Medicine is undergoing a great revolution that is radically transforming health care by not only improving diagnostics and therapeutics but also providing new understandings of disease prevention [1,2]. Contributing to such transformation are some intimately intertwined factors such as the introduction of systems medicine (application of systems biology approaches to disease); big data (availability of large quantities of data); new technologies (that allow study of new magnitudes of individuals’ data); and patients’ increasing participation, engagement, and involvement in health care [1-6]. Large amounts of data are necessary to make sense of systems medicine, and, in turn, sophisticated technologies are crucial for big data to be adequately analyzed and interpreted. Similarly, individuals’ participation, engagement, and involvement in health care are crucial for a constant supply of big data because, by sharing their data with research, individuals can contribute to the accumulation of a vast amount of information [7]. The convergence of these elements is typically referred to as “4P medicine,” a term initially coined by systems biologist Leroy Hood to indicate a type of medicine that is predictive, preventive, personalized, and participatory at the same time [2,4,8-11]. However, 4P faces both technical and societal challenges [2]. While there appear to be rapid technical developments in the capacity to process people’s data, the transition from a reactive to a proactive, empowered approach to medicine seems to pose greater difficulties [1-3]. The key concept of 4P medicine that “the individual is at the center of action-taking related to health and health care” [12] raises the question of not only how to best educate patients, physicians, and public about the opportunities offered by 4P medicine but also how to acquire the amount and type of data necessary for predictive and personalized medicine to be realized [2,12-14]. Reaching out to larger communities of individuals may require looking beyond the traditional researcher-led enrollment system conducted in university hospitals and research centers. Patient-activated networks and patient-initiated research activities may help address this challenge [2,3]. According to many, it is exactly these patient-initiated phenomena that will be the most powerful tools in the strive to push forward the 4P agenda because they may increasingly provide data while fostering individuals’ empowerment and boosting transparency and accountability in science at the same time [3,12,15-18]. We set out to analyze the opportunities and challenges these activities offer and focus on the ethical oversight requirements of such phenomenon.

The major strategies adopted to promote a participatory approach to medicine involve increasing individuals’ participation, engagement, and involvement in their health and in health research [5,19]. According to Woolley, participation in medical research encompasses activities that involve not only an active, intentional role but also more passive forms of inclusion [5]. The concept of public engagement in scientific studies, on the other hand, does not depend on individuals’ participation in research, that is, individuals can feel engaged even if they do not actively participate in the research. Engagement can be higher or lower according to scientists’ effort to communicate their intentions and request public’s collaboration in collecting data through so-called participant-centric initiatives [5,20]. Finally, involvement characterizes activities where members of public can play an active role in initiating research, selecting the scientific questions to address, and designing a study and implementing it [5,21]. This latter type of initiatives is commonly referred to as participant-led research (PLR), participant-driven research, or participatory research, and it includes a wide spectrum of approaches such as self-experimentation, self-surveillance, analyses of genetic information, and genome-wide associated studies [22,23]. In this study, we decided to adopt the label “PLR” over others because we believe it better emphasizes the main characteristic of this activity, which is not only initiated but also conducted by those who participate in it.

PLR has been facilitated by the integration of a wide range of increasingly affordable technologies into everyday life like computers, smartphones, tablets, and wearable gadgets and by the emergence of social media platforms where people can nowadays share (health-related) information about themselves to be used for a variety of research purposes [23]. For example, platforms like PatientsLikeMe and the Quantified Self offer people the opportunity to design, conduct, and analyze their own studies by uploading different types of data about themselves [12,18,24]. An example of a successful PLR initiative, the results of which were eventually published in an international scientific journal, is the lithium study, in which a group of patients suffering from amyotrophic lateral sclerosis and belonging to the PatientsLikeMe community initiated and conducted a study on the effects of lithium on their condition [25]. The study findings were later confirmed by a standard clinical trial [26]. While the lithium study involved testing a substance on one’s own body, most PLR activities involve uploading one’s data to Web-based platforms (often genetic data derived from direct-to-consumer genetic testing) with the scope of starting or contributing to a research project [15,27,28]. Users may want to publicly share their data on Web-based platforms for a variety of reasons. For example, studies have found that users may want to learn about themselves or contribute to the advancement of medical research and, in the case of genetic and genomic data, want to improve the predictability of genetic testing because they find it fun to explore genotypic and phenotypic data [29,30-34].

