
The Challenge
The AI data ecosystem consists of many elements from AI models and workflows, data repositories, data, metadata and the myriad ways they are sourced, computational infrastructures, and data generators, as well as many different stakeholders including researchers, data users, data re-users, data donors / human subjects, and model developers. Each element poses its own opportunity for putting ethics into practice; and each participant poses their own normative principles and ability to weigh ethical considerations. Increasingly, the ethical development and application of AI requires an integral understanding across multiple ecosystem elements, and the ability to address and honor the concerns of a multitude of stakeholders.
How do we develop social and technical approaches to defining, measuring, and implementing ethics across the ecosystem? What approaches can ensure robust assessment and control of AI models, data, and methodologies over time and as they are applied in new situations with regards to their ethical use?
A systems approach is needed to take into consideration the interrelated, interdependent elements of the AI data ecosystem and account for the fact that changing any one element may affect others. Attention is needed now to intentionally prepare for the ubiquitous use of AI in biomedical research and establish an ethical AI/data ecosystem.
The Innovation Lab
The Innovation Lab was an intensive, interactive, and free-thinking workshop and aimed to stimulate thinking in promising new research approaches and theories at the intersection of artificial intelligence and machine learning, biomedical and behavioral research, and ethics. Participants engaged constructively in dialogue with each other, the facilitators, and the Subject Guides to develop collaborative research proposals. Collaboration was encouraged by bringing diverse minds together to embrace this challenge.
The Innovation Lab was held online and ran over five days from March 14-18, 2022, concluding after final presentations. The approach of the Innovation Lab was not to discuss ideas that were already well developed but not yet published. Rather, the goal was to bring individuals from different disciplines together to interact and engage in free-thinking on first principles, to learn from one another and create an integrated vision for future research projects. Although the sharing of these ideas was encouraged within the Innovation Lab, we asked that their confidentiality would be respected outside the Innovation Lab. At the Lab, interdisciplinary teams worked together to ideate and develop a roadmap for how to tackle selected challenges in this field. Through the course of the five days, teams formed, pitched, and refined plans (based on input from subject guides and other participants) for interdisciplinary pilot projects that advanced scientific questions related to social and technical approaches to defining and implementing ethics across the ecosystem.
Who Applied
This Innovation Lab brought together people with expertise in behavioral and biomedical research, IRB review, data and technology standards, data repositories, data management, AI model development, computer science, computational tools, social science, ethics, tech law and data sharing policy, patient advocacy, and community engagement interested in forming new collaborations and developing innovative approaches to address ethical opportunities and challenges arising in the AI data ecosystem. The interdisciplinary collaborations formed during these events resulted in new peer-reviewed publications or prepared teams for submitting proposals to NIH funding opportunities to further develop, refine, and test, hypothesize, and develop original research project ideas.
Approximately 30 applicants were selected to participate on the basis of their interests, expertise, and other characteristics solicited in the application. Most participants were academic faculty, industry and legal experts, researchers, and program directors, from early to senior career stages. Original thinkers from outside academia (e.g., industry) or earlier career stages were also encouraged to apply. Consideration was given to balance across a range of diverse disciplinary experience and expertise. Participants were willing to engage in frank disclosure and assessment of ideas in a collegial and professional fashion. Meeting discussions were considered a private communication and not shared outside of the meeting, unless approved by the contributor. NIH staff reviewed applications and selected the final list of participants. All research-related application information has been kept confidential.
Meeting Outcomes
View a summary of the meeting and outcomes.
