Advocating Values through Meaningful Participation: Introducing a Method to Elicit and Analyze Values for Enriching Data Donation Practices in Healthcare

The secondary use of routinely collected patient data made possible by the broad consent form is seen as a prerequisite for developing data-driven health technologies. In Germany, relevant stakeholder groups (e.g., ethics committees and data protection authorities) specified the broad consent form; however, only one group of patient representatives was consulted, which may indicate asymmetries in engagement. This situation informed our research on medical data donation and emphasized foregrounding patient values. Drawing on participatory design, value sensitive design, and emerging research on value-led participation, we propose a method consisting of (1) a workshop concept for participatory value elicitation composed of four carefully coordinated phases and (2) an analysis procedure to examine the empirical data collected. This analysis allowed us to derive design requirements for medical data donation user interfaces. We conducted three workshops with patient advocates of vulnerable groups and patients in residential care of a psychosomatic unit. Our findings provide new directions to improve user interfaces for medical data donation: First, user interfaces need to enhance patients' reflective thinking about the potential consequences of their data donation; second, a decision facilitator supporting patients' value-based decision-making (e.g., by providing simple language or tailoring descriptions to patient needs); and finally, a data intermediary relieving patients' decision-making and giving them control over their data after donation. Moreover, we emphasize the need to increase the use of participatory approaches in health technology development.


INTRODUCTION
Data-driven health technologies are a promising approach to the improvement of individualized medicine.A prerequisite for such advances is the secondary use of routinely collected patient data [72].As a result, initiatives, policymakers, and healthcare providers are increasingly calling for national and even cross-national health research data platforms [63] to facilitate collecting and sharing data.In support of this development, a GDPR-compliant standard 1 , the "broad consent" form, has been developed for all German university hospitals [85].By signing the broad consent form, patients agree to the secondary use of their data across all participating university hospitals and federal states without any predefined medical research purpose.In an iterative consultation process, the requirements for the broad consent form were specified by "all relevant stakeholder groups and authorities" [85].More precisely, all 52 German ethics committees for medical research and all 18 German data protection authorities at the federal and state levels were involved.Surprisingly, there was only one consultation group with patient representatives from the Federal Joint Committee.This unequal involvement might have led to asymmetries of engagement among stakeholders in this consultation process.Dahl and Sharma [16], for example, emphasized that selecting between the requirements of multiple stakeholders with differing interests, perspectives, and values might cause bias by factors like professional background, social status, or eloquence.We hypothesize that such power asymmetries in the consultation process of the broad consent form development could have bypassed patient values, especially concerning their autonomy and privacy [38].Particularly in healthcare, large-scale data collections require special attention as patients may belong to vulnerable or stigmatized groups [56,80], accompanied by the risk of potential privacy violations (e.g., [19]).
In our ongoing research on medical data donation, working closely with patients, physicians, medical ethicists, and legal experts, we realized the need to consider patient values through a participatory design process.We believe that considering values is a prerequisite for designing socially responsible technologies in healthcare.We developed a method to understand patients' values in medical data donation processes based on the broad consent form.In this regard, we investigated the following research questions: RQ 1: How can we engage patients to systematically explore, articulate, and reflect on their values regarding medical data donation?RQ 2: How can we translate the collected values into design requirements to inform medical data donation user interfaces?
To tackle these research questions, we designed a method consisting of (1) a workshop concept for participatory value elicitation and (2) an analysis procedure to examine the empirical data gathered.We considered two strains from the field of human-computer interaction (HCI) and computer-supported cooperative work (CSCW) that informed our method: first, participatory design (PD) to understand better what it takes to achieve active user involvement (e.g., [4,59,69]) and the theoretically founded approach of value sensitive design (VSD) [28] to account for human values in technology design.We then leverage these two strains by scrutinizing foundational research on value-led PD (e.g., [41,50]).
We facilitated three consecutive workshops with patient advocates and patients in inpatient care.We gathered empirical data through elaborate, sequential activities that supported participants' capabilities and critical reflection on medical data donation processes.We systematically analyzed the empirical data and uncovered design requirements strongly endorsing patient values: First, enhancing patients' reflective thinking on the potential consequences of their data donation; second, a decision facilitator to support patients' value-based decision-making; and finally, a data intermediary to increase patients' control over their data long-term.We hope our method, consisting of a workshop for participatory value elicitation and analysis procedure, serves as a blueprint, supporting researchers, practitioners, and policymakers in the context of healthcare and beyond.Thus, we make the following contributions through our research: • we provide a literature overview of PD and VSD accompanied by related work focusing on healthcare leveraged by essential research emphasizing value-led PD; • we present findings from three workshops for participatory value elicitation (one with patient advocates of vulnerable groups who we consider as patients based on their or a family member's rare disease and two with patients in residential care of a psychosomatic unit); • we suggest an analysis procedure to evaluate the empirical data gathered systematically; and • we show how to derive empirically grounded design requirements informing the design of health technologies, i.e., medical data donation user interfaces.
This article is structured as follows: First, we provide a literature overview of related work that strengthened our method in Section 2. In Section 3, we synthesize the insights and introduce our method, i.e., a workshop concept for participatory value elicitation accompanied by preliminary measures and participant recruitment for three workshops conducted and an analysis procedure to examine the empirical data gathered.We then present our results in Section 4, separated by participants' contextualized understanding of their values, on the one hand, and, on the other, how they materialized values for an idealized data donation.Based on these results, we discuss the design requirements derived and reflect on the suitability of our method in Section 5. Finally, we describe the limitations of our research in Section 6 and give directions for future work in Section 7.

RELATED WORK
In the following, we provide an overview of PD in Section 2.1 and VSD in Section 2.2 accompanied by related HCI and CSCW research in the healthcare domain.We then leverage both strains by considering research emphasizing value-led PD in Section 2.3.In each section, we focus on the strengths and gaps of each approach and conclude with a summary, helping us to develop our method.
2.1 Supporting Participation in Healthcare PD flourished as an action-oriented research approach called "cooperative design" in Scandinavia in the 1970s and was fundamentally influenced by the Norwegian Nygaard [10].Nygaard's work was characterized by an explicit political stance on sociotechnical designs through the collaboration of designers and users [5,10,69].In HCI and CSCW [62], PD gained increasing traction internationally throughout the 1980s.The potential of PD was seen as an opportunity to enable collaboration alongside research in commercial settings (e.g., designing systems in work environments) [10].Since then, PD is considered a central approach to democratic decision-making [4,59,69], for example, by considering aspects that help frame active participation (see [36]).
Research, however, has addressed several shortcomings of PD, such as how to create participation opportunities in environments beyond the work environment [39] or how to approach an entire design process through PD without focusing only on refining artifacts [75], as researchers tend not to apply PD methods systematically (see [76]).Regarding healthcare, Rothmann et al. [67] argue that PD approaches may conflict with different sociodemographic backgrounds and needs of patients, the organizational hurdles of medical institutions, and the increasing complexity of health technologies.
Despite this assumption, research shows new directions for healthcare contributing to PD. Davidson and Jensen [17] investigated whether involving older adults in PD leads to more beneficial health technologies.In their study, older adults co-designed new design ideas for mobile health applications.These ideas were then evaluated by experts from academia and industry regarding novelty and creativity.The authors found that despite participants' lack of prior experience with mobile devices, they could contribute creative ideas and anticipate co-design activities.For future research, the authors suggested that PD involving older adults can help to address their needs, support their capabilities, and enrich the design of health technologies.Similarly, Harrington et al. [37] approached PD through a design workshop to explore local communities' experiences, needs, and concerns to inform future health technologies.Their findings based on design workshop activities (e.g., using poster boards, paper prototypes, or storyboards) showed that the health needs of a community are strongly interwoven into sociocultural interactions.The authors concluded that PD approaches should go one step further and use design workshops in such a way as to not only understand the health needs of a community but also reflect on social and political perspectives and contexts.
In summary, PD is seen as a unique vehicle for inducing users' agency and autonomy in design processes [21] and to overcome power imbalances maintaining individuals' privacy (e.g., of vulnerable groups [55]).Embedded in a particular community, PD can then enable a broader discussion on political and legal burdens.Yet, in healthcare, we assume the challenge remains to unfold patients' different viewpoints, needs, and concerns in more detail.We believe that for developing new design interventions in healthcare, more pluralistic ideas should be approached [3], for example, through creative, collaborative activities among participants.In the following, we consider these insights from the perspective of values.