PLR activities have some clear limitations, such as an evident self-selection bias due to the fact that its participants usually come from a selected, hyperactivated, and highly educated subset of the general population, leading to little variance in the sociodemographic characteristics of those who join it [35]. However, there is much consensus on the opportunities offered by this novel approach to science [19,22]. First, PLR promotes individuals’ empowerment, one of the pillars of the patient-centered health care model, by allowing people to make more informed and autonomous decisions on their own health [22]. From a patient or participant perspective, taking part in PLR could fulfill individuals’ need for involvement and self-determination that can otherwise be lacking in investigator-led research and that is currently urged in patient care [36]. It can also create opportunities for social support among individuals sharing the same condition or health concerns [21]. PLR supports the democratization of research, a system where anyone has a human right to contribute to science and actively participate in the research process, and significantly helps cut research-related recruitment and logistic costs [18,37,38]. Furthermore, such activities can provide great support to large-scale and longitudinal research studies, accelerate the pace of their execution, and explore areas that standard medical research often overlooks or cannot reach [13,22,39]. Just as in other forms of crowdsourcing, the underlying belief that fuels PLR activities is that the more people are allowed to participate, the more accurate and complete the generated information will be [40]. However, due to the general lack of qualified supervision that distinguishes standard research practice, these bottom-up initiatives challenge existing ethical paradigms and raise questions that scholars have only recently started to address [13,40].

Just like investigator-led research, PLR poses evident questions related to its ethical, legal, and social implications, potentially resulting in barriers to the optimal integration of its outcomes into scientific evidence and, ultimately, into health care [3]. Beyond issues of accountability, the main concerns are whether PLR can be conducted (1) in a scientifically thorough fashion and (2) in an ethically appropriate manner [14,37]. The first concern follows the consideration that being self-reported, self-collected, and mostly generated without an experienced researcher’s supervision, data produced by PLR may not meet the highest scientific standards characterizing investigator-led research and contain major biases that may lead to questioning data’s reliability and validity [22,26]. In response to this worry, recent scholarship has proposed that PLR can, at least in principle, reach the same level of scientific accuracy of standard research, provided that participants are adequately trained on how to collect and report their data [16,41]. Furthermore, with the widespread use of these approaches, PLR holds the promise of introducing increasingly novel methods for validating its results to secure their publication in international journals and their integration into health care practice [19]. The second concern builds on the assumption that PLR activities bear potential risk of harm for those taking part in them [15,23]. For example, testing off-label drugs without a researcher’s supervision might lead to serious health consequences as much as sharing one’s identifiable genetic data on Web-based platforms may lead to privacy issues that can result in discrimination by employers or insurance companies [7].

Following these considerations, some scholars have argued that PLR should be ethically regulated, and a debate is currently taking place regarding what forms of ethical oversight mechanisms should be adopted in such contexts [16]. While some scholars believe that we should try to capture such research within existing regulatory frameworks (requiring, for example, ethical review by an Institutional Review Board; IRB) [42], Vayena et al have proposed that existing ethical standards should be applied to the specificities of participant-led health research with alternative mechanisms [16]. They have proposed a distinction into 3 categories representing different levels of similarity between a given PLR project and standard research, the level of risk involved, and the type of agent conducting the research; they suggest crowdsourcing review as an alternative method of ethical oversight [12,16,21,23,42]. In particular, a given PLR activity will fall into the first category (and, thus, it will be subject to the standard form of ethics review) if it is performed by state or for-profit institutions; if the activity does not meet the “institution-plus” criterion, it will fall either into the second category (if it involves more than minimal risk to participants) or in the third category (if it involves no more than minimal risk to participants) [16]. While the second category will demand a form of ethics review equivalent to an expedited review (eg, a faster review conducted by the IRB chair and one or more experienced reviewers), no formal ethics review is morally required for the third category [16].

The need to adapt ethical standards to PLR also finds justification in the great amount of evidence showing that participants are more willing to donate their data when an ethical review body has approved the study protocol [43]. But establishing the appropriate ethical oversight mechanism in PLR is not without challenges. In fact, these activities represent a revolutionary movement in contrast to mainstream research, and, as such, they often position themselves as opposed to the traditional elitism of standard research practice [38,44]. Furthermore, applying the same mechanisms of ethical oversight that are also employed to review standard research is likely to impose a burden on PLR in terms of finances, time, and logistics, with the possible consequence of discouraging participation [13,21]. Thus, we can expect some forms of resistance by this novel approach to science against any top-down attempts to regulate it.

The literature suggests a substantial lack of information on individuals’ perspectives regarding ethical oversight in participant-led health research [13,16,38,42]. Previous research has explored the reasons why users decide to publicly share their data on Web-based platforms for research purposes and what prevents them from doing so [29], but no study has so far investigated participants’ perspectives on what ethical oversight methods should be in place in such settings. Even scholarship from other disciplines that has investigated users’ attitudes toward Web-based platform policies has so far addressed issues such as privacy [45] and copyright [46], but overlooked medical research ethics. Considering that most participant-led health research activities take place on Web-based platforms and involve publicly sharing individuals’ data in anonymized, coded, or identifiable forms, the goal of this study is to investigate whether the users and visitors of a Web-based data-sharing platform apply the same ethical principles of standard research to Web-based data sharing and, in particular, to Web-based participant-led health research projects. Furthermore, we aim to investigate the mechanisms of ethical oversight that users think should be adopted in such a context.

