top of page

Researching trust in EHR game-changer

Dell Medical School, The University of Texas at Austin // 2021

mobile phone and stethoscope
SUMMARY

Do users trust blockchain technology when it's used to manage personal health information? Almost half show concern about security & privacy.

Researchers at The University of Texas at Austin were designing an app that enabled patients to manage their electronic health records (EHRs) and securely share that information with healthcare providers via blockchain technology. This addressed the issues of siloed records, interoperability, information security, and manual processes. However, as many people are not familiar with blockchain, the client needed to know how the app would perform with users before finalizing the design for development. Our class, led by UX research industry veteran Eric Nordquist, were enlisted to conduct this research over the span of 3 months. Student teams conducted a heuristic evaluation, competitive research, and a usability study to generate insights so we could make data-informed recommendations to the client. As this was a student project, the results as far as we know are that the designers made improvements to the app design based on common insights from the 6 student teams. A research paper on MediLinker was published in June of 2023.
Medilinker-Task-1-1_edited_edited.jpg
PROBLEM + IMPACT

The R&D team needed data-informed usability insight on the prototype design

Research Questions
  • Do the designs successfully explain how the app protects a user's information?
  • Do the designs clearly convey what blockchain is?
  • Can users navigate the app and know how to use key features?
  • Would users want to use this app?

Impact to Patients
  • Elimination of need to repeatedly fill out identical paperwork
  • Enablement of secure, direct control over patients' personal electronic health records (EHRs)
  • Simplification of the process of sharing health information with providers, reducing administrative burdens

Impact to Healthcare Practitioners
  • Improvement of access to complete, timely patient information
  • Enhancement of the ability to provide quality care through efficient data management

Impact to R&D Team
  • Transforming EHR management using blockchain technology
  • Allowing secure and immediate data sharing across the healthcare system
  • Reducing fragmented, siloed patient data, streamlining healthcare processes
CONTEXT

Contributed UX research and presentation design

What I Did
  • Conducted heuristic evaluations, competitive research, and designed research studies (participant quotas and screener surveys)
  • Created and facilitated usability testing sessions, including script development and remote testing
  • Administered SUS questionnaires, managed usability note-taking, and performed thematic analysis
  • Led the design of presentation and research readouts

​Team

Vanessa Sanchez
Laura Huerta

Anna Shulpina
Laura Manzanarez

Timeline
3 months

Tools
Figma, FigJam, Zoom, Google Sheets, Google Docs
RESULTS

Most participants liked the idea of the app, but 5 out of 12 participants had concerns about security and privacy

Medilinker-SUS.png
Medilinker-Takeaways.png
Medilinker-Recommendations.png
REFLECTION

Begin readouts with actionable insights

Outline your process and methodology
To provide your audience with some background and sense of how you did things, begin your read out with 1-2 slides on process and methodology.
Try recruiting from online forums like LinkedIn
While this may not be the best strategy in all situations, for graduate school it's pretty good. It will allow you to reach more of your network. You may even be surprised by who volunteers.
Pair insights with concrete recommendations prioritized by usability error severity
Make your reasoning and recommendations very clear with actionable insights and help the team understand which issues should be prioritized in the next iteration.
Lead with consolidated recommendations for your team
In a real team, folks may not have time to dig through a lengthy slide deck. For their convenience, it's a good idea to provide a consolidated list of action items in the first section of the report.
RESEARCH DETAILS

Our research revealed 9/12 participants visit more than 1 clinic per year

Medilinker-Process.png
Medilinker-Methodology.png
Our methodology was thorough but had limitations
We included a participant quota, 21-question screener and recruitment script in the recruitment phase of our methodology. In the design of the study, we chose to begin with 5 baseline questions to build rapport followed by 7 tasks to asses key user flows in the clickable prototype. After each task, participants were asked 7 questions. At the end of each session, participants were asked 10 system usability scale questions. Each session was schedule for 1 hour, though some went shorter or longer. Tasks were untimed and we used the think-aloud protocol. Each session included 1 notetaker and 1 moderator and was recorded with the participant's consent to facilitate notetaking later. The biggest limitation was that we could only recruit from our own friends and family.
Medilinker-Participant-Demographics.png
Our limitations as students resulted in biased sampling and uneven distribution of participants
As students, we recruit who we can from our personal networks. Although we strived for an equitable sampling, our participant demographics admittedly were skewed towards people between 20 and 30 years old, most of which visit clinics fewer than 5 times per year. The lived experiences of people in this age group correlate with greater familiarity and comfort with technology. They are also less likely to be caretakers or live with an acquired disability. 

We evaluated 7 key tasks

The tasks were:
  1. Register
  2. Add a clinic
  3. Send requested info for verification
  4. Disconnect from a clinic
  5. Update driver's license
  6. Consent to research study
  7. Switch users
Medilinker-Task-3-1.png
Medilinker-Task-3-2.png
Medilinker-Task-3-3.png

We recruited as equitably as possible from friends and family, hitting our target sample number of 12

Findings may be skewed
We recognize that some of our qualitative data may be skewed due to our relationships with participants. We did our best to pair moderators up with strangers.
ML-Timeline.png
bottom of page