Foregrounding Values in Health Technology Design
Values reflect individuals' beliefs about what they regard, for example, as good or bad, desirable or undesirable [1]; they transcend specific situations and generally apply to social life [22].Cheng and Fleischmann [14] conclude that values serve as "guiding principles." Friedman and colleagues [29] define human values more broadly as "what is important to people in their lives, with a focus on ethics and morality."In 1996, Friedman [26] emphasized the need to consider values in system design, highlighting that values might be at odds with economic aims, challenging to articulate, and incapable of being sufficiently expressed.Questions arose concerning which and whose values in technologies should be considered and how values can be made feasible [74], which finally paved the way for VSD.
VSD distinguishes between direct stakeholders affected by (or interact with) a technology and indirect stakeholders (who seldom interact with) but are nonetheless affected by it.By approaching these stakeholders' values and potential value conflicts that might emerge among them, VSD offers the "tripartite methodology, " consisting of conceptual, empirical, and technical investigations.Friedman et al. [29] highlight this methodology's iterative use throughout the technology design process.
Despite the plethora of VSD methods for eliciting values in HCI and CSCW research (see [29]), Le Dantec et al. [48] critiqued that these methods must be revised to focus more on value discovery (e.g., more effective empirical investigations), allowing researchers and participants to continuously reflect critically on their values.Parvin et al. [43] argued that values should be investigated within the situations they occur, as values do not follow universal definitions.Shilton [74] noted that VSD methods face the complex challenge of general adoption.There are only a few evaluations of whether and how certain VSD methods (e.g., Envisioning Cards [27]) have been effective outside of academic settings [74].
From a healthcare perspective, Lee et al. [49] pointed out the problem that physicians tend to consider values directly related to patient treatment.Values tied to patients' needs, philosophy of life, and background are thus disregarded.Concerning the latter, research has investigated how values can be foregrounded in healthcare settings: Berry et al. [9], for example, investigated how patients with multiple chronic conditions can be supported to reflect on their values regarding their healthcare through different prototypes (e.g., a list or conversation canvas to explore values, selfcare, and health status).The authors argued their approach helps patients to articulate their health needs, for example, personal information they are willing to share with physicians or healthcare providers.Lim et al. [51] argued that patients' values with multiple chronic conditions might misalign healthcare providers' expectations.The authors gave patients a worksheet, accompanying questions, and cards to contextualize, rate, and reflect on their insights.This approach supported patients in investigating their values and everyday activities regarding self-care.The authors stated that their approach helped patients communicate their values and understand better their needs when sharing health information.
In summary, we recognize the flexibility of VSD methods, which can be used iteratively and extended with other methods [29].We affirm that, given the design of health technologies, VSD can be supportive.However, there is a need for enabling participatory value elicitation to anticipate patients' different life situations and medical histories, as values should be examined in contexts in which they occur [43,58].For example, through methods (such as worksheets or cards for contextualizing and reflecting on personal insights [51]) to better understand patients' needs regarding health information disclosure.In the following, we leverage the insights of PD and VSD with foundational research fostering value-led PD.

Establishing Participation Led by Values
Research [4,59,69] emphasizes that researchers and other stakeholders should negotiate their values and contribute equally to sociotechnical designs.However, unfolding different stakeholders' values through PD might challenge researchers to rethink the relationship between methods and participation [40].In this regard, Frauenberger et al. [25] argued that researchers must be clear about choosing and using methods to work with different stakeholder values.Iversen et al. [40] emphasized that it is not only a question of methods to capture values from participants but also how to work with values in the design process, as they can drive PD activities.Values can be partially undefinable or mutually dependent, expecting researchers to provide more in-depth elaboration on how to design with values.Thus, values should be negotiated in dynamic and dialogical processes promoting values' emergence, development, and grounding [40].
In response, Iversen and Leong [42] extended their work [40,41] by solely focusing on establishing, negotiating, and debriefing the emergence of values in a PD process.Out of this investigation, they found that researchers might implicitly mediate values that can influence the outcomes of PD activities.To prevent such influence through researchers' values, they emphasized that value-led PD endeavors should carefully connect expectations (e.g., of a (research) project) to participants' values.To close this gap, Leong and Iversen [50] demonstrated that a value-led PD should approach meaningful alternatives for realizing alternative information technology conceptions.Such alternatives can surface participants' values, leading to material (e.g., a product or artifact) or immaterial outcomes (e.g., participants' understanding) towards system design.To arrive at meaningful alternatives, the authors emphasized more awareness of researchers and their responsibilities towards participants' capabilities (e.g., by carefully facilitating workshops).
In line with this research, Grönvall et al. [34] suggested the continuous negotiation of values in agonistic spaces2 for community-based PD.However, the authors emphasized that these spaces only intervene in stakeholder conflicts when they allow for equal participation.If this condition is met, the authors assumed that values could then be jointly reshaped, thus promoting mutual learning experiences to negotiate dynamic stakeholder values.Peer [64] further emphasized activities that allow participants to think critically about their values concerning data literacy.By conducting workshops to reveal the values of local communities (e.g., on data infrastructures), they facilitated activities for creating data story sketches (two-dimensional representations) or data sculptures (three-dimensional representations).The author reinforced that such activities promote researchers' engagement for a deeper understanding of values and their interpretation regarding certain groups of people.Jonas and Hanrahan [45] went one step further by asking how the shared values of researchers and participants can inform a design.Their research focused primarily on reflexivity and advised researchers to be aware of how their values might conflict with those of participants.Especially when it comes to the participation of marginalized groups, the authors argued that a deeper understanding of research contexts should be gained through researchers' critical selfreflection.
A comparable strand of value-led PD discusses participation and inquiry of values from the lens of experienced-centered design (ECD).ECD aims to establish a dialogue, for example, among individuals or a group of individuals, to promote new forms of knowledge [54].Wright and McCarthy [83] describe the strong humanistic nature of ECD, giving voice to those otherwise excluded from, for example, technology design or experiential design projects, on the one hand, and, on the other, creating opportunities for people to enrich their social experiences with and through technology.Using an ECD-based approach, Foley et al. [23] showed how agency and social interaction among vulnerable groups (e.g., dementia patients) can be promoted by relating shared experiences and participatory activities.The authors advocated that researchers' caring support for inclusive design practices in ECD can support vulnerable groups' everyday lives (e.g., interacting with electronic devices).
In summary, we found several insights in research on value-led PD, helping to design our method.When we (researchers) work with values, supporting and encouraging the ongoing negotiation of values among participants through PD activities (e.g., in workshops) is vital.However, research projects in healthcare can be complex, especially regarding vulnerable or stigmatized groups to ensure their agency and inclusion (e.g., [23]).Before realizing value-led PD, researchers need to reflect critically on their attitudes and those of participants.Furthermore, researchers should carefully document their considerations and methods, supporting future value-led PD efforts [50].In the next section, we specify our method based on the insights gathered from existing research in the discussed fields.

METHOD
Based on the insights gathered in the related work (see Section 2), we pose our first research question: How can we engage patients to systematically explore, articulate, and reflect on their values regarding medical data donation?We answer this research question by proposing a workshop concept for participatory value elicitation.
Research in VSD (e.g., [61,81,84]) provides various methods that helped us develop our workshop concept.However, we found no consensus on how these methods can systematically capture various prevailing participant values of a specific context.Although VSD recommends, for example, a value source analysis (see [29, p. 77]) to identify values of direct and indirect stakeholders, it is not clearly described how this method can be practically applied.Parvin et al. [43] further note that the identification and application of values in technology design should instead yield to using values as hypotheses fostering problematic situations.Similarly, Le Dantec et al. [48] suggested the approach of value discovery, namely that values should be examined in context and that notions of values should be defined and continuously reflected.Regarding the latter, the authors argued that empirical investigations can support understanding values.
We found further directions in research on value-led PD (e.g., [40,42,50]) to understand better what it takes to work with values in participatory settings, on the one hand, and, on the other, our researchers' responsibilities to support participants in partaking in PD activities.As a result, we specified a workshop for participatory value elicitation consisting of four sequential phases: 1 explore, 2 contextualize, 3 translate, and 4 reflect; accompanied by activities, i.e., the value questionnaire, value map, and value scenario (see Figure 1).We (the three authors of this paper) refined our workshop concept iteratively by discussing drafts within our research group and by testing a first version in a pilot study (2.5 h; breaks excluded), as suggested by Thabane et al. [79].For the pilot study, we recruited six participants with an academic background (two students and four research associates).The results of the pilot study led to several improvements: First, we decided to explain our research ethics in detail to participants (see Section 3.2).For this, we welcomed each participant individually and clarified in a personal conversation how and why participants' data are collected, stored, and processed.Second, we improved the description of each activity, i.e., value questionnaire, value map, and value scenario.Informed by research (see Section 2.3), we critically reflected on our facilitators' responsibilities and roles, for example, whether our facilitation supported participants' capabilities and level of knowledge to engage with each activity.
Consequently, we further conceptualized our facilitator responsibilities and roles: One facilitator provided the thematic introduction to the workshop context, another explained the activity descriptions and took responsibility for the timekeeping, and the last took notes on the workshop procedure and activity outcomes.Further, all three facilitators initiated a discussion at the end of each phase in dialogue with participants to create a comfortable atmosphere (see [34]).In the next section, we detail our workshop for participatory value elicitation, participant recruitment, and analysis procedure.

