6

Reflecting on the Workshop and Looking to the Future

In the final session of the workshop, panelists who are potential funders of bioethics research both nationally and internationally reflected on and reacted to the workshop presentations and discussions. Panelists included Tania Simoncelli, the director of policy for science at the Chan Zuckerberg Initiative; Dan O’Connor, the head of humanities and social science at Wellcome Trust; and David Castle, an executive-in-residence at Genome Canada and a professor of public administration at the University of Victoria in British Columbia. The session was moderated by Jeffrey Kahn of Johns Hopkins University, who then called on participants to share their final observations. The rapporteurs’ summary of the research topics and areas for further attention that were suggested by individual participants throughout the workshop is included in Box 6-1.

EVOLVING TECHNOLOGIES, RESEARCH MODELS, AND COLLABORATIONS

Ethically Leveraging Digital Technology and Machine Learning for Health

From the information presented, it is clear that the use of digital health technologies, artificial intelligence (AI), and machine learning in biomedical research and clinical care are placing enormous pressure on current privacy, ethics, and regulatory frameworks, Simoncelli said, and one theme that emerged from the workshop was an urgent call for a new system of data sharing and governance. As discussed by Michelle Mello of Stanford University earlier in the day, the ethical issues can be grouped into two main

categories, existing issues compounded by emerging digital technologies and new issues. The reuse of data for purposes that were not originally intended or anticipated is an old problem, for example, but the reuse of data generated by digital technologies brings fresh concerns. An example of a new problem is what Mello termed “the end of anonymity,” Simoncelli recounted, where the de-identification of data is no longer sufficient to protect patient privacy. New laws and a new system architecture addressing these issues are needed, she said, and developing them will be extremely challenging. If embedding bioethics into these issues from the start is a priority, as Pilar Ossorio of the University of Wisconsin and others suggested, then perhaps people in the field are already behind, Simoncelli said. As noted by John Wilbanks of Sage Bionetworks, these issues are not likely to be a priority for Congress in the near future. In the interim, Simoncelli said, there is an opportunity to begin mapping out an oversight framework, and perhaps one area where funders could play a role would be to help fund the research that would lay the groundwork for such a framework.

New Collaborations Across Disciplines, Institutions, and Stakeholders

Simoncelli noted that another key theme that had surfaced throughout the day is that one challenge to developing new rules, standards, and solutions to address the ethics issues surrounding the use of digital technologies, AI, and machine learning in health is that none of the stakeholders involved has the full complement of expertise required. For example, she said, regulators are just starting to gain experience with these new technologies that continue to evolve at a rapid pace, and digital technology developers are not trained in ethics. As pointed out earlier in the day by Mello and Ossorio, simply broadening the circle of trust to include digital health technology developers is not the solution because these companies operate under a very different set of norms from those embraced by the clinical research enterprise. This diversity of stakeholder expertise presents an important opportunity for new kinds of collaborations that are multidisciplinary and multi-institutional, Simoncelli said. Although one approach is for stakeholders to bring missing expertise on board (e.g., a technology developer could hire a bioethicist), the discussions advocated for the development of new collaborative spaces where people can come together and “learn each other’s language.” As an example, Simoncelli said that the Chan Zuckerberg Initiative is a new kind of philanthropy that seeks to address complex challenges by bringing together scientists, engineers, data scientists, policy experts, and advocates.

Incorporating New Models of Research

Another challenge identified for existing regulatory and ethical frameworks is the emergence of alternative approaches to data collection, including citizen science and personal science. As discussed earlier by Eric Hekler of the University of California, San Diego, there is a wide spectrum of ways in which individuals are taking a scientific approach to answering their own questions about their own health and well-being (see Chapter 3). There are many interesting models of community or participant-driven and patient-driven research projects, Simoncelli said, and members of the public are undertaking these efforts because of gaps—both real and perceived—in research that have left the needs of the community unmet. She observed that communities are demanding a more inclusive approach to research and more accountability from the scientific research community. This is another area of great opportunity, Simoncelli said. Patient-driven or patient-partnered research initiatives can benefit research by providing patient insights into their disease areas. Public engagement in science broadly can lead to more support for science. Co-development with communities can lead to new knowledge that is more relevant to the interests and needs of the communities.

