Psychosis iREACH
The project aims to develop and design a digital platform that uses machine learning technologies to deliver CBTp- informed intervention for the families or caregivers of individuals with psychosis. Aims to boost caregivers' confidence in taking care of their loved ones and enhance their abilities in caring for, connecting with, and communicating effectively with them, Psychosis iReach offers professional knowledge into caregiving skills and provides training in emotion management for caregivers of psychosis patients.
Basic Information:
My Role
UX Researcher
UX Designer
My Responsibility
User Research
UX Design
Usability Testing
Data Analysis
Research Methods
Affinity Mapping,
Data Analysis,
Empathy Map,
System Usability Scale, Usability Testing,
Journey Map,
User Interview
Project Duration
Sept. 2021- June 2023
Project Info
Research Project at the University of Washington, Seattle
Team
2 UX Designers
1 UX Researcher
1 Developer
1 Project Manager
Project Objective
Understand the pain points of caregivers
Assess the usability and user experience of the Psychosis iReach Chatbot feature
Prioritize recommendations so the platform can address users’ deepest pain points
Discovery:
Initial Assumption
Caregivers of individuals with psychosis lack professional caregiving knowledge for psychosis patients and may face an imbalance between taking care of patients and managing their emotional stress through the process of caregiving.
Target User
Caregivers of individuals with psychosis
Early Stage Research：Evaluate Assumptions
Conducted user interviews was to evaluate our assumption and gain understanding of our users’ needs, wants, and current pain points.
Research Methods:
User Interview
​
Objective:
Evaluate our assumptions and understand the pain points of caregivers
​
Interview Location:
Zoom
​
Interview Length:
15 minutes
​
Interview Participants:
5 Caregivers of Individual with psychosis
​
Data Analysis Methods:
Affinity Diagram
​
To avoid bias, we recruited participants represent diverse ethnic heritages, age ranges, genders, device user time, and attention to technology trend.
After conducted the one-to-one interview with users, we analyze the qualitative data found from these interviews utilizing Affinity Diagram:
We grouped things by which task/screen it was associated with.
Themes are found between participants’ quotes within questions:
After discovering themes across multiple interviews, we worked together to synthesize the data and agree on insights.
Summarizing from the interviews to understand Caregiver Needs:
Design:
Practice skills with Chatbot
Our product comprises two components: the learning module and real-time practice with the Alfie chatbot. In the real-time practice section, caregivers engage with Alfie, responding to diverse scenarios. In the event of errors, the coach chatbot intervenes, offering corrections and explanations for the incorrect choices, and guiding caregivers toward more effective reactions to enhance their care for loved ones.
Patient who is impacted by psychosis
Alfie Chatbot represents a case scenario of an individual impacted by psychosis, which informs users about their concerns. In the practice modules, users will interact with Alfie in ways that will require users to demonstrate the skills they learned during the Psychosis iReach training.
Coach who guides caregivers and provides suggestions
The Coach Chatbot guides users to correctly respond to different scenarios and offers users knowledge into professional caregiving techniques for supporting their loved ones, along with guidance on stress management and self-care practices to enhance their well-being.
Caregiver
Caregivers will practice the skills they learned from the module with Alfie Chatbot and get constructive feedback from the coach.
Design Iterations - Before:
Practice your knowledge with the Virtual Chatbot
Before users enter the virtual practice session with the chatbot, we will ask users to create an Avatar for their Alfie chatbot. This can also help users to resonate with the platform.
Lacks gender inclusivity.
No skip function
No Skip Function to allow users who want to jump directly to the practice without the customization process.
In the Chatbot practice session, Alfie Chatbot represents a case scenario of an individual impacted by psychosis, which informs users about their concerns. Users are required to select the accurate option from the three choices provided.
Meaningless Profile Photo
The profile photo of Alfie Chatbot does not reflect the user-customized avatar.
Clustered Message Bubbles
The clustering of message bubbles creates confusion in identifying which side they belong to.
Useless Text Entry Field
The text field is unnecessary since users only need to choose options in the multiple-choice question, without entering any text.
After a caregiver makes a choice, if an error occurs, the Coach Chatbot guides users to respond correctly to various scenarios and offers a detailed explanation of why the chosen option is incorrect.