Workshop for Participatory Value Elicitation
In a short presentation, we familiarized participants, i.e., patients, with our research context, namely values regarding medical data donation.Based on insights from our pilot study, we learned that the information provided must be presented carefully to avoid influencing participants' values [41].Aligning with research [54,83], we are aware of the central role and respect towards participants' characteristics (e.g., health constraints, articulation skills, or even motivation to participate), as they provide valuable insights from their lived experiences.We, therefore, note that our workshop concept can be adjusted in duration regarding varying conditions of a cohort without affecting the result of each phase, i.e., activity outcomes.We detail the four sequential phases of the workshop for participatory value elicitation in the following.
3.1.1Phase 1 explore.In this phase, participants explored their values regarding the research context, i.e., values they perceived as important when donating medical data.In the first step, we equipped each participant with a material box (see Figure 2).We then instructed participants to write down values on yellow sticky notes to arrive at an individual compilation of values [48,74].Values are commonly expressed by general terms, such as trust or quality [2].Hence, we engaged participants to feel free to use, for example, keywords or descriptions expressing their values.We aimed for participants to become familiar with articulating values regarding our research context.In the second step, we introduced participants to the value questionnaire (see Figure 1, phase 1 explore).Based on an iterative process within our research group, we developed this questionnaire, encouraging participants to specify the most important value they chose from the previous exploration.Based on five questions (see Appendix A), we intended to enable participants to investigate, describe, and articulate values.In a short discussion, we asked participants to present and discuss their value questionnaires within the group to alter value similarities or differences (see [40][41][42]).3.1.2Phase 2 contextualize.In this phase, we engaged participants to assign values derived in the first phase to stakeholders of our research context on a value map (see Figure 1, phase 2 contexualize).We argue a value map supports a contextualized understanding of stakeholders responsible for upholding participants' values.By revealing stakeholder relationships, commonalities, and value conflicts, a value map helps to negotiate values.
Using pink sticky notes, we started by arranging the main stakeholders of the workshop, i.e., patients, in the center of a pinboard and a limited selection of direct and indirect stakeholders (e.g., medical researchers and physicians) around them.Participants were then asked to add direct or indirect stakeholders they perceived as relevant to the context [60].We indirectly assisted participants in realizing the value map, aiming not to influence them.The following steps were applied recursively, i.e., participants could return to one of the steps if necessary to express concerns or negotiate issues in the workshop group: First, participants assigned the values from the first phase to a stakeholder they considered suitable.Second, the participants rearranged stakeholders based on the values assigned and assumed relationships.Finally, we engaged participants to consider commonalities or value conflicts between two or more stakeholders by implicitly asking questions.We marked each relationship using tape and green sticky notes describing explicit reasoning for a commonality or value conflict.This phase ended with a discussion where the whole group reviewed the value map, focusing on stakeholder relationships.

Phase 3 translate.
In this phase, participants imagined an idealized situation when donating medical data by materializing a value scenario (see Figure 1, phase 3 translate).In contrast to traditional scenarios (see [13]), value scenarios consider both direct and indirect stakeholders [61] envisioning long-term system effects, cultural aspects of society, and values in their use (e.g., describing positive effects or consequences of system use).In our understanding, value scenarios are not limited to a narrative.We intended to encourage different levels of expertise and capabilities.Therefore, we encouraged participants to think of visual representations, for example, drawings or collages supporting participants' storytelling (see, e.g., [30,81]).First, we prompted participants with a short instruction for creating a value scenario (see Appendix B) [57].Second, participants had to select at least three values from the value map they deemed indispensable.We then divided participants into subgroups (consisting of two or three participants) in which the value scenarios were created.We equipped participants with prototyping materials (e.g., Styrofoam, cardboard, or drawing utensils) to support their creativity (see [64]).We finally engaged each subgroup to present their outcome, and all participants provided feedback by highlighting similarities or ambiguities in the value scenarios.

Phase 4 reflect.
In this phase, participants critically reflected on the past three phases and accompanied activity outcomes (see Figure 1, phase 4 reflect).We asked participants to compare their value scenarios with the value map.In doing so, we sought to understand, on the one hand, how a value scenario could improve a situation (e.g., have value conflicts of stakeholders been solved) and, on the other hand, whether the translation of values in the third phase led to a shift in participants' attitudes based on their values from the first phase [71].We outline how we prepared the workshops, recruited participants, and analyzed the empirical data gathered in the following.

Preliminary Measures and Participant Recruitment
We are aware to treat the empirical data gathered from our three workshops as vital.Our university does not have an institutional review board for research projects; however, we are concerned about conducting our research activities ethically.Therefore, we developed an internal research ethics review encompassing the values prevalent in our research group, including legal requirements (e.g., the rights of participants under the GDPR), which were reviewed and approved by our university data protection officer.Our research ethics are primarily about protecting participants' identity, with the overarching goal of cultivating socially responsible cooperation between us (researchers) and participants (see [65]).For example, we reflect these research ethics in our consent forms containing information about our research group, contact details, and study background.This consent form explains the processing and storage of empirical data collected.Regarding the latter, we detail, for example, that we store transcripts securely on a password-protected server and delete audio recordings at the end of a study.Furthermore, we are convinced that our research context focusing on patients of vulnerable groups expects us to be sensitive and attentive when using their data; thus, we aimed to create a trustworthy relationship between the participants and ourselves.
If possible (see Section 3.2.1 and Section 3.2.2),we informed the participants before the workshop, i.e., we sent your study information to participants individually.On-site and before the start of the workshop, we again detailed our research ethics and motivation and gave reasons for our study purpose and procedure.We intended to provide participants sufficient time to decide whether to participate in our study freely.We chose not to offer incentives but reimbursed travel expenses and provided catering.It is also important to us to keep participants informed about the progress of our research beyond the study.Therefore, at the end of each workshop, we asked participants if we could use their contact information to share our progress (e.g., submitted publications or research efforts), sustaining their participation in the future and highlighting their contribution to our research.
We recruited 16 participants in total.We decided to refrain from participants' gender in our results in order to prevent the possible marginalization of, for example, non-binary participants.Also, we deem that participants' identities did not influence our results in any way.However, we demonstrate that our workshop is viable for participants of different ages and vulnerabilities with rare diseases or mental disorders (e.g., physical impairments or psychosomatic illness).Participants had an average age of 55.7, ranging from 37 to 84.Regarding this, we found that based on medical histories, participants, i.e., patients, provided valuable insights into caveats and hurdles they faced when, for example, consenting for admission to medical treatment or residential care.Due to COVID-19, we also paid strict attention to existing hygiene protection measures when conducting the three workshops.In the following, we give an overview of each workshop group, explain the participant recruitment process and their characteristics, and discuss the varying conditions for each workshop.

3.2.1
First workshop with patient advocates.We (the three authors of this paper) facilitated the first workshop for participatory value elicitation in May 2022 with five patient advocates of vulnerable groups.We arranged the workshop in the facilities of an associated partner institution.Anticipating that the first workshop would require more time in terms of clarification and execution (than in the pilot study), we decided to extend the duration of the workshop to 4.5 h (breaks excluded).
This cohort was recruited due to our research activities: one of the participants contacted us after reading a press article presenting our research and emphasized their interest in participating.After an informal meeting, we decided to meet regularly.We argue that these personal conversations established a shared understanding and commitment.Furthermore, we found that this participant is intrinsically concerned about improving medical data donation practices due to their personal experiences (e.g., consenting to multiple exhausting medical treatments).This mutual exchange lasted over six months; as a result, this participant voluntarily supported us in recruiting additional participants with comparable backgrounds (e.g., medical histories) through their network.
Based on these initial discussions, we realized that the involvement and input of this cohort, i.e., patient advocates, to participate in our research was valuable.In their volunteer work, the patient advocates support, for example, patients who need advice on how to deal with their rare diseases.Since the educational opportunities for such patients are limited, patient advocates assist with their expertise, mediate between patients and medical institutions, and strengthen patients' rights by giving them a voice (e.g., informing and educating society about a certain rare disease).In addition, we considered patient advocates themselves as patients since personally or a close family member's exhausting health conditions shaped them (e.g., suffering from a chronic illness or physical impairment).Based on these experiences, patient advocates expressed a critical view toward medical institutions or hospitals as they have faced emotionally stressful admission or inpatient care.

