3

Ethical Questions Concerning Nontraditional Approaches for Data Collection and Use

___________________

¹ This text was revised after prepublication release.

Over the past decade, many new tools for individuals to monitor their own health have emerged, and the public is increasingly engaging in scientific research activities, serving as advocates for their own health. Citizen science (also sometimes referred to as personal science, do-it-yourself science, patient-led research, or participant-led research) is a form of research that takes place outside of the regulatory environment of traditional scientific research,² said Camille Nebeker, an associate professor in the University of California (UC), San Diego, School of Medicine, and the session moderator. Originally, citizen scientists partnered with professional scientists to support ongoing research by contributing observations (e.g., counting birds). Community health workers are often engaged as citizen scientists, Nebeker noted, partnering with public health researchers. More recently, individuals have begun conducting their own research without any partnership with professional scientists. Participants in this new system of unregulated personal research may have education in another field and a passion to find a cure for a particular condition, but they have not had training in the practice of scientific research

___________________

² Currently there is not a widely accepted definition of the term “citizen science” (Heigl et al., 2019), but it has been referred to as “the general public engagement in scientific research activities when citizens actively contribute to science either with their intellectual effort or surrounding knowledge or with their tools and resources” (EC, 2014).

or research ethics. “They don’t know what they don’t know,” Nebeker said, and there is the potential for harm.

In this session, Eric Hekler, an associate professor in the Department of Family Medicine and Public Health at UC San Diego, provided an overview of citizen science and discussed what the National Institutes of Health (NIH) could do to advance this area. John Wilbanks, the chief commons officer at Sage Bionetworks, discussed the governance of unregulated research using mobile devices and how individuals engaging in self-study might be persuaded to do so ethically.

CITIZEN AND PERSONAL SCIENCE

Citizen science has been defined by the European Commission as “the general public engagement in scientific research activities,” Hekler said (EC, 2014). There are many dimensions to citizen science, including engaging the public in activities related to health research, he said. A subcategory of citizen science is personal science,³ which has been described as individuals taking a scientific approach to answering questions about their own health and well-being (Heyen, 2020). Individuals might conduct personal research to understand, for example, how they are recovering from a procedure or their response to eating certain foods. As individuals engage in personal science to optimize their health, well-being, or any other personal goal, questions arise about power and privilege, Hekler said. Specifically, who has the right to define what “success” is in personal science, and how can the biomedical research enterprise support individuals in their pursuit of success?

Hekler illustrated examples of citizen science activities based on who is leading the research: citizens, professional scientists, or both as co-developers. These citizen science research activities, he said, can be for the benefit of science, historically marginalized populations, and/or communities and individuals themselves. Who leads the research and who benefits from the research can then be considered against the contribution that the research makes to the health system—whether by enabling overall improved health through activities targeting infrastructure and systems, improving prevention efforts, or improving diagnosis and treatment.

There is great diversity and variation within citizen science activities, Hekler said. Activities vary in terms of who is leading, who it benefits,

___________________

³ In Hekler’s view, “personal science” can be considered a subcategory of the broader concept of “citizen science.” Specifically, if an individual is using the scientific method to support their own goals and desires, they are engaging in personal science. Citizen science is broader and can also involve when individuals take part in research work under the guidance or supervision of others (e.g., bird migration projects).

and the contribution the research makes to the health system. Hekler went on to describe four different types of citizen science activities, beginning with Foldit,⁴ a computer game designed by scientists to allow the public to contribute to scientific research on protein folding following an introductory puzzle where players learn what a protein looks like and how to use the game tools to fix common structural issues. Presented with puzzles based on well-understood proteins, the methods in which the puzzles are approached by members of the public are analyzed by researchers in order to improve protein folding software. Propeller Health is another example of a simple digital tool designed by professionals to track a patient’s asthmatic symptoms and potential triggers (e.g., pollution).⁵ An example of co-development between professionals and citizens is MakerNurse, a collaboration that enables nurses to design, make, and share solutions they develop in the course of providing patient care.⁶ Finally, Hekler described OpenAPS, a personal science and patient-led activity that has expanded scientific knowledge and contributed to improved diagnostics and treatment.⁷ OpenAPS is an artificial pancreas system designed by Dana Lewis, an individual with type 1 diabetes who is not trained as an engineer or a scientist. Lewis and Scott Leibrand built the system to meet Lewis’s own needs and then shared the design for the technology online. There are now more than 2,000 people using similar devices they built using Lewis’s open-source specifications. OpenAPS is an example of a personal science activity that has not only helped individual patients but has also contributed to scientific knowledge, spurring peer-reviewed publications and helping to increase the pace of development of artificial pancreas systems, Hekler said.