We will adopt survey methodology in this study. We target the users of the Web-based data-sharing platform Open Humans who have subscribed to the Open Humans newsletter [47]. Initially, we developed a conceptual model that seeks to explain individuals’ ethical oversight expectations of data-sharing platforms with their past data-sharing behavior, their previous participation in research, their perceived importance of having transparent guidelines on Web-based data-sharing platforms and perceived knowledge of them, their attitudes toward data sharing and self-experimentation, and their future data-sharing intention (see Multimedia Appendix 1). Subsequently, we developed a Web-based questionnaire using the survey platform SurveyMonkey [48] on the basis of the literature discussing the main ethical principles applying to standard research and, in particular, Emanuel et al’s ethical framework for clinical research [49]. We developed items to measure the variables that potentially have a relationship with users’ opinions on ethical oversight, such as past data-sharing behavior [29], previous participation in research [50], attitudes toward self-experimentation [51,52] and data sharing [30-34], perceived importance of having transparent guidelines, and knowledge about the platform’s guidelines [30,31]. To content and face validate the questionnaire, we conducted a pretest with a convenience subsample of 6 participants. We contacted potential pretest participants through the Open Humans platform and asked them to fill out the Web-based survey by providing specific feedback on the clarity and appropriateness of each survey item. Data collection continued until data saturation was achieved.

Once the pretest had been conducted and the survey questions refined, we created a research project on a dedicated page of the Open Humans platform where we described the scope of the study and provided a link to the Web-based survey. Open Humans research projects are meant to ask an engaged audience of participants to join and contribute to research. Past and current research projects include the Genevieve Genome Report (matching participants’ genome against public variant data), the Twitter Archive Analyzer (to explore social media usage), and the Keeping Pace project (seeking to study data about how participants move around and to understand how seasons and local environments influence their movement patterns). Altogether, Open Humans research projects have so far involved more than 3000 users.

An invitation to visit the research project Web page and fill out the Web-based survey was sent through the Open Humans regular newsletter to all subscribers (approximately 4640 users) on June 13, 2018. Those who do not have an account on Open Humans are requested to create one to be able to fill out the survey. We expect a response rate of 5%-10% (between 200 and 400 completed surveys out of approximately 4000 survey invitations sent), in line with previous research [53]. To increase the response rate, a second newsletter including a reminder about our research project and the upcoming deadline was sent to potential participants in mid-July 2018.

By clicking on the link to the survey, participants are directed to a dedicated page of the Open Humans platform describing the project’s goal (“Data sharing and ethical oversight”) and its academic and nonprofit nature, providing the name and contact details of the principal investigator, and providing the informed consent form as well as a downloadable version of it. Users can only be redirected to the actual survey if they consent to participate in the study by clicking on the corresponding button. If they agree, the participants are directed to the survey on surveymonkey.com, where a short introduction reminds them about the study’s scope and guarantees that no data are extracted from participants’ accounts. An anonymized unique identifier is attributed to each Open Humans platform account to detect multiple entries from the same individual. If such multiple entries are found from the same individual, we will keep the first entry for analysis. To start the survey, participants are asked to confirm that they “have joined the project on openhumans.org, read the description, and accepted the corresponding consent form.” The survey displays 1-3 questions on each page according to the questions’ length, and participants can review and change their answers through a “back” button. We ensured that survey questions are appropriately displayed on mobile phones. Completion of the survey is estimated to take approximately 15-18 minutes. To avoid missing data, answers to the questions are mandatory, except for sociodemographic variables. We offer no remuneration for participation in this study. However, we ask participants whether they would like to receive their responses to the survey and the aggregated answers from all respondents for comparison.

Once the survey has been closed, we will import the data into SPSS (IBM Corp, version 24.0). We will compute frequencies and correlations and use chi-square tests to assess the relationships between participants’ ethical oversight expectations and their past behavior, future intentions, beliefs, attitudes, and knowledge. A priori power analysis suggests that a sample size of 142 respondents would be adequate to detect a moderate effect with alpha=.05 and power=.8 [54]. Thus, our proposed minimum sample size of N=200 will be more than adequate for the main objective of this study. Data missing at random will be handled using multiple imputation [55]. Depending on the nature of the comments provided in the open-ended questions, we might also be able to gather important qualitative insights. In this case, two researchers will code the comments and label them. Similar labels will then be merged into broader themes to provide a comprehensive description of the findings [56].