At the same time, patients and patient communities are taking on more and more of the burden as they broaden their roles and responsibilities in the research enterprise, Simoncelli said. A question that needs to be asked is whether that is a burden they should have to take on and, if so, how the professional research enterprise might support their efforts. Hiratsuka discussed the role of community leaders in defining the research questions to be asked, the principles of respect and reciprocity that must be adhered to when asking research questions, and what benefit and harm mean in the context of their community. This role of the community is important, but, as Hekler said, it can be somewhat uncomfortable for the professional research community to adapt to the new role of being present but not engaging as they are used to, Simoncelli said. Although there are many new models of citizen and personal research, they are not well studied. Further study is needed to understand which of these new models are most effective and how the research enterprise can best collaborate using them.

VISION AND SCOPE OF BIOETHICS

Making the Business Case for Bioethics

Underlying many of the issues discussed throughout the workshop is the idea that the National Institutes of Health (NIH) needs to spend more money on bioethics, O’Connor said. There are only so many ways

to redistribute funds, he said, and more money needs to be dedicated to bioethics. From his perspective as a funder, he said, presenting a compelling vision of what the world will look like—how it will be improved—after the money is spent can influence a funder to support a grant, even when resources are tight.

In this regard O’Connor suggested that NIH needs to develop a vision for how bioethics fits into the future of the biomedical research enterprise. The bioethics community and its collaborators in other fields need to come together and craft a compelling vision of why the world would be better if more money were spent on bioethics. For example, as speakers raised during the workshop, the science will be better if the public is involved, if it is more diverse, and if it addresses and attempts to overcome historical injustices. O’Connor said that bioethics needs to acknowledge the issue of “redistribution of wealth” (i.e., equitable benefit from research). He recalled the discussion by Roberts of gene editing as an example of how those conducting the research have little stake in structural change and often benefit from preserving the status quo. Gene editing, he said, would improve their lives from good to better. There is little attention paid to equitable access and addressing structural injustices. Similarly, as discussed, researchers in for-profit technology companies that develop AI and machine learning for health operate under a different set of norms than those in biomedical research.

Bioethics as a Field of Disciplines

“Bioethics is not a discipline,” O’Connor said, “it is a field of many disciplines.” Discussions of bioethics are often limited to clinical bioethics. O’Connor observed that “bioethics,” as a term, was used broadly throughout the workshop. Disciplines in the field of bioethics often include law, philosophy, and clinical research. Based on the scope of the workshop discussions, he suggested that the disciplines of sociology, the history of medicine, critical medical humanities, science and technology studies, and literary theory are also necessary.

Bioethics as a field includes disciplines that NIH has classically not funded, and O’Connor highlighted the need to look at the structures of grant making and ensure that the decision makers are diverse and have the disciplinary backgrounds needed to meet the interdisciplinary challenges of bioethics. Panelists discussed diversity as it relates to bioethics research and the training of the bioethics workforce. In the past, he said, the vision put forth by the bioethics community was essentially that all research would be conducted ethically. This vision is insufficient to encompass the expansive scope of what is meant by “bioethics” as used throughout the workshop. O’Connor highlighted the need for a broader strategic vision for what bioethics will be. This vision should define the standards for what is ethical

and make judgments accordingly, and it should demonstrate that bioethics can be both the study of and the enabler of science, he said.

Castle expanded the discussion of the disciplinary status of bioethics, sharing his perspective on how considering bioethics to be a discipline can be a problem. To illustrate, he described the Genomics and its Ethical, Environmental, Economic, Legal and Social Aspects (GE³LS) program of Genome Canada.¹ The program is similar to the ethical, legal, and social implications program in the United States and to the ethical, legal, and social aspects programs in Europe. The scope of GE³LS includes biomedical, environmental, agricultural, social sciences and humanities, and legal scholarship. Castle said that social sciences and humanities research tended to be part of the large-scale applied research programs and that it is mandated by Genome Canada that there be an integrated social sciences and humanities research component built in. One challenge for these projects, he said, has been to form interdisciplinary teams in which there is genuine interaction between the natural and social sciences. Another challenge is to sustain interdisciplinary research interactions and to be able to distinguish the research outputs from having large-scale, integrated programs that deliver socioeconomic benefits.

A benefit of Genome Canada’s integrative approach, Castle said, is that GE³LS provides a new funding stream for social sciences and humanities researchers for both conceptual and applied research. In some cases, he said, they can see the impact of their work in the real world. A disadvantage of this approach, which Castle suggested is related to the disciplinary status of bioethics, is that it can be difficult to recruit researchers to the projects. One reason is that they might not believe that they have the kind of disciplinary or interdisciplinary focus that is needed, a perspective that may be shaped and entrenched by responding to traditional funding sources. Another reason is that social scientists are used to working with other social scientists and humanities researchers, and the idea of conducting integrated, interdisciplinary work with natural scientists or clinical scientists is challenging, especially for early-career researchers who are trying to establish themselves within the conventions of their particular disciplines.