The Implications of Personal Science

Personal science has implications for professional science and raises questions about what is known and how science is conducted, Hekler said. First, personal science reveals layers of possible biases within professional science. The first layer of possible bias, stereotyping, occurs when individuals are grouped as a research subject or a patient and are essentially told that they should not be taking matters into their own hands and that they should “wait for service” from the scientific research community. The next layer is omission bias. When faced with a potential risk, people tend to favor inaction over action, Hekler said, even when inaction carries the same

___________________

⁴ See https://fold.it/portal (accessed April 15, 2020).

⁵ See https://www.propellerhealth.com (accessed April 15, 2020).

⁶ See http://makernurse.com (accessed April 15, 2020).

⁷ See https://openaps.org (accessed April 15, 2020).

or greater risk. In the clinical research context, essentially asking patients to wait for professional scientific research to provide solutions can result in individual inaction with potential consequences. Furthermore, professionals can be biased toward making sure that solutions are “foolproof” before being released, which could be thought of as a variation of professionals falling into an omission bias. These two variations of the omission bias, occurring both among patients and professionals, could then lead to the last layer—a situation of learned helplessness when people have been told to wait for solutions from the scientific enterprise that never materialize. This can be the case particularly for those with a rare condition that is not a focus of NIH-funded research, Hekler added.

Next, personal science raises questions regarding the process of forming scientific consensus, Hekler said, referring to the work of Boaz Miller, who framed three conditions that can indicate whether there is knowledge-based scientific consensus: calibration (i.e., consensus requires that the parties agree that they want to agree); social diversity; and triangulation of evidence (i.e., having a diversity of methods for generating evidence) (Miller, 2013). The last two, diversity of people and methods, allow for a wide range of knowledge in the process of building trustworthy consensus. If professional science is not hearing from the diversity of voices and methods that personal science brings, it is potentially ignoring valuable information that can promote scientific knowledge, Hekler said.

Finally, personal science asks professional science to “go beyond pure science into ‘ought’ thinking,” Hekler said, drawing from the “is–ought framework” of philosopher David Hume. Professional science sees the world as it is, thinking about science and facts, and focusing on the past and present (which perpetuates systemic biases). A benefit that personal science can bring is that it enables individuals to see and act toward a possible future world or experience as they believe it ought to be. More diverse participation and views in research highlight the need to think further about ethics and morality, along with values and principles, all with an orientation toward the future. Bridging between “is” and “ought” thinking is extremely difficult, Hekler said.

Considerations Moving Forward

Hekler listed four key questions to be addressed in order to start bridging “is” and “ought” thinking and to be considered in future orientation of work and thinking for ethics research moving forward.

• How might professional scientists improve their awareness of systemic implicit biases that, unintentionally, may compromise the capacity to support those served by science?

• How do scientists, unintentionally, compromise equitable participation, contribution, and benefit from the applied sciences? How might scientists improve equity in science?
• How might scientists increase the trustworthiness of scientific consensus, via diversifying the people contributing and diversifying their methods, without compromising their capacity to work toward consensus?
• How do evidence and values interact effectively? How might scientists bridge “is” and “ought” thinking within NIH and beyond?

One way to think about these issues is relative to mechanisms that NIH possesses to potentially strengthen the trustworthiness of scientific consensus and start to bridge “is” and “ought” thinking, Hekler said. Appropriate methods, checklists to define quality, study sections (groups of experts in a given field who review grant applications), and funding opportunity announcements are all mechanisms that work together synergistically. Hekler listed his suggestions for each of the methods and processes.

Methods

Create new and expand on existing NIH-acknowledged best-practice methods in citizen and personal science, keeping emerging open science⁸ practices in mind. Open science enables personal science to build upon the work of professional science. For example, the OpenAPS algorithm was built on a commercial continuous glucose monitor and an insulin pump, Hekler explained.