3.2.2
Second and third workshop with patients in inpatient care.We (the three authors of this paper) facilitated the second workshop with six participants and the third with five participants in September and October 2022.These workshops were conducted in an art therapy room at the university hospital facilities since patients (participants) received residential treatment in the psychosomatic unit.We had to reduce the workshop duration to about 3 h (breaks excluded) on the advice of our research partner, i.e., a physician of the psychosomatic unit, due to the participant's physical and mental stamina.However, we carefully maintained the proposed workshop procedure.To comfort participants and support them if necessary, the physician accompanied both workshops (in a nearby room).Since participants were undergoing inpatient treatment, we relied on the physician to recruit participants.Unlike the first workshop, we could not make initial contact with participants before conducting the workshops.This was due to the guidelines of the university hospital, namely that participants' data (e.g., name, e-mail address) could not be shared with us; thus, only the physician could accomplish recruitment.
The cohorts of the second and third workshops were characterized by their psychosomatic illnesses (e.g., depression or tinnitus).Patients suffering from these symptoms receive psychotherapeutic measures such as art therapies for four to five weeks in inpatient care.Compared to people without a psychosomatic disease, the participants suffered from a low mental stress level, which manifests in rapid exhaustion.We regularly consulted our contact person, i.e., the physician, to adequately prepare ourselves to provide a suitable workshop situation for the participants.However, we found that some participants felt pressured by the upcoming activities during the workshop.These circumstances led to one-on-one supervision to lower affected participants' doubts about realizing an activity.For example, one participant could not see or read well and remember what they read (e.g., from the value questionnaire).Another participant was mentally stymied and pressured by the idea of failing at completing an activity.In these cases, we approached the individuals respectfully and attentively.At the same time, our engagement resulted in a more personal relationship and open exchange.This group of people, in particular, requires special attention since even a simple declaration of consent is very difficult to process mentally.

Analysis
We specified an analysis procedure to answer our second research question: How can we translate the collected values into design requirements to inform medical data donation user interfaces?As discussed in Section 3, research suggests performing a value source analysis (see [29, p. 77]) to identify the values of direct and indirect stakeholders.Despite these efforts, we missed more systematic guidance to gather and analyze values.We, therefore, propose an analysis procedure tailored to our workshop concept going beyond purely identifying and applying values (see [43]).
We recorded over 9 hrs of audio materials, which we transcribed verbatim.Due to our research context, we facilitated the three workshops for participatory value elicitation in German to avoid the distortion of the meaning and description of a value by its translation [48]. 3We revisited the photos taken from the activity outcomes, i.e., value maps and value scenarios, in cases where we needed to clarify the transcripts (e.g., one participant only stated the position of a value on a value map without defining the value).We, furthermore, enriched our data with personal notes taken during the workshops.As mentioned, we examined the values of a vulnerable group of people, i.e., patients with mental disease and physical constraints.We (researchers) needed to anticipate this situation and reflect carefully on the data.Hence, our intensive exchanges with participants before (e.g., considering participants' needs to prepare the workshops carefully) and during the workshops (e.g., one-to-one mentoring with participants) helped us to gain more empathy with our participants [70] but also to assess the empirical data gathered from their perspective.Our interpretation is grounded in the relationship between our understanding of the context and the material, i.e., a "reflective act of interpreting meanings" [53].
We concentrated on two parts of the workshops to conduct our analysis: First, on participants' quotes when creating and discussing the value maps (phase 2 contextualize incorporating the outcomes of phase 1 explore) and second, on their interpretation when presenting (phase 3 translate) and reflecting on their value scenarios (phase 4 reflect).We used the analysis tool MAXQDA 4 to conduct a qualitative content analysis [53] by following an inductive coding strategy in the first and a deductive in the second step.To recap, our analysis aimed to investigate participants' values regarding medical data donation.Thus, we employed the following analysis procedure (see Figure 3) to approach our goal: Step 1.In the first step, we investigated participants' contextualized understanding of values.We used the value maps to develop a coding scheme (see phase 1 explore).In doing so, we learned how participants substantiated, interpreted, or even excluded values (e.g., deemed as not meaningful) in the context of medical data donation.Although VSD offers value-oriented coding manuals (e.g., [31,46]), we argue that these were unsuitable for our analysis.We posit that creating a coding scheme based on a value map resulting from a negotiation process leads to a systematic overview of participants' contextualized values to direct and indirect stakeholders.
In order to initiate inductive coding, the first author started from scratch and created codes based on the qualitative data, i.e., values of the value maps, as there was no predefined coding manual aligning with our analysis.The first author screened the material (e.g., photos and notes) and familiarized themselves with the transcripts.In places where the transcripts did not provide clear insights, for example, into the definition or interpretation of a value, personal notes of all workshop facilitators were consulted.The first author compared participants' contextualized understanding of values, i.e., codes, and organized these codes into categories. 5In cases where participants named values differently, but the participants' descriptions indicated an equal value (e.g., "expediency, " "purpose, " "goal, " or "good purpose"), the author merged these values into one code.We (the three authors of this paper) met in regular meetings to restructure and refine the categories carefully, if necessary.Following each category, we thoroughly selected participants' quotes as representative descriptions to strengthen the categories and our results.This analysis step was needed to understand better how participants funneled values chosen from the value map and why some values were not used in a value scenario.We present the findings of the first analysis in Section 4.1.
Step 2. In the second step, we analyzed how participants materialized values regarding an idealized scenario for medical data donations.We used this step to answer our second research, i.e., to translate values into design requirements.Here, we considered the six value scenarios from phase 3 translate.To ensure consistency of our analysis between the first and second steps, we applied the developed coding scheme to the value scenarios using deductive coding.To perform deductive coding, the author assigned the coding scheme to each value scenario to understand how participants materialized values for medical data donation processes.For example, certain value scenarios revealed that values such as "time" or "explanation of consequences" might enhance patients' decision-making when donating data.We (the three authors of this paper) discussed the suggestions participants envisioned regarding new interactions of medical data donation scenarios.We present the resulting findings in Section 4.2.
We used the suggestions derived to arrive at design requirements.The first author reviewed the suggestions and then revisited the full transcript of the discussion of the phases 3 translate and 4 reflect.Screening the discussions was necessary to eliminate ambiguities regarding participants' suggestions, as the value scenarios are visual representations with few descriptions.For example, the author found that a value such as "easy-to-use, " indicating accessibility, was considered to be simple vocabulary, structured contents, or the use of visual elements (e.g., icons).We (the three authors of this paper) discussed the three resulting design requirements and ranked them by their importance for healthcare data practices.We present and discuss the design requirements in Section 5. Based on our analysis, we detail our results in the following section.

RESULTS
Corresponding with the two analysis goals (see Figure 3), our findings are divided into two sections: The first section summarizes participants' contextualized understanding of values regarding medical data donation (see Section 4.1).We included in this section the values mentioned in the value maps and participants' expectations respectively to stakeholder responsibilities (see Figure 4).Furthermore, we extracted value conflicts from the value maps that could be uncovered and might contradict values.The second section presents how participants materialized values for an idealized data donation in the healthcare domain (see Section 4.2).Based on the value scenarios, this section foregrounds participants' interpretations of values potentially realized within the value scenarios.
We assigned color codes to each value representing in which workshop they emerged, i.e., for the first workshop, for the second, and for the third.We grouped the participants as P1 to P5 in the first workshop, P6 to P11 in the second, and P12 to P16 in the third.We detail our results in the following.

