Case Study: Visualizing Bipolar Disorder
GROUNDED VISUALIZATION DESIGN CASE STUDY
Researcher and Practitioners
VSRS team
Jaime Snyder, Project lead
Caitie Lustig, Graduate student
James Kitchens, Graduate student
Justin Petelka, PhD student
Liam Albright, Undergraduate student
Nikita Nerurkar, Undergraduate student
Collaborators
Elizabeth Murnane, Dartmouth University
Iris Gottlieb, Graphic Artist
Stephen Voida, University of Colorado, Boulder
Tara Walker, PhD student, University of Colorado, Boulder
Community Partners
National Alliance for Mental Illness – local chapters
Depression and Bipolar Support Alliance
Application Domains
Personal data visualization, bipolar disorder, self-tracking tools, quantified self, visual methods, human-centered computing, empirical studies in HCI, collaborative and social computing design and evaluation methods
Project Abstract
This study investigated the visual metaphors and images that people with bipolar disorder associated with their condition in order to better understand how personal health data could be visualized in more meaningful ways, through apps or other forms of technology. Through interviews and participatory design activities, this study explored the gaps, tensions, and intersections between traditional data visualization conventions used in many popular tracking tools and the visual metaphors and images used by individuals currently under treatment for bipolar disorder to describe their mental health experiences. This study (1) supports patient-centered approaches to the long-term management of serious mental illnesses, (2) informs the design of self-tracking tools for specialized audiences, and (3) provides insights into the role of visualization in shaping data-driven knowledge.
Related publications
Snyder, Jaime. “Visualizing personal rhythms: A critical visual analysis of mental health in flux.” In Proceedings of the 2020 ACM Designing Interactive Systems Conference, pp. 269-281. 2020.
Petelka, Justin, Lucy Van Kleunen, Liam Albright, Elizabeth Murnane, Stephen Voida, and Jaime Snyder. “Being (in) visible: Privacy, transparency, and disclosure in the self-management of bipolar disorder.” In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, pp. 1-14. 2020.
Snyder, Jaime, Elizabeth Murnane, Caitie Lustig, and Stephen Voida. “Visually encoding the lived experience of bipolar disorder.” In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems, pp. 1-14. 2019.
Murnane, Elizabeth L., Tara G. Walker, Beck Tench, Stephen Voida, and Jaime Snyder. “Personal informatics in interpersonal contexts: towards the design of technology that supports the social ecologies of long-term mental health management.” Proceedings of the ACM on Human-Computer Interaction 2, no. CSCW (2018): 1-27.
Matthews, Mark, Elizabeth Murnane, Jaime Snyder, Shion Guha, Pamara Chang, Gavin Doherty, and Geri Gay. “The double-edged sword: A mixed methods study of the interplay between bipolar disorder and technology use.” Computers in Human Behavior 75 (2017): 288-300.
Matthews, Mark, Elizabeth Murnane, and Jaime Snyder. “Quantifying the Changeable Self: The role of self-tracking in coming to terms with and managing bipolar disorder.” Human–Computer Interaction 32, no. 5-6 (2017): 413-446.
Snyder, Jaime, Mark Matthews, Jacqueline Chien, Pamara F. Chang, Emily Sun, Saeed Abdullah, and Geri Gay. “Moodlight: Exploring personal and social implications of ambient display of biosensor data.” In Proceedings of the 18th ACM conference on computer supported cooperative work & social computing, pp. 143-153. 2015.
Funding
Jaime Snyder (PI), “Visualizing Bipolar Disorder: Exploring the Representation of Personal Data in the Treatment of Serious Mental Illness,” Sponsored by the UW Royalty Research Fund (RRF, #65-6521). $34,766. 02/01/17—09/15/18.
Jaime Snyder (PI), “Collaborative Design: Personal Data related to Serious Mental Illness,” Sponsored by Group Health Foundation: Lessons Learned from Engaging with Communities, $7,500. 6/2018.
Introduction
How well did I sleep last night? How relaxed am I right now? Will I feel better or worse tomorrow?” In the current digital age, we can expect these questions about our personal health and well-being to be answered with data. Knowing oneself through personal data has been enabled by recent work in Personal Informatics (PI) that has improved the instruments, methods, and algorithms that support self-tracking.
Visual displays of personal information play a critical role in self-tracking, as they are often the ways that PI systems present personal data to end users. Visualizations provide opportunities for individuals to know about themselves and their communities and support the interpretation of large quantities of information even if individuals are not trained in data science or statistics. The design choices made in the visual encoding of personal data influence how data-driven self-knowledge is shaped.