Quality Checklist

Build the appropriate quality checklists that a study section could use to ensure adherence to best-practice methods in citizen and personal science, while not compromising the fundamental epistemological⁹ requirements in the approaches taken by self-studies. Evidence-based practice is grounded in clinician experience and prior knowledge. That prior knowledge, however, is grounded in epidemiology, which is population based. Variations of epistemological questions should be considered as a way to add value to self-studies, even though epistemology does not produce generalizable knowledge by itself.

___________________

⁸ The term “open science” can be defined as a set of practices that increase the transparency and accessibility of scientific research (van der Zee and Reich, 2018).

⁹ Epistemology involves the philosophical study of how knowledge is acquired and disseminated.

Study Sections

Create new study sections filled with a diversity of citizen and personal scientists along with traditional professionals who, together, support rigor in this type of science. Professionals from sociology and the humanities, who have a long history and tradition of studying and understanding structural biases, should be included.

Funding Opportunity Announcements

Create funding opportunity announcements that legitimize and fund a wider range of contributors (e.g., participant-led work) and methods.

In closing, Hekler reminded participants that article 27 of the Universal Declaration of Human Rights states that “Everyone has the right to participate … [and] share in scientific advancements and its benefits,”¹⁰ which is increasingly becoming a rallying cry from leaders in advancing personal science.

GOVERNANCE OF UNREGULATED HEALTH RESEARCH USING MOBILE DEVICES

Today’s smartphones have far greater capabilities than simply spoken communication, Wilbanks said. A typical iPhone, for instance, has facial recognition technology, a barometer, a three-axis gyroscope, an accelerometer, a proximity sensor, and an ambient light sensor. And individuals today have access to many other advanced technologies beyond the smartphone. On the retail website Alibaba, one can purchase an electroencephalograph, a pulse oximeter, and other medical devices at a relatively low cost. Due to the advantages in the affordability and availability of these types of materials and devices, the research capabilities in personal science are accelerating quickly.

Wilbanks and colleagues set out to examine potential ethical and policy questions related to unregulated health research using mobile devices in the United States. Specifically, the study addressed how independent individuals and entities pursuing self-led research could be influenced to do what is ethically right when they are not legally obligated to do so and might not agree that it is necessary. With funding from NIH,¹¹ Wilbanks

___________________

¹⁰ See https://www.un.org/en/universal-declaration-human-rights (accessed April 15, 2020).

¹¹ Addressing ELSI Issues in Unregulated Health Research Using Mobile Devices, No. 1R01CA20738-01A1, National Cancer Institute, National Human Genome Research Institute, and Office of Science Policy and Office of Behavioral and Social Sciences Research in the Office of the Director, National Institutes of Health, Mark A. Rothstein and John T. Wilbanks, Principal Investigators.

and colleagues conducted a series of qualitative interviews with thought leaders including application (app) and device developers, researchers using mobile devices, patient and research participant advocates, and regulatory and policy professionals. Four working group meetings were also convened to gather input from stakeholders. The findings and recommendations were published in a special issue of The Journal of Law, Medicine & Ethics and were disseminated to app developers and policy makers at two workshops (Rothstein et al., 2020).

Preventing Harms from Unregulated Health Research Using Mobile Devices

A range of options for preventing harms from unregulated health research using mobile devices were offered by the members of the working group. One option considered was to extend the Common Rule to apply to all research and all researchers.¹² The Common Rule was designed to safeguard the welfare and interests of research participants and society. Although the vast majority of other countries use the approach of extending the Common Rule, Wilbanks said there is little political support for taking this approach in the United States. Another option proposed within the working group was to maintain the status quo. The supporting arguments for not taking any action were that there have not been any adverse consequences thus far and that regulation could drive research underground or result in some valuable research not being done. In the end, neither of these options was considered to be practical.

Instead, the research group focused on approaches to persuading people to conduct their self-study in an ethical manner. Steps defined included establishing outer boundaries, providing education and assistance, and appealing to their self-interest and their sense of decency. The approach focuses on communicating to individuals that an ethical approach to science results in better science and that it is therefore in an individual’s own self-interest to engage in personal science in an ethical way.