DISCUSSION

Taking on the Burdens of Self Science

There is a collective action problem among citizens and patients, suggested Maria Merritt of Johns Hopkins University. She recalled the debate about whether individuals should be accepting free services from a digi-

___________________

¹ For more information, see https://www.genomecanada.ca/en/programs/genomics-society-ge3ls (accessed April 15, 2020).

tal platform in exchange for access to their data and the point made by Simoncelli that patients are bearing more and more of the burden that really should not be theirs to bear. There is not an organized way in which individuals can protect their own interests or articulate them collectively, Merritt said. With respect to the concerns raised about not being able to regulate these emerging technologies within the existing system, she pointed out that the European Union has a very different regulatory regime from the United States, and she suggested it might be worth comparing the approaches to governance.

Castle said that this is a key action area for intellectual inquiry and for funding. There is enough evidence available on how big data is being handled and on how corporate interests retain the autonomy to self-regulate to take action. He suggested that the time to reform or regulate these issues has passed and that what is needed is a fundamental re-orientation to a changed world. He acknowledged the concerns about the sale and reuse of data but added that “a fortress mentality” around health data can actually stifle innovation. As an example, he said that where he lives in Canada, the previous privacy commissioner publicly admonished the Ministry of Health twice, claiming that they were keeping useful data out of the hands of innovators and preventing possible research and development partnerships that could help the population. Issuing this type of warning is very unusual for them, Castle said, because the unelected officials in the ministry were overly risk averse.

Simoncelli elaborated on the increasing burden on patient communities. There are many contexts and ways in which patients are seeking access to their data, she said. For example, although patients have the right to their health records under the Health Insurance Portability and Accountability Act, one’s data are not necessarily easy to obtain. Another area where patients are taking on an increasing burden is rare disease research. More and more, rare disease communities are building the research-enabling infrastructure needed to attract researchers and to accelerate research on rare diseases that have been neglected by traditional academic research. She said that there is no real infrastructure to support this patient-driven model of research; groups are organizing themselves and creating what they need from the ground up. In that regard, she suggested, this approach is part of a broader patient movement focused on accelerating research on the terms of patients, and this movement could bring current bioethical issues to the forefront.

Bioethics for the Benefit of Society

Discussions about broadening the conception of what bioethics is and developing a strategic vision for bioethics are important, a workshop par-

ticipant said. That vision, she said, is that bioethics is necessary to make society better. She said that she formerly worked at NIH in a funding role, and she added that there was pressure from the leadership of the scientific programs that any funding for bioethics had to be in the context of how bioethics can make scientific research better. As a result, she said, it was difficult to make programmatic decisions to support bioethics work in a more comprehensive way. The justification for work in bioethics is not that it is an adjunct to doing good research. The justification for bioethics is that making society better involves using all of the tools available to improve people’s lives, including structural and biomedical tools as well as bioethics and other interdisciplinary tools, she said.

Castle agreed and said that Genome Canada has a fundamental social mission of doing the scientific research and technology development to improve lives and increase prosperity. The challenge, he said, is to create programs that deliver on that mission. Castle said that the organization is updating review criteria and guidance to include examples of what meaningful integration of natural and social sciences looks like. It is not enough to simply add on a bioethicist as an afterthought and call it “interdisciplinary,” he said. In addition, the Genome Canada program structure permits the funding of projects that stem from a social sciences and humanities viewpoint. As an example, he mentioned a funded project for which the central question involved using genomics tools to enable remediation in forestry. Development of the project started with a consultation process, led by indigenous communities, to set the priorities from their perspective about what was happening on their lands. The science was then structured around those priorities, resulting in an integrated project with a clear deliverable back to the affected communities, he said.

At Wellcome Trust, O’Connor said, the funding strategy within the humanities and social sciences and the disciplines that make up the broad field of bioethics is to determine what the organization excels at that can add value for finding solutions. For example, Wellcome Trust funds literary studies, which O’Connor said can add to the understanding of how people and communities experience illness. The skills and expertise found in the humanities and the social sciences can help to elucidate the lived experience and form questions in a way that science does not. O’Connor said that the disciplines within the field of bioethics need to confidently offer to contribute their skills and expertise to solutions.