To probe the limitations and potential biases of visual conventions in the representation of personal data and to identify alternative approaches that better align with lived experiences, this study examined an “edge case” in personal tracking. The edge case in question is individuals self-tracking to manage their bipolar disorder (BD), a serious mental illness characterized by severe and unpredictable mood swings i.e., episodes of depression, mania or hypomania, and mixed states.
In addition to the various charts, spreadsheets, forms, and checklists that clinicians provide, many individuals managing BD also use self-tracking apps. In this context, how these PI tools visually represent personal data to users is critical for supporting healthy self-reflection, a sense of agency, and effective communication with care providers. However, popular applications like Fitbit offer a distinctly quantitative representation of an individual by displaying personal data through time-series graphs and line charts. In highlighting regularized patterns using standardized baselines, there is a risk that representations of self that emerge from these visualizations can foster unhealthy self-scrutiny and unrealistic normative expectations of health. This study investigated the visual metaphors and images people with bipolar disorder associated with their condition in order to better understand how personal health data could be visualized in more meaningful ways, through apps or other forms of technology.
Fourteen people took part in our study (9 female, 5 male; average age = 45.9; age range = 20–64). Participants self-reported 1) being over the age of 18; 2) having a diagnosis of BD (Type I N=9, Type II N=3, Type Not Otherwise Specified N=2); and 3) not having been hospitalized for mental health issues within six months. Participants were recruited through local chapters of the National Alliance for Mental Illness (NAMI) and the Depression and Bipolar Support Alliance (DBSA), campus health care clinics, and email lists. Participants were also invited to share the study with their personal social networks. All 14 participants completed Interviews 1 and 2 and 11 returned for Interview 3, marking a high retention rate throughout the study, despite this population’s high risk of instability and rate of life transitions. Interviews were scheduled for one hour, and most took place in a faculty office on campus.
Phase One. Context Study
| Aims | Learn about participants’ experiences with BD, including formal and informal self-tracking practices. The visual elicitation activity explored the expressive capacity of a basic visual form: the line. |
| Activities | Recruited participants via community organizations – local chapters of the National Alliance for Mental Illness, Depression and Bipolar Support Alliance, campus health care clinics, and email lists. Conducted screening interviews over the phone Interviewed participants for one-hour sessions with one visual elicitation activity. Visual elicitation activity 1 asked participants to draw lines that represented their experiences with BD Analyzed data generated from this phase through inductive social semiotic analysis |
| Outcomes | Visual motifs surfaced through inductive social semiotic analysis of data generated from context study and co-design activities. Able to show that people untrained in data representation, or without specific professional data skills, use basic visual elements to represent change over time. Visual elicitation activity surfaced narratives and enabled participants to reflect on their experiences. |
Details
Following a screening interview over the phone, we met with each participant for a one-hour interview session (Interviews 1), which focused on learning about participant experiences with BD, including formal and informal self-tracking practices. The session was audio recorded with permission and consisted of a semistructured interview and a visual elicitation activity.
Visual activity 1: Lines. The first visual elicitation activity, during Interview 1, explored the expressive capacity of a basic visual form: the line. We asked participants to draw a single line that represented their experiences with BD. We then asked them to add a second line to their drawing that represented someone close to them over the same time represented by the first line. Someone close could be a family member, a friend, a therapist, or even a pet. Last, we prompted participants to add a third line that represented what they would have considered an ideal state during that period. Examples of output from this activity are shown above. We asked participants clarifying questions, including whether specific visual features had particular meaning to them. For example, “I see this part of the line as really bumpy, but it straightens out over here. Does that represent a specific event? How would you describe the differences between those times?” At times, this resulted in augmentations, additions, or corrections to the lines.
Data generated from the context study included audio recordings, transcripts, PDFs of digital drawings, and screen recordings of drawing activities performed on the digital tablet. Some participants also shared examples of personal tracking documents and artwork, which we photographed with their permission.