Potential Opportunities for Various Stakeholders

The study report details the investigators’ recommendations for action by states, NIH, the U.S. Food and Drug Administration (FDA), the Federal Trade Commission (FTC) and the Consumer Product Safety Commission (CPSC), the Centers for Disease Control and Prevention (CDC), consumer

___________________

¹² For information on the Federal Policy for the Protection of Human Subjects, known as the Common Rule, see https://www.hhs.gov/ohrp/regulations-and-policy/regulations/commonrule/index.html (accessed April 15, 2020).

technology companies (including app stores), and individual unregulated researchers (e.g., citizen scientists, do-it-yourself researchers). Wilbanks highlighted several of the top-level recommendations from the study in each area (see Rothstein et al., 2020).

• States can take action in the absence of federal laws and regulations. Wilbanks cited the state research law enacted by Maryland as an example.¹³ This law broadly defines “person performing research” and requires that those persons conform to the Common Rule.
• NIH has the ability and the opportunity to educate and assist unregulated researchers in conducting ethical self-research. One recommendation from the study, Wilbanks said, is for NIH to establish an Office of Unregulated Health Research to provide more accessible information regarding ethical practices for self-research. NIH cannot fund unregulated research studies, but it can fund studies about unregulated research to elucidate the actual risks and benefits.
• FDA only regulates mobile apps that function as medical devices and that could pose a risk to patient safety if they fail to function as intended, Wilbanks said. However, there are ongoing interagency collaborative efforts to reduce regulatory duplication and identify areas unaddressed by current regulation that the study recommended be continued. Wilbanks added that the health app developer and health device developer communities are closely watching the regulatory environment, and he said the threat of FDA regulation is a powerful incentive that could be put to better use.
• FTC has the potential to play a greater role in the area of consumer protection, Wilbanks said. CPSC has influence over the types of products that are marketed and the claims that are made about products. The study recommended that FTC make targeted enforcement actions against developers of unregulated mobile research applications who engage in deceptive or unfair trade practices, Wilbanks said. The study also recommended that FTC increase monitoring of the security of and improve the security practices of the hardware and software components of tools for self-research. He noted that software security vulnerabilities are often propagated by developers who simply copy and paste bad security source code shared by others.
• CDC expertise in epidemiological surveillance could be extended to surveillance for security vulnerabilities related to mobile health research tools. The sharing of flawed code can create vulnerabilities

___________________

¹³ Md. Code Ann., Health-Gen. §§ 13-2001–2002.

• that can promote misinformation and allow for leakage of personal information, Wilbanks said. The study recommended that CDC work with NIH and private entities to understand the scope of the issue and to establish a system to monitor trends in security vulnerabilities over time.

• App stores, such as those maintained by Apple and Google, are powerful de facto regulators of which products reach consumers’ mobile devices and computers, Wilbanks said. For example, Apple requires apps developed with ResearchKit to secure approval by an institutional review board (IRB). The study recommended that Google implement a comparable policy and that both companies require developers to upload the IRB approval letter and make it available to consumers. App stores could also prioritize where apps appear in search results based on adherence to certain norms, require that terms of service and privacy policies for health apps explicitly ban third-party data transfer, and require device developers to encrypt data both at rest and in transit.
• Citizen science organizations reach unregulated researchers through social media, meet-ups, and mailing lists, and Wilbanks said these groups need to be better leveraged to promote ethical self-research. Often, these groups are only engaged when they are being “scolded.” One study recommendation calls for these organizations to create guidance for their members on how to transparently communicate the goals, risks, and benefits of their research.

DISCUSSION

To open the discussion, Nebeker described Quantified Self as another example of a citizen science organization.¹⁴ Quantified Self is a community of people supporting one another in the conduct of self-study. In some cases individuals engaging in self-study are living with a significant health problem that they feel the health system is not addressing sufficiently, Nebeker said. In other cases, individuals might be looking to improve personal well-being or performance (e.g., better sleep patterns, faster marathon time).