Bioethicists’ Role in Rapid Response

Kahn said that one of the ways in which bioethicists are valuable is that they are available to help address emerging ethical issues as they arise

(e.g., issues associated with the novel coronavirus outbreak²). One barrier to providing a rapid response can be the need to go through investigator-initiated research proposals for support, and Kahn suggested that there is an opportunity to discuss this further. O’Connor said that rapid reaction depends on the availability of people who have the depth of training and broad expertise needed. The challenge for funders is to justify supporting an activity that is not needed all the time, even though it is of high value at those times when it is needed. This is particularly difficult for a public funder, such as NIH, that is accountable to the government and the taxpayers, O’Connor said. As a private foundation, Wellcome Trust has a moral duty to fill the gap and fund these types of activities that its public colleagues cannot. However, Wellcome Trust does not have the scale and scope that NIH does. He suggested that there is an opportunity for NIH to make some funding available to enable the rapid availability of bioethics capacity, and he reiterated that this requires making the case confidently for why this is of value. Wilfond noted that the National Center for Advancing Translational Sciences’ Clinical and Translational Science Awards Program originally required institutions to have an ethics component as part of their awards. Although this is no longer a requirement, he said that an element that has been sustained consists of the institutional research ethics consultation services, which he said offer support to investigators who do not have embedded ethicists in their programs.

Redefining What Information Can Reasonably Be Expected to Be Private

The definition of private information as it pertains to human subjects should be considered, said a workshop participant. Per the Common Rule, “private information includes information about behavior that occurs in a context in which an individual can reasonably expect that no observation or recording is taking place.”³ When that definition was originally written, he said, the sense was that people did not have a reasonable expectation that information in publicly available sources was private (e.g., listings in the phone book, court records) or that what they might do out in public would remain private (e.g., researchers observing people in a public space). Today, given the volume of information available on the Internet and the ways in which those data can be combined to develop a picture of a person’s life,

___________________

² At the time of the workshop, the U.S. Centers for Disease Control and Prevention had just confirmed the possibility of community spread of COVID-19 in the United States. See https://www.cdc.gov/media/releases/2020/s0226-Covid-19-spread.html (accessed April 20, 2020). On March 11, 2020, the World Health Organization declared COVID-19 a pandemic. See https://www.who.int/dg/speeches/detail/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19---11-march-2020 (accessed April 20, 2020).

³ 45 CFR 46.102(e)(4).

the workshop participant asked whether anyone could have the reasonable expectation that any information about their lives or health is private. He suggested there is a need to revisit this definition and what it was originally intended to protect and to consider how it might be refined to protect that interest in the current technological environment. Castle agreed that further discussion is needed on what would be a reasonable expectation of privacy in light of current norms. He suggested that, in addition to legal and ethical analyses, there should also be empirical analysis to understand, for example, the contextualization of data and how thresholds of expectations change as the technological landscape changes. O’Connor added that moral norms and notions of privacy change over time. Some of the disciplines in the field of bioethics, such as anthropology, sociology, and science and technology studies, can help to define current experiences and expectations of privacy, which can inform the discussions of the definition of private information.

Workforce Training Considerations

Castle also highlighted the need for training those who work in data-intensive sciences and the associated support staff on their roles in safeguarding data privacy. A workshop participant agreed that the researchers who are actually using these data (and who do not interact with the patients contributing the data) need training in the ethical management of the data (e.g., data use, combining data). Castle said that there is a wide spectrum of issues that are being incorporated into curricula to help data scientists and those who work with them to be better stewards and users of data. Ethical issues that researchers might face include, for example, structural bias in the data; unintended consequences of algorithm development use; the potential for dual use of algorithms, applying data from one domain to another domain it was not intended for (e.g., applying AI to data for law enforcement purposes); or using data from social media for behavioral modeling that is then used to fundamentally change people’s choice architectures.

CLOSING REMARKS

A main theme of the workshop was disruptive shifts in relationships along a number of axes, Kahn said—shifts in the relationship that individuals have with data that are from or about them and with the professionals who use their data; shifts in the notion of how communities and citizens relate to science, and how science relates to communities and citizens; shifts in how researchers ought to be thinking about inequality and research participation and the implications of research; and shifts in the relationship research has with the workforce pipeline in bioethics. All of these disruptive shifts are making the world hopefully better, he concluded, but certainly different.

This page intentionally left blank.