Contextualized Understanding of Values Regarding Medical Data Donation
Participants of all three workshops were able to contextualize values regarding medical data donation on three values maps in phase 2 contextualize (see Figure 4).Furthermore, participants shared their insights and negotiated discrepancies (e.g., varying descriptions and meanings of values).
The value maps from the second and third workshops with the patients in inpatient care overlapped in the level of detail.However, the value map from the first workshop with the patient advocates differed, for example, the number of values and degree of detail concerning the description of values.We attribute this difference to the duration of phase 2 contextualize in the first workshop.We had to shorten the duration of the second and third workshops to counteract possible exhaustion or stress because participants, i.e., patients, were in clinical treatment.Furthermore, patient advocates have profound knowledge of consent forms and healthcare data practices due to their occupation or vocation.Based on the first analysis step (see Table 1), we found that the values of time, recognizable use, comprehensibility, and purpose were frequently discussed during the workshops.However, values that occurred less frequently (e.g., sustainability or autonomy) were reflected upon and negotiated more extensively during a discussion, leading us to determine the importance of these values, which nonetheless informed our categories.In the following section, we detail the categories and subsequent discussions among the participants.

4.1.
1 Patients should donate their data sustainably to counter repetitive health data collection.Participants broadly viewed medical data donation as an important contribution to benefit healthcare, for example, by supporting the development of specific medications or therapeutic interventions for patients with a rare disease.However, the fact that medical institutions or healthcare providers collect patient data regularly was a cause for concern.Based on lived experiences, participants strengthened this concern by indicating that, for example, physicians of varying medical institutions repeatedly asked for consent.In this regard, participants mentioned that repeated consenting leads to consent fatigue.They assumed that this consent fatigue would reduce thinking about, for example, the benefits or consequences of donating data and make consent more routine.One participant stated that the sustainability of a data donation (e.g., repeated data donation, which does not appear necessary, without apparent purpose) contrasts data collection in healthcare: "[Sustainability] meant that I would like to donate data once without being asked again.[...] I don't always want to be asked from hospital to hospital." (P2) Additionally, participants noted that patients of vulnerable groups may feel neglected or excluded from medical research if they refuse to consent.One participant encountered this issue by foregrounding autonomy.i.e., patients with a rare disease should be supported and educated (e.g., about the medical research purpose) when they donate data to be able to improve their health status: "I think [a small affected group with very extreme conditions] would be a good example of this dependency [on medical data donation] that it would often be more important for [those] patients to do studies for small diseases with severe limitations [...]." (P9) 4.1.2Patients should retain ownership of their personal information despite data donation.A key question in the first workshop with patient advocates was whether patients completely lose the right to their data (e.g., by allowing healthcare providers to use patient data).Directed at data ownership, participants discussed ownership of donated data towards secondary use: "The data belongs to the patient.That is a donation.What does that mean?[...] On the one hand, [medical institutions] want to use our data completely, but, as far as I remember, it's not that easy.They can't even say that they're using it and that it's theirs.So, the question is: When do I get rid of my data completely?Is that even possible?Who is allowed to use the data?Can someone profit from it without the data donor being considered?" (P1) Out of this discussion, participants debated further expectations, namely traceability of patients' data regarding data ownership.For example, one participant argued that physicians or medical researchers should think about how to increase patient participation, for example, by giving patients information on data usage and potential research outcomes after a period: "[...] [university research groups] collect some data, such as skin samples and anything else, and then they write a research paper about it or whatever.If nothing comes back, neither the promise that the individual patients can call again and get information, [...] that is a no-go." (P5) 4.1.3Patient awareness of data donation processes should be enhanced to avoid personal disadvantages.Participants discussed different views on preserving patients' awareness of medical data donation.One promising direction was seen regarding education, i.e., physicians or hospital staff should support patients to understand better how their data might be used (e.g., for medical research purposes): "[...] in the end, [hospitals] formulate what patients are then allowed to read and have to sign." (P1) We further found that participants extended this discussion by questioning the integrity of physicians and hospitals.Participants draw connections to education (e.g., the seriousness with which patient data are handled).One participant concretized this assumption, stating that the non-transferability increases the meaningfulness of a data donation (e.g., no disclosure of patient data to third parties for commercial purposes).They assumed that this non-transferability would enhance the integrity of physicians and hospitals when educating patients on potential secondary data use: "[Non-transferability], so not that somehow data are collected or sold.[...] I don't think it's so great if it's then sold but rather processed in general." (P8) To support integrity of physicians and hospitals, one participant assumed that information retention could encourage a decision to donate data.They argued that reduced information disclosure would both address privacy concerns by not compromising patient awareness and support redecision for patients to consent or decline secondary data use: "[...] I must be able to say: I would like to donate for this purpose and not that 17 others have access afterward.To have the possibility, I want to be able to decide again in five years [...]." (P5) Similarly, we found that information retention channeled into the anonymity of data.For example, one participant assumed potential access to patient data might result in personal disadvantages (e.g., a negative impact on the workplace): "For me, it is important here that under really no circumstances, on the concrete, a natural person can be inferred.[...] And the importance [of anonymity] is, I think, quite clear that just in today's [...] increasing elbow society, the admitting of personal illnesses [...] have negative vocational and also social consequences, and, therefore, [anonymity] is for me important." (P9) 4.1.4Patients should experience mutual exchange through a balanced patient-physician relationship when giving consent.Participants underlined the need to be taken more seriously, for example, by physicians during admission for medical treatment.Participants argued that a balanced relationship requires physicians' empathy, exchange on the same level, and sensitivity to patients' life circumstances or health conditions.We found that an integral element of this mutual exchange between physician and patient is communication, which one participant contemplated as meaningful interaction: "[Communication] means sustainable and profound communication.Clear communication must be conducted at all levels and through all channels.At least, that's what I found extremely important with the data donation.So, what happens with the data?Communication is not a one-time process but an ongoing one.It can also occur multilayered in different channels and depends on the corresponding [stakeholder] group that receives it." (P2) We, however, found that participants in the first workshop missed patient-oriented communication towards the moment of giving consent.Participants related to trust as a precondition for consent required between patients and physicians or hospitals.In addition, trust was seen as a concern shaped by lived experience, which might influence new situations: "I would prioritize trust at the patient and physician level and, of course, at the hospital level, but from the patient's point of view, that's where it's most direct." (P1) "I find that mistrust is interesting since I also have a professional background, and there is always trust that people get into an airplane [...].But if that is lost when an aircraft crashes, it can cost so much regarding the trust." (P2) Strengthening this assumption, one participant hypothesized that to achieve voluntariness of consent.In other words, physicians should be aware of the patient's condition in stressful situations and should not insist on consent: "[...] I had this situation in my mind where you usually sign.This is very often when you have acute problems, call up a physician's office, and then sign various consents, which is more like a doorstep transaction.Maybe having that signed more after the acute treatment.So that it's not given under duress." (P9) 4.1.5Patients should have the opportunity to consent for purpose-bound data donation, which can help increase the willingness to disclose data.We found that the participants' willingness to disclose data depends on the purpose of a medical data donation.One participant noted, for example, when consenting, uncertainty exists about what data (e.g., personal or health data) are used for which medical research purpose or whether third parties will have access: "I think that if there is a focus on a certain goal, why am I signing this now?Why am I participating in this now?Like earlier, now also [in this workshop] that we also signed [the consent form], and why this is explained in the first place: What is the concrete goal now?Then there is also greater comprehensibility and acceptance [...]." (P8) In this regard, we found that participants emphasized transparency as a key aspiration of a purpose-bound data donation accompanied by recognizable use, clarity, or privacy.One participant in the first workshop concluded: "[Transparency], the more I think about it, includes quite a lot for me.I also [included] recognizable use, clarity, and privacy.But for me, it all fits under transparency.So, clarity about the scope of the data use.Goal and purpose and transparency within the process and subsequent use.So that's what I mean by transparency.And that enables us to make decisions and have confidence in the process and the use of the data." (P3) 4.1.6Patients should be capable of making informed choices by communicating complex consent forms.There was general agreement among participants that consent forms in healthcare contain substantive barriers.For example, information about the secondary data use is buried behind paragraphs and a dense legal vocabulary.Participants emphasized explanation of domain-specific terms or multilingualism for non-native speakers.They channeled these barriers toward the low comprehensibility of consent forms.One participant argued that comprehensibility of consent forms should be ensured to mitigate concerns through the declaration of consequences:  Regarding this, we found participants stressed the practicability of how consents are realized.For example, broad consent might be incompatible with patients' concern for revocability.One participant acknowledged that it would impose an unreasonable administrative burden on medical researchers conducting multiple studies if they had to revoke consent from each patient individually: "[...] patients who would, of course, like to be informed about every study and every analysis that is to be made with their data at some point and [...] to be able to decide about [secondary data use], again and again, i.e., to be involved, and that then for the researchers this is very costly and also leads to the impracticability of studies and the administrative effort based on this information and consent." (P9) In the next section, we broaden our understanding of this section by investigating how participants used these values to envision new interactions of data donation scenarios.

