Family Health Informatics

Understanding Sleep Tracking in the Context of Family Health

March 2016 - Present

As an independent research project, I collaborated with Dr. Laura Pina, a researcher at the UW Computer Science Department, since its beginning in Spring of 2016. Throughout this project, I have learned and applied the full user-centered design cycle in the rigorous and ethical setting of academic research. I studied how health, specifically sleep, changes in context when viewed from the perspective of family dynamics. I worked with eminent professors in the field (Drs. Julie Kientz, Sean Munson, and James Fogerty) as we carefully crafted this project in the emerging area of study in the intersection between collaborative family health and personal informatics. I am honored to have co-authored a CSCW paper with them.

Currently, we are working on a paper set for CHI 2019, wherein we will elaborate on our findings from a deployment of a sleep tracking probe used by families. The probe was the result of a year of design and development iterations, where I was the lead designer and front end developer.

 

Project Overview

The design of health tracking technologies typically focuses on individual self-tracking and self-management, not yet addressing family health in a unified way. But in families composed of parents and children, the health of individual members is often interrelated: the health of children can have an impact on the health of parents, and vice versa. Specifically, this project explores how sleep affects and is affected by family dynamics and what are the implications of designing for tracking at the family level.

 

In this project

  • We first laid the discovery work examining opportunities for family-centered health informatics (Spring/Summer 2016)

  • Then, we iterated on designs and development of a family sleep tracking probe (Fall 2016 - Summer 2017)

  • And finally, we deployed the probe in Fall 2017 - Winter 2018. We are currently analyzing data from this deployment.

My Role

From Spring of 2016 to the present, my role in this project has evolved from researcher to designer/developer and back again to researcher. These 3 phases showcased the broad range of my skillset, which also includes managing and adapting to teams of fluctuating members.

Phase 1: Examining Opportunities for Family-Centered Health Informatics (Spring/Summer 2016)

To understand the design space of collaborative family health (sleep) and personal informatics, we conducted discovery research which led to design implications for a health tracking tool tailored to the unique contexts of family dynamics. The work was based on a study by Dr. Teresa Ward who had previously looked into the sleep habits of children suffering from Juvenile Ischemic Arthritis (JIA), a chronic condition where sleep is a major concern.

 

For this phase of the study, my responsibilities stemmed from being a qualitative researcher:

  • Recruited families from a participation pool maintained by UW and UW Medical Center

  • Planned and wrote the protocol and interviewed 24 families at their homes

  • Participated in large scale coding for analysis

  • Co-wrote the CSCW 2017 paper

Research Methods

Contextual Interviews

Our chief elicitation method used for this project was contextual interviews with 24 families at their homes. Because Family Health means an interconnection of the health of the individuals as a family unit, we stipulated that all members be present for the interview. We also looked into how chronic illnesses can affect family dynamics: 10 of the 24 families had children who suffer from JIA. In particular, we explored how these families view, discuss, and do with regard to sleep.

 

Participatory Design with Children

To further understand sleep and sleep tracking from the perspectives of children, we contacted Prof. Jason Yip at the School of Informatics and took part in 3 design sessions with his KidsTeam UW (Six 7-10 y.o. children). We gained insights into their conceptions of health technologies and how they perceived health in the context of their respective families.

 

Literature Review

Background research consisted of articles in HCI devoted to family collaboration and technologies as well as personal informatics (i.e. tracking tools for health). As this was and still is an emerging area of study in CSCW and CHI, we looked into all available aspects of family collaboration and tool use.

Findings & Design Implications

Using Grounded Theory to analyze our data, we uncovered themes central to the understanding of family health informatics:

 

  • Family Members Collaborate to Manage Children’s Health: usually one parent becomes the primary caregiver, while the other parent or older siblings take on more supportive roles. Viewing tracking and monitoring as a collaborative practice means that a family-centered health tool would need to accommodate the dynamics of multiple roles within the family.

 

  • Revealing Ripple Effects Between Family Members: Both adults and children expressed wanting to understand how the behavior of one family member affects others. Engaging the entire family in tracking means going beyond individual tracking to leverage family data for comparison and to identify such ripple effects.

 

  • Need to Know Why, Not Just What: family cohorts wanted to move away from raw quantitative information to more qualitative insights, wanting data representations to be more high-level, easily interpreted, and more actionable.

 

  • Sharing Data within a Family Raises Privacy Concerns: Tensions that emerged stemmed from privacy concerns about sharing too much information with the family. Older children as well as divorced parents were particularly sensitive to the idea that tracking sleep can reveal more than sleep information.

Family informatics technologies need to account for the different needs of family members (e.g., differences in the amount of sleep for adults versus children), and should take advantage of shared goals and contexts to support family members in tracking on behalf of each other. Expanding existing models of personal informatics to family-centered informatics is a good place to start, but a family-centered tool needs to accommodate coordination and collaboration of individuals as they work toward collective and personal goals.

Reflections

As my first project as a qualitative researcher, I learned many techniques and theories from Drs. Pina, Kientz, and Munson - especially on coding as dictated by Grounded Theory. Coding a huge amount of data (from 1+ hour long interviews with 24 families) was also a big eye-opener for me. Another learning opportunity came in the form of logistics for conducting family interviews. To get at the "contextual" in Contextual Interviews, we planned for the setting to be at their place of residence. But for the sake of our safety, all researchers on the team had to "buddy up." It's an interesting reminder that even academic researchers in their protective environment of their institution need to face the realities of face-to-face interactions with participants.  