Message Text Redundancy
After users make their selection, the interface still displays the entire option list
Functionality of the “Hide Text” Button
Meaningless Progress Bar
The progress bar is limited to showing only the percentage, offering no other useful details.
Clicking the 'Hide Text' button may leave users uncertain about which text will be hidden.
Test:
Usability Testing based on the "Before" design
I developed usability testing protocols and conducted usability tests for 6 participants who are all caregivers of individuals with psychosis. I conducted individual 45-minute interviews with each participant.
We grouped things by which task/screen it was associated with.
Taking a closer look at the Affinity Map:
Summarize our findings:
Main recommendations that were discovered during the tests and addressed in the iteration:
​
For Coach Chatbot:
​
​​
​
​
​
​
​
​
​
​
​
For Alfie Chatbot Customization Process:
​
​​
Design Iterations - After:
Practice your knowledge with the Virtual Chatbot
Welcome page screen to introduce the three characters that will appear in the virtual coach feature includes a skip function to allow users to skip the introduction and jump directly to the real-time practice.
Setting up Alfie Chatbot:
To make this function more personalable, we designed the Alfie Chatbot customization function before users enter the virtual practice session with the chatbot.
Avatar Customization
Allow users to choose and design their own Alfie Chatbot
Upload Photo
Users have the option to upload a photo for their Alfie Chatbot instead of customizing an avatar.
Inclusiveness
Add the “They” choice for gender options.
5-step Avatar Customization
Divided the customization process into 5 steps. Users can create an avatar that resembles their loved ones.
Skip Function
Add the Skip function for users to skip the Alfie Customization Process.
After Setting Up Alfie's Avatar, Start the Exercise:
Optimized the Progress Bar Design
Enhanced the progress bar design to display both the module name being practiced and the completion percentage
Rearranged the Message Bubbles to Avoid Clustering
Separated the message boxes to improve readability, making it easier for users to identify the side from which the message originates.
Removed the useless text entry field
If the caregiver made a mistake:
“Try Again” Button
Provide users with the opportunity to reattempt problems they made mistakes on, incorporating feedback from the coach's suggestions.
Eliminate Text Redundancy
Only show the user’s choice and the coach’s suggestion for clarity and simplicity
Interactive Progress Bar
Test:
Usability Testing based on the "After" design
I developed usability testing protocols and conducted usability tests with 10 participants. I conducted individual 45-minute interviews with each participant.
1
Complete a list of tasks
2
Think Aloud
3
Follow-up
Questions
4
System
Usability
Scale
​
Visualizing the result:
100 %
tasks completed
10 / 10
rated the overall chatbot functionality as 5/5
9 / 10
rated the Alfie Chatbot functionality as 5/5
8 / 10
rated the progress bar functionality as 5/5
We utilized two usability testing methods: Affinity Mapping and System Usability Test Scale. ​Here are the Affinity Mapping results:
Taking a closer look at the Affinity Map:
Main pain points that were discovered during the tests and addressed in the iteration:
​
​Users may have problems understanding the different characters involved in the virtual exercise​
​​​Fix: Add a help page to explain the characters further
​
When users are customizing the avatar, they may want to go back to the previous step.
​​​Fix: Add a "Previous" button on each step
​
Users may want to edit their Avatar afterward, and there is no place to edit the Avatar.
​​​Fix: Add a "Edit Your Alfie Avatar" function
Utilizing System Usability Scale:
We utilized the System Usability Scale, a valid and reliable 10-item, 5-point Likert scale that has been widely adopted by studies that focus on the family caregivers population.
According to the developers of the scale, scores > 68 out of 100 indicate higher levels of usability.
Therefore, participants generally feel that our system’s user interface is easy to use.
​Mean SUS Score
93.1
out of 100
Takeaway:
In this project, I had the opportunity to practice both my UX design and research skills while employing my quantitative skills to generate practical solutions. I have strengthened my design knowledge and enhanced my understanding of the relationship between theory and research in the field of human-centered design.
Talking with our participants during rounds of user interviews allowed us to understand and evaluate the needs and values of key stakeholders while effectively presenting information and solutions to users.
I have found great enjoyment in working with peers and people from different technical backgrounds and witness how our work can create a difference in people’s lives.