Suggestions for New Interactions of Medical Data Donation Scenarios
Based on the first analysis step, we made a granular delineation of values to understand how subgroups materialized their value scenarios (see Figure 5).Each subgroup chose at least three values from the value map and considered in addition to the patients, physicians who are the first point of contact when consenting to data donation, on the one hand, and, on the other, medical researchers who provide information about medical research purposes, data processing, and data use.One subgroup, for example, created an overview of information flows that assigned responsibilities to stakeholders (see Figure 5, 01), while another decided on a timeline (see Figure 5, 02 and 04).Yet another team used color-coded sticky notes within a matrix to distinguish challenges and solutions (see Figure 5, 03).One subgroup included other stakeholders (see Figure 5, 06), namely, data protection officers, insurance companies, and healthcare providers.We noticed in this value scenario that no clear distinctions were made between the values assigned to each stakeholder.Some values (e.g., transparency and purpose) were fluidly placed across all stakeholders, making it difficult to delineate them.We found that each subgroup included further values (e.g., compliance, formatting, or control) in addition to those listed in Table 1 that had previously received less attention in phase 2 contextualize.Informed by the subgroups' discussions and reflection, we differentiate between the first workshop with patient advocates (see Section 4.2.1) and the second and third with patients (see Section 4.2.2).In the following section, we present the findings of the second analysis step.

4.2.1
Value scenarios foregrounding the patient-physician relationship.In the first workshop, we found tendencies fostering the patient-physician relationship.In both value scenarios, we found suggestions to support patients when donating data (see Figure 5, 01) and to increase the transparency of data donation processes by deploying a neutral instance (see Figure 5, 02).
The subgroup of value scenario 01 critically considered the moment of obtaining informed decision-making regarding medical data donation.They emphasized this moment as crucial to enhance the education of patients' data donation by physicians.However, the subgroup assumed that neither physicians nor hospital staff were adequately resourced to provide this educational outreach to patients.They emphasized a decision facilitator would ease the burden regarding the patient-physician relationship by foregrounding a substantial ethical stance and representing a specific set of values, such as control, compliance, and welfare: "The decision facilitator has other values.They [uphold values such as] control, compliance, and patient welfare.[...] Here, we have these ethical values." (P3) We found that the attributes of a decision facilitator oscillate around patient welfare to support a 'good' decision, i.e., on the assurance that patients are informed about medical research purposes.For example, the subgroup argued that medical researchers have to provide precise information on the medical research purpose of the research and data use and processing.This information would be forwarded and revised by a decision facilitator.Another participant echoed this assumption, assuming that a decision facilitator could support the mutual exchange between physician and patient significantly: "I found the idea [...] of a [decision facilitator], whether you call it a mediator or a consultant or an educator, quite interesting.Because I think that much more trust can be built up [in comparison to] a physician, who can sometimes be verbally off the mark." (P2) In value scenario 02, the subgroup drew similar conclusions.They chose three values as prerequisites for a patient-physician relationship, namely trust, respect, and autonomy.To reduce patient uncertainty (e.g., about how data are used and processed), the subgroup called for a neutral instance to increase transparency about medical data donation processes: "[This neutral instance] does not have to be the physician, but people who are qualified and also have the time for [a consultation]; also, an explanation about the procedure of such a data donation [...]." (P5) Another participant added that this neutral instance strives to inform patients of potential outcomes of medical research benefiting their health: "[...] patients have a right to receive feedback on the [medical] research results obtained with their data.So that if [patients] have contributed to the development of a groundbreaking new therapy, they are also informed about it accordingly [...]." (P9) 4.2.2Value scenarios dismantling patients' obstacles to donating data.We found that the value scenarios of the second and third workshops (see Figure 5, 03 to 06) aimed to reduce barriers (e.g., in consent forms).Based on personal experiences, the subgroups considered primary values such as anonymity, recognizable use, or even clarity and time.These values were enriched by additional values supporting patients' capabilities (e.g., formatting, information, easy to use, and appreciation), which can be broadly compared to the findings of Section 4.2.1.We found suggestions for making medical data donation processes more accessible for patients.For example, the subgroup of value scenario 05 aimed to mitigate formal hurdles from consent forms by emphasizing a time-independent opportunity to reflect on consent or revise consent (e.g., by an expert representing an ethics committee) to improve transparency.Regarding this, the subgroup of value scenario 04 stressed the linguistic problems of consent: "In the whole context, it was also particularly important to us in the group that the whole [consent form] is kept short and understandable and offers possibilities for plain language [...]." (P8) Similarly, in value scenario 03, the subgroup negotiated on plain language.They approached the idea of assistance (comparable to a decision facilitator, see Section 4.2.1) by alternating ways how patients might be assisted at the moment of consent: "[...] perhaps this simple language could be implemented as a kind of summary.So, a kind of data traffic light is established on the [consent forms]: green, yellow, and orange.Each explains something about the storage period, the usability of the data, and the degree of personalization." (P9) To further support patients' accessibility, the subgroup of value scenario 05 suggested visual elements or an adequate content structure: "[...] I could imagine doing a questionnaire on a tablet and just clicking, and sometimes when I have to fill out questionnaires like this: 'And now rate and do this.' So, sometimes, I would simply prefer a yes/no answer.Of course, sometimes it has to be differentiated [...]." (P13) In the next section, we synthesize our results and discuss the design requirements by reconciling participants' contextualized understanding of values and how participants materialized values in idealized data donation scenarios.

DISCUSSION
Informed by research (see Section 2), we conceptualized a workshop for participatory value elicitation (see Section 3.1) and conducted three workshops (see Section 3. 2) to answer our first research question: How can we engage patients to systematically explore, articulate, and reflect on their values regarding medical data donation?To examine the result of our workshops, we specified an analysis procedure (see Section 3.3) to answer our second research question: How can we translate the collected values into design requirements to inform medical data donation user interfaces?
To recall, our research is motivated by a particular German urgency: National initiatives, policymakers, and companies increasingly call for health research data platforms (see, e.g., [63]).These platforms collect patient data based on the provision of broad consent.The consultation process of specifying the broad consent form in Germany seems to be characterized by asymmetries of stakeholders' engagement and influence that might have bypassed patients' values.We hope our research can contribute to and inspire the future design of medical data donation user interfaces.In the following, we discuss the design requirements derived based on Section 4.2 before we reflect on the suitability of our method, i.e., workshop concept and analysis procedure, in Section 5.2.

Design Requirements That Endorse Patient Values
To deepen our understanding of participants' values and their use within the value scenarios, we (1) investigated the findings from the second analysis step of Section 4.2 and (2) revised them with the first one of Section 4.1.Based on this iterative approach (see Section 3.3, Step 2), we propose the following design requirements for medical data donation user interfaces anchored in the empirical data gathered.

