Projects

Overview: I am the lead researcher on the development of health visualizations for older adults. I apply a user-centered approach to iterate on visualization designs of smart home data. This first involves taking data from home based monitoring technologies and aggregating them into measures of wellness. The information is then presented as visualizations for the older adult stakeholder. Using focus groups, user interviews, and usability studies, I refine the designs from early mockups to interactive prototypes. I play the role of designer, developer, and user researcher in this project.

Methods: User-Centered Design, Focus Groups, User Interviews, Usability Testing

Impact: Most of the work in smart home visualizations focuses on presenting information back to health care providers or researchers. This is the first time that there has been an older adult focus, representing a distinct information need from that of other stakeholder groups. The visualizations are an important resource to empower older adults in managing their health and wellness while coming full circle in demonstrating the utility of a smart home environment.

Overview: To inform the design of health visualizations, I led a study to examine how older adults' graphical perceptual needs differed from those of the general population. I modeled the study after a classic psychophysics experiment by Simkin and Hastie (1986). I developed the tasks as an online survey using HTML, JavaScript, and PHP and sampled over 200 participants to compare performance across age groups. Participants were asked to make proportional and comparison judgments on segments of graphs representing different basic graphical elements.

Methods: Survey Design, Web Programming, Survey Analysis

Impact: Though cognitive graphical perception studies abound, they were conducted primarily within a university setting. Therefore participants represented a skewed population. This was the first time where we examined graphical perception with older adults. In comparing with the general population, we found older adults took longer but were on par in accuracy of graphical perception. We also found moderate differences in performance across graphical elements.

Overview: I led a project to evaluate navigation within a multifunctional wellness tool for older adults. The tool was an interface that consisted of various applications nested in a hierarchical structure. However it was unclear how intuitive it would be to navigate within the hierarchy. To test this, I developed an online tree test that tracked the navigation within an abstract menu representation of the wellness tool. This was compared with traditional in-person usability studies.

Methods: Usability Testing, Online Tree Testing, Web Programming

Impact: We demonstrated that tree testing provided a flexible way to evaluate navigation of health systems comparable to usability studies. This approach had not been applied for health systems previously. The findings led to a reorganization of the information architecture for the wellness tool.

Overview: I was co-lead investigator responsible for integrating Microsoft's Lab of Things platform into an applied smart home environment. The platform provides a central hub to manage deployment of sensor systems across homes. As a use case, we deployed sensors across three different home environments and used the data to create a real-time visualization application.

Methods: User-Centered Design, Programming

Impact: The pilot study consisted of a two month deployment across three settings. We have extended the pilot study and are now implementing smart home sensors over a six month period in older adult homes.

Overview: I conducted first impression testing to compare 5 different variants of a web page design. It was unclear which of the designs would promote the most engagement with consumers. I approached this from both an empirical and qualitative standpoint. In Qualtrics, I developed a survey that tracked first click on a static page. I complemented this performance data with an aesthetics questionnaire. This allowed me to measure how effectively participants engaged with the designs and their perceived appeal of the designs. For recruitment, I used Mechanical Turk to collect data based on a pre-screening qualification.

Methods: Survey Design, Qualtrics, Mechanical Turk, Statistical Analysis

Impact: From a budget of $500, I was able to get high volume response (n=823) for a short turn around time (1 week). The results helped identify which designs performed better and were perceived of as more appealing. This was also the first time anyone had pioneered the use of Mechanical Turk as a resource for research within AWS.

Overview: To better understand where older adults go to when searching for health information and the differing levels of trust that older adults place in each resource, I co-led a survey with Shomir Chaudhuri. We designed and implemented a paper survey distributed to 1,500 older adults in retirement communities throughout the Seattle area.

Methods: Survey Design, Statistical Analysis

Impact: We found that though older adults have access to various forms of health information they prefer in person communication the most. Even within older adults we found age group differences with a sharp distrust of Internet, television, and radio resources.