The study protocol, including survey questions, has been approved by the Ethics Commission of the Federal Institute of Technology, Zurich on May 7, 2018, with the title “Ethical oversight in online data-sharing platforms” (Ref.: EK 2018-N-36). We expect there will be no risks to participants in this study.

This is the first study to explore individuals’ ethical oversight expectations regarding research projects on open data-sharing platforms. For this reason, we developed our own conceptual model in an effort to explain participants’ ethical oversight expectations through their past data-sharing behavior, their previous participation in research, their perceived importance of having transparent guidelines on Web-based data-sharing platforms and perceived knowledge of them, their attitudes toward data sharing and self-experimentation, and their future data-sharing intention (see Multimedia Appendix 1). To build measures for our variables of interest, we relied on the literature on the ethical principles of standard research [49,57-59], on what characterizes PLR [12,16,19,37,60-62], and on barriers to and facilitators of data sharing in varous contexts (eg, biobanking) [30-32]. Below we have described how survey questions were created on the basis of the literature across all variables. The survey questions can be found in Multimedia Appendix 2.

Research is likely to benefit from fostering more engaged research participation [3]. The integration of digital tools into everyday life is facilitating people’s active involvement in research, and PLR activities are becoming an increasingly popular phenomenon [70]. PLR initiatives hold great promise for science, not only because they are a tangible expression of a response to the push for a more participatory medicine but also because they offer significant opportunities for advances in a variety of fields by offering novel solutions to complicated challenges [22]. Yet, if not duly overcome, some challenges are likely to become burdensome bottlenecks to the successful realization of these activities. In particular, ethical oversight mechanisms need to be adapted to PLR to ensure that the same ethical standards of investigator-led research are fulfilled while PLR values are, at the same time, respected and promoted [16,21,23]. Some alternative solutions to standard ethics review (such as crowdsourcing ethics review) have been proposed [16], but evidence is currently missing on the perspectives of individuals who are directly involved in PLR activities. Our study represents the first attempt to explore what ethical oversight mechanisms users and visitors of a Web-based platform hosting research projects think should be in place. The results will also inform a broader debate on PLR and its potential impact outside the medical realm. In fact, the potential significance of PLR is much broader than medical research and also invests health or wellness research and nonmedical human research (this reflection was contributed by an anonymous reviewer). On the basis of the type of data that our participants report to be engaged with, we will be able to establish which fields of application are suitable for the preferred PLR ethical oversight mechanisms.

A number of limitations to this study are worth mentioning. First, we will recruit our participants from a relatively small, US-based data-sharing platform; therefore, the results—while representative of the Open Humans population—might not be generalizable to other platforms. The decision to restrict our inquiry to the Open Humans community has been dictated by the availability of the platform to join this study. Furthermore, because this is the first study of this kind, we aim to collect an initial set of information from a single, circumscribed group of people. A second limitation is that while the users of Open Humans can share their data for research purposes by contributing to dedicated research projects, other platforms such as PatientLikeMe represent more common venues to start leading PLR activities. However, Open Humans represents a fast-growing platform for initiating and conducting research projects with a diverse and highly involved community. Third, we will ask our participants what potential ethical oversight mechanisms they think should be in place within research projects on the Open Humans platform, thus, introducing a hypothetical bias [71]. Soliciting their opinions solely on already established oversight mechanisms might have provided a more valid and reliable account, but in the current state, it remains unclear what these would have entailed. Finally, we will compare engaged and nonengaged users on the basis of their answers to the past data-sharing behavior questions. Having access to their account activity data would have represented a more reliable and objective measure of engagement. However, we purposely decided not to include this type of data to ensure that participants feel comfortable in sharing their opinions and experiences with the research team.

This study is the first attempt to elicit individuals’ perspectives on the ethical oversight of Web-based PLR activities and to compare such views with their experiences, beliefs, knowledge, and sociodemographic characteristics. Future research should replicate our effort in novel methodological and contextual ways. First, because our study will follow a quantitative approach, qualitative research might be a valuable approach to further investigate people’s expectations regarding which forms of ethical oversight should be applied to PLR. Second, as we restricted our enquiry to a single platform, exploring and comparing different geographical settings will certainly provide more insights on users’ expectations and allow for a better refinement of any policy recommendations. Furthermore, it would be interesting to study whether practitioners’ beliefs about ethical oversight can differ on the basis of research type (including medical research, wellness research, or scientific research involving human data).

To maximize the extent to which it can benefit from research, society has a legitimate priority and concern about protecting research participants and ensuring that a high-quality research is conducted in an ethical manner [23]. As PLR is capable of producing generalizable scientific knowledge, just as standard research, it is crucial to understand what criteria are important to PLR participants to determine who should be in charge of ensuring that standard ethical principles are satisfied [23].