Promoting reflective thinking on the scope of data donation for meaningful decision-making.
In general, research in HCI describes reflection "as reviewing a series of previous experiences, events, stories, etc., and putting them together in such a way as to come to a better understanding or to gain some sort of insight" [7].Cho et al. [15] emphasize that reflection should address various personal interests to help people harmonize their goals.In this regard, lived experiences can form the basis for reflective thinking about and planning for the future by supporting individuals to change their knowledge, behavior, and values [12].Therefore, we assume that reflection is not spontaneous but must be encouraged.
In Section 4.2.2, we argue that our findings regarding time and declaration of consequences implicitly pointed to reflection.In the participants' view, reflection should be enabled before consent to medical data donation is given: First, by detaching reflection from situations of the data donation (e.g., patient admission for medical treatment).Second, by enabling patients to reconsider personal disadvantages or potential consequences (e.g., no treatment if no consent is given).Regarding the latter, we found that reflection on consequences might reduce patients' uncertainty about how personal or health data are used and processed (see Section 4.1.1).We argue that reflection can be realized through time to reflect, helping patients comprehend the possible consequences of donating data (see Section 4.1.6).
In data-driven health technologies, we suggest that reflective thinking (e.g., [78]) can be realized by promoting patient decision-support.However, previous research highlights that realizing reflection in technology design remains a challenge [8].Inspiration for future research can be taken from "customizable designs" by allowing users to decide what data aligns with their goals based on lived experiences [15].Rohani et al. [66], for example, assume that personalized reflection can help patients with psychosomatic illnesses (e.g., depression) through visual tools.Patients can then blend personal insights into their behavior through technology-mediated reflection supporting therapeutic interventions.Such a design can support user reflection by considering their lived experiences for more meaning-making processes.
5.1.2Enabling informed consent through value-based decision-making.In healthcare, informed consent is defined as a process "whereby individuals are to be fully informed, in understandable language, about all aspects of research studies to enable informed decision making [about whether to participate in a research study]" [52].Similarly, VSD considers a lasting yet value sensitive definition of informed consent: Friedman et al. argue that to achieve informed consent, users must be provided with "the right sort of information at the right time" [28, p. 113] without any hindrance to making a 'good' decision.For example, Denning et al. [18] argue that patients expect particular values (e.g., "safety, " "psychological welfare, " "autonomy, " or "notification") when their health data is collected, used, or processed with mobile devices.
Our participatory workshops yielded similar results.Based on participants' value scenarios, we found a particular challenge to conceal various values in healthcare data practices (e.g., anonymity, recognizable use, or clarity; see Figure 5).However, as shown in Section 4.2.1, a decision facilitator might counteract this challenge by foregrounding ethical issues, revising information for medical research purposes, and supporting patients before consent We regard a decision facilitator as a potential bridge between patients, physicians, or medical researchers acting in patients' best interests.Interestingly, a decision facilitator was not necessarily considered human.We assume that a decision facilitator should enhance patients' awareness (see Section 4.1.3)to accommodate diverse information needs and to educate them about the facets of data donation; for example, that there are no consequences on treatments in the absence of consent or when consent is withdrawn later (see Section 4.1.6).We hypothesize that adequately educating patients about medical data donation practices can reduce their skepticism about medical institutions (e.g., how they use and process data), thus mitigating their privacy concerns (see Section 4.1.7).
Similarly, Gomez Ortega et al. [33] argue that individuals may be reluctant to share their data because of privacy concerns or lack of contextualized data donations (e.g., clarifying the purpose).The authors suggest that information about clear and meaningful objectives for data donation should be made available to enable people to understand their attitudes and perspectives when donating personal information.However, Franzen et al. [24] assume people's concerns about sustaining privacy arise from their attitudes and behaviors.Therefore, future work should further explore how to move to more personalized approaches that, for example, allow data donors to reflect on their viewpoints for achieving meaningful decision-making (e.g., through incentives to participate and contextual insights [33]).

5.1.3
Transparency through mediation to ensure comprehension in data donation.Previous research has explored approaches to increase the transparency of the secondary data use: For example, a value-based consent model for decreasing the administrative burden and enabling patients to translate their values autonomously into decisions for consent [32].Other research investigated hurdles of a formal and visual nature to enhance transparency (e.g., through visual icons and accompanying text descriptions) [35] or proposed interventions to stimulate considerations on privacy [20].
Compared to our results, we found that complex data practices or opaque medical research purposes compromise transparency.To counter these challenges, we suggest a neutral instance, i.e., a data intermediary (see Section 4.2.1 and Section 4.2.2), to increase transparency and control regarding patients' data donations.In contrast to the decision facilitator (as described in Section 5.1.2),who acts before consent is given, a data intermediary bundles several responsibilities acting after a data donation takes place.A data intermediary seeks to balance asymmetries of data suppliers (e.g., patients) with data collectors (e.g., medical institutions) through an alternative approach to data processing (see Section 4.1.4).An example is to centralize and standardize data suppliers' ability to manage and control the flow of personal data [44].This approach would help to achieve patients' control over their data and to favor propositions of individual values, such as "transparency, understandability and convenience" [47].In healthcare, a data intermediary can enhance patients' data ownership (see Section 4.1.3)by informing them that their data are used for a particular purpose using mechanisms like opt-in or opt-out.Such mechanisms seem necessary in situations in which the medical research purpose has to correspond with patients' values (see Section 4.1.1 and Section 4.1.5).
Regarding this, Barnett et al. [6] propose a data intermediary as support in apps that allow users to customize how their data are shared with third parties.A particular benefit of a data intermediary is to realize traceable processes of data use on data platforms [77]) to enhance digital participation.The concept of a data intermediary (e.g., for data stewardship [82]) is not entirely new as an opportunity to carefully manage data and provide a level of confidence to data providers.However, more research on data intermediaries' social, technical, and legal capabilities in handling health data is needed, especially concerning secondary data use.
In summary, we hope our design requirements emphasize useful directions from a patient perspective when designing medical data donation user interfaces.First, user interfaces should enhance patients' reflective thinking, helping them to assess the possible consequences of their data donation.Second, we emphasize that patients' value-based decision-making should be accommodated by a decision facilitator (human or non-human) that considers patients' capabilities (e.g., the need for simple language or tailored descriptions according to varying information needs).Finally, transparency should be increased through a data intermediary supporting patients' decision-making and control of their data even after data are donated.

Reflections on the Suitability of our Method
We emphasize that values should be investigated in the context in which they occur, serving as hypotheses, as argued by Parvin et al. [43].We considered Shilton's suggestion [73] to use contextual methods to determine values guided by phenomenological and inductive approaches.Before conducting the workshops, we critically reflected on our roles.Based on research (see Section 2.3), we conceptualized these roles in Section 3. In the aftermath, however, we found that our facilitator role was subject to shortcomings, which we reflected on regarding our method, i.e., a workshop for participatory value elicitation and analysis procedure.
First, the value questionnaire (phase 1 explore) allowed participants to explore their values by defining them in writing from an experiential perspective.Although we iterated the questions carefully beforehand, some participants had difficulties answering the posed questions (see Appendix A): "[The value questionnaire] was not quite so easy.[...] I believe everyone has their preconceived understandings of values" (P3).However, we indicate that the value questionnaire encouraged participants to explore their values introspectively, which they had not previously articulated explicitly.We align with Iversen et al. [41], who note that values can be partially undefinable or interwoven, which might have hindered participants from delineating their values.Furthermore, Borning and Muller [11] argue that values cannot be understood as heuristics, making it challenging for individuals to imagine values in a particular context and define them precisely.We recommend that future iterations of the value questionnaire might present examples of values and their descriptions unrelated to a workshop context to allow participants to assess the questions better.
Second, the value map (phase 2 contextualize) systematized participants' values by assigning them to direct and indirect stakeholders.Participants described the value map as a supportive vehicle to contextualize their understanding of values and negotiate incomprehension (e.g., stakeholder responsibilities regarding values assigned) within the workshop group.We were careful not to influence participants in any particular direction when creating the value map.Informed by Iversen et al. [42], we carefully reflected on our research context in advance and critically questioned our expectations.Nevertheless, we hypothesize that we implicitly influenced participants through our intrinsic concern to improve medical data donations.For example, we specified stakeholders (e.g., physicians and medical researchers) as starting points that might have affected participants' stances.In future workshops, participants should figure out on their own, i.e., without the assistance of a facilitator, which stakeholders they consider important in a given context.
Third, the value scenarios (phase 3 translate) showed participants' perspective of an idealized situation in a specific context (see [61]).Echoing Leong and Iversen [50], we argue that value scenarios supported participants to think about possible alternatives.We further found that the values scenarios enabled introverted participants to participate in constructive discussions.At the beginning of the activity, we showed generic scenario examples (e.g., narratives, sketches, or a combination of both).Some subgroups used these examples to create the 'right' value scenario.Other participants were inhibited, especially in the second and third workshops, fearing they would do something wrong when not following our examples.We repeatedly reinforced that these examples were only meant to serve as inspiration.We further provided prototyping materials like cardboard, Styrofoam, or small building blocks to unleash participants' ideas and creativity.Grönvall et al. [34] inspired us to create a space for equal participation.However, we found that the subgroups of the second and third workshops with patients in inpatient care expressed similar ideas, i.e., two-dimensional representations.We attribute this to the spatial situation, which allowed ideas to be negotiated across subgroups that might be eventually adopted.Future research should consider spatial separation of subgroups to prevent influencing each other.Furthermore, research should examine whether an activity description of a value scenario is sufficient without offering visual examples to enhance participants' storytelling, which could lead to outcomes such as three-dimensional representations (e.g., [64]).
Finally, we emphasize that our workshop concept enabled participants to recognize the ambivalence of values.Based on our analysis, we found that participants continuously negotiated values across the four workshop phases and regarding the research context.One participant argued that "there is a certain ambivalence as to whether I now formulate my fundamental values or values about data donation" (P4).According to Iversen et al. [40] and Frauenberger et al. [25], we intensively engaged with our concern to develop a consistent method that helps to work with values of vulnerable groups of patients.However, regarding participants' lived experiences, we assume they expressed very emotional viewpoints of their values in the face of exhausting admission processes and physically demanding medical therapies.We attempted to address participants' values equally, such as allowing introverts to partake in our activities.Yet, we encourage future research to focus more on individual values.In other words, values illuminate personal experiences and are closely linked to them, thus offering valuable insights for technology design.
In summary, we hope our method will serve researchers and practitioners as a resource to elicit values through participation and to analyze the empirical data gathered.By contributing to research on value-led PD (see Section 2.3), we believe that our proposed workshop concept allows participants to explore their values creatively, negotiate them continuously by contextualizing, and finally, materialize their values for an idealized situation.