Nebeker provided research ethics guidance to a Quantified Self participant-led research project to study personal daily lipid levels. The leaders were interested in thinking through study risks, benefits, and risk-management solutions. Because this was a group of individuals engaging in self-study to gain individual knowledge, the project did not meet the federal definition of human subjects research as it was not considered to be “generalizable knowledge,” which is a classifier for defining research

___________________

¹⁴ See https://quantifiedself.com (accessed April 15, 2020).

subject to the Common Rule or IRB review, she explained. However, the group did publish a discussion of the process of conducting responsible and safe participant-led research (Grant et al., 2019). In this participant-led research, participants studied their daily lipid levels over time, and each of these single-subject studies led to observations that the leaders wanted to share via a peer-reviewed publication. For consideration of this manuscript, the publisher requested IRB approval documentation. Because an IRB was not involved prospectively (because the study did not meet the definition of research with human subjects), the group “worked the system” and requested an IRB to declare the study as exempt from IRB review as it involved existing data, Nebeker said. This is not ideal, as the federal definitions within the Common Rule do not speak to self-study or self-experimentation, and yet journals are bound by the traditional conventions requiring research involving humans to obtain an IRB approval, which leads to the question of how to better support the responsible and ethical conduct of self-study.

Power and Privilege in Citizen and Personal Science

Acknowledging and Balancing the Power Differential

Hekler drew from his own experience to discuss further the gaps in power and privilege in citizen and personal science. He spoke about how he first met Lewis when she was giving a talk at a conference about her OpenAPS project. He was very interested in the project and offered to help by conducting a clinical efficacy trial. Lewis politely declined, and Hekler realized that he had made the mistake of assuming he knew what he could do to help. This is an example of the psychological bias called the Dunning–Kruger effect, which, he said, could be summarized as “confident ignorance.”

Afterward, Hekler and Lewis together identified the way in which he could contribute, and they applied for and were awarded a grant from the Robert Wood Johnson Foundation to study how to open pathways to innovation for citizen scientists. During the planning for this grant, Hekler said he began assuming the role of principal investigator (PI) for the study because he has training as a PI. However, he and Lewis came to realize again that this was Lewis’s study and that she should fulfill the role of PI. As part of the study, traditional and nontraditional experts were convened to discuss opening pathways for innovation and to specifically consider the issues of power and privilege. Another learning experience for Hekler was being assigned by Lewis to be a nonparticipating note-taker for the entire workshop. Although, at first he felt that he was better suited to help synthesize the group outcome, he said, he came to realize that contributing

from his perspective as a white male of privilege would have compromised the desire for equitable participation from all who can, particularly nontraditional patient scientists. He said that he now tries to be conscious about the impact of gaps in power and privilege when working with citizen-led research projects.

It is critical to understand and address issues of power and privilege, Hekler said. There are implicit biases and structural forces that assign status to people. It is not enough to invite particular individuals to participate (i.e., tokenism), he said. The environment must be such that these people also feel comfortable and safe in participating. Furthermore, he said, those who traditionally hold power (e.g., professional researchers) need to recognize when they are falling into the trap of confident ignorance and emotional blindness (i.e., never having experienced that which someone else is describing and thus not being able to relate and truly understand).

Self as Subject

The principle of respect for persons participating in research is embodied in the informed consent process, Nebeker said. The challenge is to ensure that same respect for persons in a self-study. How does an individual consent himself or herself? For the Quantified Self participant-led study, which involved the collection of blood via finger prick, she said that an explicit process was developed to help self-researchers through the decision-making process, considering both the potential risks and the benefits. She noted that even though each person was conducting self-study, there was the potential for outside influence (e.g., not wanting to disappoint others in the group).

Hekler said that the question underlying self-study is who has the power and privilege to be able to ask the questions and define success. For example, he said that Lewis and Leibrand have the capacity to develop OpenAPS (e.g., time, resources, education, a network). Acknowledging this power and privilege, Lewis’s intention in securing the grant from the Robert Wood Johnson Foundation was to enable pathways to innovation for others, Hekler said. There are many examples of approaches to building the context and the capacity that can enable individuals to use research to support their personal needs, Hekler said (e.g., community-based participatory research partnerships, youth participatory action research networks, and initiatives such as the Patient-Centered Outcomes Research Institute).

Structural Inequalities

Structural inequalities prevent many people from engaging in self-research. Too often the institutional reaction is to solve the problem for

those affected, Hekler said. But there are often unintended consequences resulting from a lack of understanding of the context (e.g., food aid provided to developing countries can adversely affect local economies and perpetuate food insecurity). Instead, initiatives should focus on building the capacity for people to ask the research questions that are relevant to their own context and lived experiences and, ultimately, solve their own issues, with advice, support, and plausible solutions that worked elsewhere provided by professionals.