Phase 2: Design and Development of “DreamCatcher” (Fall 2016 - Summer 2017)

Based on findings from Phase 1, I was tasked with creating a design for visualizing sleep data that would best serve both children and adults. Iterations on the design involved going back to KidsTeam UW to gain insights from the perspective of children and also recruiting families to evaluate the design from within the context of family dynamics. I used Sketch and Adobe Illustrator to build mockups and information visualization of different features.This iteration did not end by the time development of the probe began, as we wanted to ensure the relevancy of the design for what it was intended, a probe that can give us an understanding of how sleep tracking can operate as a collaborative family tool. The following screens show the initial designs:

 

 

 

 

 

 

 

 

 

 

 

 

 

The challenge of designing for a family was how to convey information visualization of sleep data that makes sense to both young children and adults and to limit the tensions with privacy that we found in the previous phase of the project. To simplify what can be a complicated and tension-filled probe deployment if teenagers were involved, we opted to constrain the age range of the children to 5-12 years of age. Still, juggling the somewhat opposing values of not overloading the cognitive capabilities of young children while at the same time engaging the attention of adults was challenging.

What Kids Want

Evaluations from KidsTeam UW yielded findings that children wanted the ability to control what they see and that the screens were not “boring.” This included giving them the ability to choose from among multiple avatars and seeing design elements that evoke whimsy and fun.

Development

After extensive review of data visualization tools, we decided to use Chart.js and Fitbit API with Angular and Bootstrap to visualize sleep data and the Tractdb database to store the data. Fitbit was chosen as the tracking tool because it provides developers with a public API. In a team of three, where I was lead designer and front end developer, we built DreamCatcher in Spring/Summer of 2017.

 

For DreamCatcher to function as a probe, we designed for interactions that can provoke discussions within the family as well as for interactions that would allow them to report to us directly about their level of engagement (answering prompts). Our prototype evaluations with children and families showed that just by having the sleep data of all family members on one screen, without having to log in or out to view individual data, allowed for more discussions on the health of the family as a unit instead of concentrating on the self. We showed data at the daily and weekly levels as well as family and individual levels.

Family Data All in One Place

Individuals must enter how they feel before they can view their sleep data for the day because prior studies on sleep have revealed that mood can impact how one has slept. Trends of all family members’ sleep can be viewed on a weekly basis, allowing the family to connect sleep patterns with one another.

Prompting Reflections

Individuals or the whole family can record their thoughts based on prompts that pop up from time to time. This encourages discussions between family members while at the same time, giving data to researchers.

Individual Views

Individuals can drill down to the components of their own sleep, on a daily or weekly basis. These screens are not restricted to just the respective person; the whole family can view anyone's data if they choose to.

Reflections

We conducted multiple evaluations of our probe, and even went so far as to make adjustments only days leading up to deployment. Family, friends, and coworkers all contributed by testing whatever working version we had on hand. As such, the work was exhausting, yet when I saw children enjoying the fruits of our hard work, it was a gratifying feeling. Though, I have to say that as a researcher, designer, and developer in this phase, there were times where I felt the tension from assuming all 3 roles.

Phase 3: Deployment of “DreamCatcher” Probe (Fall 2017 - present)

We recruited 10 families for a study lasting 6 weeks, during which time the families wore Fitbits at night and interacted with the probe during the day. The data we captured during the study included: individual sleep data, audio logs, mood reports, who viewed whose data, and the number of times individuals viewed their own data. At the end of 6 weeks, we conducted an in-depth interview with the whole family.

 

Analysis

With a goal to submit a written article to CHI 2019 (deadline September 2018), we proceeded to analyze the wide range of data first using the top-down approach based on findings from Phase 1 and then going over the data again with a bottom-up approach. We grounded the analysis by keeping in mind theories in Family Health, namely the Bowen Family Systems Theory and the idea that artifacts can be boundary-negotiating.

 

We found that

  • Tracking can be a family affair. Children not only became involved in learning about their own sleep but also those of their parents and siblings. This sometimes led to discussions on the overall health, not only sleep, at the family level.

  • Making sense of the data can be difficult when families feel that they are not actionable. But we found that increasing knowledge of others’ sleep raises empathy and general awareness that can factor in how they treat each other the rest of the day.

  • The success of the study hinged on the intra-family boundaries that occur naturally. Parents who did not view sleep as a priority passed on that stance to their children, and individuals who were reluctant to openly discuss their sleep with others hampered family coordination.

  • Individuals sometimes assume the responsibility of the entire family, which can cause undue stress on that individual. Fostering collaboration can ease the stress of these individuals.

Reflections

A learning opportunity from this phase of the study came from a sensitive situation when interacting with a family. Having never encountered such a situation before, it forced me to think deeply on empathy as a researcher. I'm deliberately vague as to the nature of the interaction to protect the anonymity of the family, but I hope in the future I will keep being mindful of different socio-cultural cues and the power dynamics at play between participant and researcher.