Brain Health Literacy and Research Engagement App

☝🏼

IMPORTANT NOTE: This project is extensive, thus, divided into two case studies: The UX Research (which you are seeing now) and the UX Design part. Don’t worry, research insights are NEVER separate from design choices, so reading each separately will still make sense.

My Roles: UX Researcher | UX Designer | UX Lead

Project Description:

As a part of the Collaborative for Aging Research and Engagement (CARE), we aim to help increase Alzheimer’s Disease and related dementias (ADRD) health literacy and engagement in clinical research for African American/Black (AA/B) adults aged 45 and older.

In this project we designed, developed and are currently testing an AI-supported application to facilitate active engagement in ADRD research.

Process

We conducted a series of UX research and design activities, outlined below:

I highlight the research (UXR) activities that we have conducted. Underneath are the related design (UXD) outcomes in each phase, which are described in the UX case study.

It all began with the Focus Groups and Participatory Design sessions:

We conducted several combined focus group & participatory design sessions at two separate locations. In the first location, we had 3 groups of 5-6 people; in the second, we had 4 groups of 5-6 people. The goal of these sessions was to understand participants’:

1) Connection and experiences around ADRD

2) Information-seeking practices

3) Motivators and deterrents from participating in medical research.

Me (right, standing up) guiding the discussions and brainstorming sessions with my amazing group of participants.

After the focus group session, during lunch break, we rapidly organized the key takeaways and put together an affinity map to brief participants.

🛠️

Methods: Focus Group, Affinity Mapping, User Persona, Problem Statement, HMW question, Nominal Group Technique.

From insights to design ideation

We connected deeply with community members and learned:

Their deep-rooted sense of community leads them to desire information in a compact and shareable form.

Caregiving demands require them to find information quickly and with few steps. Many participants talked about the simplicity of “Googling”.

They have profound distrust in the healthcare system, rooted in a history of racial discrimination, unethical past research, and underrepresentation of AA/B doctors. They mostly place their trust on their social network and their community.

Then came the time to get together and brainstorm…

I led the thematic analysis to achieve team alignment and determine our next design steps. These are the tabulated codes imported from Taguette.

💚

FUN FACT: The name “Lola” came from one of the user personas that our participants created in the FG/PD sessions!

The three interaction modalities: Traditional, Text Bot and Voice Assistant

Across these 3 modes, we ideated and prototyped different concepts in low fidelity to highlight reliably-sourced ADRD information, information transparency as trust builder and community-centered content. You can check out these concepts in the UX Design case study.

Traditional (App)

We created a lo-fi wireframe with a basic architecture to navigate enough to complete the 3 tasks.

Text Bot

We designed the text prompts and responses to simulate a chat bot using the Line app.

Voice Assistant

This mockup was used with a Bluetooth speaker to simulate a conversation, using prompts we designed.

🛠️

Methods: Brainstorm, Sketching, Rapid Prototyping (Figma), Mockups

We developed a protocol to test these interaction modalities separately in a Wizard of Oz test setting, as can be seen below.

Outline of study setup.

We conducted multiple sessions where participants rotated through three stations, each with a different modality (App, Text and Voice) and set of 3 tasks.

In each modality, the three tasks related to 1) ADRD information-seeking, 2) Community, and 3) Research studies.

In each station, tasks were modified to better suit its interaction mode. After participants completed all tasks, we had a semi-structured conversation about their experience and feedback.

I (left) was in charge of the sessions at the traditional (App) station, alongside Chelsea (right) who took notes.

Examples of the tasks participants performed while talking through their thought process. Tasks varied slightly for text and voice.

I recommended to the team that we organized the comments and ideas from our participants in a feedback capture matrix.

We learned that the design needed:

Clearer ways to share information to user’s community, especially in traditional app format.

Improved accessibility to accommodate users age-related visual decline, particularly: App: Reduce number of steps; Voice: Reduce information density and pace of delivery; and App and Text: Increase font size.

Emphasize content transparency by including (medical neighbor) credentials and highlighting the community impact for the AA/B community.

Support user’s mental model of direct and personified interaction (e.g.: Question and answer interaction) by improving conversational agent flow and prioritizing a “search bar” feature.

We organized participant insights into a feedback capture matrix, where the four quadrants describe what worked, what needs to change, what questions were raised, and what new ideas spark from the design. The goal of this was to identify relevant design changes for the next iteration of Lola.

🛠️

Methods: Wizard of Oz, Session Facilitation, Semi-Structured Interview, Think Aloud, Content Analysis

These findings allowed us to iterate on the design to fully integrate all three interaction modalities in a mid-fidelity prototype for its next evaluation.

Adapted Cognitive Walkthrough

Screenshots of integrated LOLA app. The app features the “Chat with Lola” feature, to text with AI

With a fully integrated prototype, henceforth named LOLA, we conducted rapid testing as adapted cognitive walkthroughs, to understand user perceptions related to 1) Usability and accessibility across modalities; 2) Cultural relevance and accuracy of information; and 3) Community-centered strategies for scaling.

🛠️

Methods: Cognitive Walkthrough, Think Aloud, Thematic Analysis

From the Cognitive walkthrough we validated several previously uncovered insights:

While audiovisual and voice input features are primordial to accommodate different user interaction styles, a tighter coupling between all modalities still needs refinement in the app for internal consistency; and on-demand, condensed content in text and audio form are preferred.

In this context, cultural relevance is strongly connected to community and sharing, with explicit references to AA/B representation. Therefore, information that is relevant and local is important to users.

Information actionability to foster engagement and trust. The app should provide clear, actionable steps for user’s health care, community sharing, and research sign-ups, to ensure that users can easily act on the information provided.

At this point, I transitioned out of the project during the Heuristic Evaluation.

In preparation for testing, our awesome developers Darshil Chheda and Satin Jain helped us prepare the Lola app for easy access to pilot testers.

Thus, we brought the app into the App Store and Play Store.

We are currently conducting Pre-Pilot Testing!

At the moment, I am conducting individual debrief interviews (n=15) with pilot testers after a trial period of 1-week, to identify usability, accessibility and cultural issues to iterate upon before our full deployment pilot (n=300) that we will execute soon.

Thank you and Stay tuned for more updates!