Institutions are not supporting citizens who want to better understand their own lives in the context of their health, Nebeker said. She recalled an example from her own institution in which one of her students who was interested in studying her own pain designed a self-study and approached the university IRB for approval. The IRB’s context for self-study was limited to that of university physicians or researchers studying themselves, Nebeker said. The members of the IRB had no understanding of why a student would want or need to conduct a self-study, and they advised her against it. This example is indicative of the regulatory definitions found in the Common Rule limiting the ability of people doing self-study to seek an external review. With the growth of citizen science applied to health, this is an area that NIH could help to guide by supporting research on developing relevant infrastructures.

Wilbanks described how structural issues could lead to people essentially being coerced into sharing their information and submitting to surveillance, sometimes for ethically questionable purposes. Those who can afford the latest technology devices can choose to engage in self-study that is relatively ethical and free of surveillance, he said. Others will often accept free devices and service in exchange for allowing access to their data because this is a way to obtain desirable devices that they would not otherwise be able to get. As an example, he said that the Apple Heart Study offered individuals who agreed to participate the opportunity to obtain Apple watches at a subsidized price. The freedom to participate in self-study without giving in to surveillance must be structurally built in, Wilbanks said.

The National Institutes of Health’s Role in Guiding Unregulated Research

Nebeker asked the other panelists to expand on how NIH might work with citizen science and personal science organizations to guide unregulated research. Wilbanks offered several potential areas where NIH could focus.

Education

NIH should invest in educating citizen and personal scientists on clinical research, Wilbanks said. People need to understand that there is a

process of gathering data, analyzing data, and drawing conclusions and that scientific conclusions are not truths but rather are claims that “exist at varying degrees of truth over time.”

Clarify the Application of the Common Rule to Research Supported by Indirect Funding

Many institutions in the United States believe that only direct funding is covered by the Common Rule, Wilbanks said. Citizen science organizations question why their members should follow the Common Rule when they are aware of institutions that claim that their research supported by indirect funding is not covered by the Common Rule. A clarifying statement from NIH would be helpful, he said.

Engage the Citizen Scientist Community as Equals

Feedback from the citizen and personal science community has revealed that they have often felt unwelcome and disregarded by the institutional science community, Wilbanks said. He urged NIH to create opportunities for the institutional, citizen, and personal science communities to meet as equals and learn from each other. He suggested attending citizen science association meetings and hearing their critiques and concerns.

Create Safe Harbors for Data-Hosting Platforms

Safe harbors are needed for data-sharing platforms, Wilbanks said. Although Internet platforms are not liable for copyright infringement resulting from information posted by users, there are no such protections for data hosting platforms. Data-hosting platforms that are trying to do the right thing can still be held liable in the event of a failure—and to the same extent as a malicious actor, he said. Legislating a safe harbor for data-hosting platforms that limits damages would be helpful, he said, although he acknowledged that this is not likely to be a priority for Congress in the near future.

Oversight Systems: Principles Versus Processes

Developing and evaluating different systems of oversight that could be applied to both citizen science and mobile health could be an area for NIH to consider funding, Wilbanks suggested. What is needed is not more principles of bioethics, Nebeker said, but rather practices that evolve with the changing ecosystem. She said that the community needs to be engaged, more authentically and more frequently, in discussions about shaping the

future of citizen science and self-study. Patients and the public have not been asked about their understanding of their data, of who wants to access their data, or of what can be learned from their data. To apply the principle of justice it is necessary to consider who is being included and excluded and who stands to benefit, she said.

The principles of bioethics are being confused with the process of bioethics, Wilbanks said. He agreed with Nebeker that the principles are sound, but he said that specific institutional processes might not meaningfully reflect those principles. For example, the objectives of researchers have become to get IRB approval and then consent and enroll patients into a study, not necessarily to meaningfully think about the risks and benefits of a study and then to meaningfully inform potential participants about them. NIH should reinforce the principles and develop new processes and practices that implement them, he added. Hekler suggested an organizational structure for facilitating oversight of the implementation of ethical principles. Rather than a centralized structure or one composed of subgroups, he proposed a resilient, diverse network of different groups with different powers that allows for checks and balances, much like the logic of the U.S. Constitution. The first step, he said, will be to understand how to balance those powers in the face of systemic biases.

This page intentionally left blank.