LIMITATIONS
We conducted our workshops in the healthcare domain by focusing on patient values.Due to our resources, we facilitated only three workshops; thus, we doubt that we have already reached theoretical saturation [68].This doubt was particularly evident in the unforeseen circumstances we faced while preparing for the workshops, hindering us from realizing a large study.We sought adequate representation of patients by relying primarily on the support of our research network.At the time of our workshops, the COVID-19 pandemic was still ongoing, so we had to postpone the workshops several times for the following reasons: In the second and third workshops, there were concerns from our medical partner facility regarding patients' health safety (e.g., direct contact with externals).Furthermore, our contact person, i.e., the physician responsible for recruiting patients in inpatient care from the psychosomatic unit, was absent for an extended period due to a COVID-19 illness.Participants also canceled due to illness or worried about participating in a workshop with multiple participants, which put a strain on achieving an adequate sample size.These circumstances and the approaching end of the research project led us to conduct three workshops.
We have experienced being more emphatic and caring, especially when working with vulnerable and stigmatized groups suffering from mental disorders or physical impairments.As mentioned, we were restricted in not having the opportunity to build a deeper interpersonal relationship with the patients in inpatient care over a longer period.An ongoing process of participation and sustained openness among participants [54] would likely have reduced initial barriers between patients and us (researchers) and, thus, given patients more confidence in accomplishing the activities.Furthermore, patients' residential care might have resulted in activities being pressured and burdened by individual suffering.Although we were sensitive to encouraging these participants to express themselves freely, our efforts might have influenced their responses and actions.
Comparing our workshop cohorts, we assume that patient advocates brought more qualified insights to the workshop due to their occupation despite their lived experiences (e.g., personal medical histories).We argue that scrutinizing only patients in inpatient care may have led to different results.In order to obtain a more inclusive sample, we assume that participants in different life situations (e.g., full-time work, caring for family members) should have been enabled to participate in our workshop.One promising direction might be to externalize the exploration of values (phase 1 explore) from the workshop process into participants' everyday lives (e.g., [9]).Despite these limitations, our studies have taught us to consider our research more as a personal concern and acknowledge the appreciation for and participation of vulnerable groups, giving our research project a deeper meaning (see, e.g., [23,55]).We have learned to value encountering vulnerable people.This has shaped our research findings and how we as researchers must act to give voice to those who otherwise go unheard.We acknowledge the efforts of our participants and the medical partner facility that supported us in conducting at least three workshops for participatory value elicitation.We hope the detailed documentation of our method and results will enrich future research endeavors.

CONCLUSIONS AND FUTURE WORK
In this article, we foregrounded patients' values in medical data donation processes through active participation.Informed by PD, VSD, and research on value-led participation, we introduce a method consisting of (1) a workshop concept for participatory value elicitation and (2) an analysis procedure to examine the empirical data gathered for specifying design requirements to improve user interfaces for medical data donation.Our workshop consists of four sequential phases and comprises various techniques, i.e., value questionnaire, value map, and value scenario.Our analysis procedure supported us in understanding better, first, participants' contextualized understanding of values and, second, how participants materialized values for an idealized data donation scenario.
Our results suggest design requirements for medical data donation user interfaces that strongly incorporate patient values: First, user interfaces should enhance patients' reflective thinking on the potential consequences of their data donation; second, a decision facilitator that foregrounds an ethical stance to induce patients' value-based decision-making.A decision facilitator (human or non-human) helps reduce hurdles and skepticism in healthcare data practices by providing simple language or tailored information based on patient needs; finally, a data intermediary that increases transparency by supporting patients' decision-making and control even after data are donated.We argue that our design requirements can enhance sociotechnical designs (e.g., of data-driven health technologies) beyond a legal and formal ethical perspective.To evaluate our design requirements, we plan to consult different stakeholder groups, especially members of ethics committees for medical research, data protection authorities, and experts in medical informatics.We also intend to use our design requirements as input for a co-creation workshop with patient advocates and experts from privacy law, medical ethics, and interaction design to examine their feasibility in the context of medical data donation user interfaces.
Regardless of our research, we hope our method will inspire researchers and practitioners alike to apply it in their respective contexts.We, furthermore, envision that policymakers will consider more participatory approaches when defining patients' rights in the future.

Fig. 1 .
Fig. 1.Concept to facilitate workshops for participatory value elicitation involving phase 1 explore, phase 2 contextualize, phase 3 translate, and phase 4 reflect, accompanied by the activities, namely value questionnaire, value map, and value scenario.

Fig. 2 .
Fig. 2. Each participant was equipped with a a material box to realize the four workshop phases.Besides a workshop leaflet, each material box contains b sticky notes in three different colors, i.e., yellow for values, pink for stakeholders, and green for value conflicts or commonalities (for phase 1 explore and phase 2 contextualize), c tape (for phase 2 contextualize), and d various writing and drawing utensils.

Fig. 3 .
Fig. 3. Overview of the analysis procedure divided into two sequential steps, namely (1) how participants' contextualized understanding of values and (2) how participants materialized values for an idealized scenario.

Fig. 4 .
Fig. 4. Value maps derived from phase 2 contextualize of the first (left), second (middle), and third workshop (right).In a moderated session, participants created a value map by articulating and assigning their values from phase 1 explore to stakeholders.

4. 1 . 7
Patient skepticism towards opaque medical data donation should be reduced by distributing responsibilities in healthcare data practices.Healthcare data practices face several hurdles, for example, how and for what patient data are used and stored or whether data are transmitted to third parties.We found that participants expressed these challenges to the healthcare domain as value conflicts among the stakeholders affected.One participant recalled challenges to addressing patient values in healthcare data practices: "[...] everything that concerns the values and this data donation is, first of all, an analytical process, so we analyze and what do we get donated by person A and person B? That is the easiest.The next step is how we interpret it, how we deal with it, and the very last step is what consequences we draw from it [...].And that's why I think it's quite understandable that in this internal relationship, that is, in the immediate area [...].(P5) Furthermore, we determined participants were particularly critical of equality being established in healthcare.One participant assumed that the distribution of responsibilities in healthcare remains unclear, resulting in one-fits-all consent forms: "[...] And I don't know, since at the end [the consent form] is, in case of doubt, approved by the hospital [...] and not the individual physician, who then hands it over[...].And in the end, [hospitals] are only the governors for all the [medical] researchers who are in the background." (P1)

Fig. 5 .
Fig. 5. Overview of values used (left) that informed participants' materialization for an idealized data donation (right) of the first workshop for value scenario 01 and 02, second for value scenario 03 and 04, and third for value scenario 05 and 06.Values are sorted by their appearance in the transcripts.

"
And [the medical researchers] have to say what they want to do, for what purpose, and only then do they get permission to access [patient data]." (P3)

Table 1 .
Overview of the values of the three workshops based on the first step of the analysis, showing the values, i.e., codes (left) and their frequency of occurrence based on the transcripts (middle) that informed each category (right).The values are color-coded, i.e., whether they were generated in the first, second, or third workshop.
"[...] often there's no time to understand [consent forms].And what is important here, [...] is that you can take [the consent form] home and have the time again to discuss it with someone who can explain it to you.So, it is mainly about having someone to explain it to you [...], that you don't just get it and have to sign it right away, but that you have the opportunity to reflect on it and think about it." (P15) And, yes, [declaration of consequences] mean for me, what are the consequences for me if I don't sign [a consent form].Then, also the fear that I won't be treated if I don't sign.[...] So, the consequence is not telling exactly what happens if I do it or what happens if I don't.That's already a big decision for me, whether I sign or not.Most of the time, you sign because otherwise, you have the feeling it won't go any further here." (P16)