Phase Two. Co-Design
| Aims | Explore how participants mapped complex notions such as change over time to basic visual forms. Understand how participants explicitly and implicitly assessed their state of being over time, and explore ways of visually documenting these changes. |
| Activities | Interviewed participants for a one-hour session with visual elicitation activities. Visual elicitation activity 2 asked participants to create a timeline that represented their experiences with BD in terms of transitions and changes over time during interview one. Visual elicitation activity 3 asked participants asked to annotate a timeline created in visual elicitation activity 2 with a set of five suggested icons during interview two. Visual elicitation activity 4 asked participants asked to find existing online images corresponding with their experiences of BD during interview two. Collaborative analysis with a professional graphic artist familiar with BD. |
| Outcomes | A range of visual metaphors and encodings surfaced from participants, some departing from linear representations entirely. Three speculative visual schemes for encoding personal data were developed collaboratively with a graphic artist. |
Details
Visual activity 2: Timeline. The second visual elicitation activity, also administered during Interview 1, required participants to think about their experiences with BD in terms of transitions and changes over time by creating a more detailed timeline. “Timeline” was defined as a diagram showing a sequence of events and how they relate to each other. Participants were encouraged to freely interpret this task. This activity explored how participants mapped complex notions like change over time to basic visual forms. A range of visual metaphors and encodings were used, some departing from linear representations entirely—for example, exploring circular or pictorial ways of illustrating change (see examples above). We again asked clarifying questions and annotated a digital copy of each timeline to note specific intended meanings.
Visual activity 3: Icons. For the third visual elicitation, performed during Interview 2, participants were asked to use a set of five suggested icons to annotate the timeline diagram they created during Interview 1. These icons represented a merging of concepts derived from conventional approaches to visually encoding time series data with ways that participants had talked about personally assessing their behaviors and moods over time during Interview 1. Participants were invited to modify or substitute these symbols as they saw fit, with the symbol system serving as a probe to inspire reflection about how participants explicitly and implicitly assess their state of being over time, and to explore ways of visually documenting these changes.
Visual activity 4: Photo elicitation. The fourth visual activity introduced a photo-elicitation task. Participants used an “incognito” instance of a Chrome browser window to access the Google Image Search interface to find existing online images corresponding with their experiences of BD. Participants were prompted to think about concepts and descriptions that had surfaced during previous discussions, and to use associated words and phrases as search terms, for example: “boundaries”, “split road”, “fries”, “ocean”, “penguins huddling”, “bear hibernating”, “lichen”, and “white noise.” Images were downloaded for future analysis.
Data generated from the co-design study included audio recordings, transcripts, PDFs of digital drawings, screen recordings of drawing activities performed on the digital tablet, and collections of online images. Some participants also shared examples of personal tracking documents and artwork, which we photographed with their permission.
Phase Three. Validation for Triangulation
| Aims | Whether and which parts of the visual encoding schemas represented in the professionally-developed sketches were useful or valuable to participants. |
| Activities | Conducted interviews with participants. Introduced participants to visual themes identified using a set of representative images and asked them to provide feedback on images and concepts. Asked participants to imagine how speculative design concepts, i.e. visual schemas, could or should change in response to their personal self-tracking practices. |
| Outcomes | Thematic grouping of images from social semiotic analysis validated. Identified sense-making challenges related to managing BD. |
Details
Between the second and third interviews, materials were analyzed collaboratively with a professional graphic artist familiar with BD. Analysis was inductive and followed social semiotic principles, focusing not only on what is represented in an image (the pictorial content) but also the communicative context in which it was created (the discursive meaning) and what is and can be done with it by specific audiences (the cultural significance).
Social semiotic analysis surfaced a set of visual motifs associated with the pictorial content of images and a set of sensing-making challenges related to the practices of interpreting personal data. Three speculative visual schemas for encoding the personal data of people with BD were created in collaboration with a professional graphic artist familiar with BD.
Interview 3 focused on determining whether and which parts of the visual encoding schemas represented in the professionally-developed sketches were useful or valuable to participants. Because several months had passed between the second and third interviews, the latter also provided insights into the stability of visual associations over time.
We began by asking if participants had experienced any significant changes or life events. We then revisited narratives and imagery from the first two interviews. Participants who returned for the final interview clearly remembered creating and talking about specific images and, in spite of some variation in mental state (e.g., feeling more or less depressed or manic or being at different points in therapy), all participants confirmed that most images still carried substantial meaning.
Next, we introduced participants to the visual themes we identified using a set of representative images from data collected during Interview 2. These images were shown to participants during the interview, and participants provided feedback about their affinity for these images and concepts, providing validation for our social semiotic analyses.
Last, we presented each participant with the three speculative design concepts and asked them to respond generally to all three, but in the interest of time, to select one to evaluate in more detail. Participants were prompted to imagine how the visual schemas could or should change in response to their personal self-tracking practices. Following our social semiotic approach, we encouraged participants to not only respond to the pictorial elements of the schemas but also to consider how each visual encoding system enabled different relationships to be surfaced and different types of comparisons to be made among potential personal data points.