AI-Powered Respiratory Health Monitoring
As the first mobile app using AI to turn cough recordings into respiratory health insights, this project makes MyAdvocate's product's user experience feel intuitive, trustworthy, and human-centered.
Indsutry
Healthcare (B2C)
Team
CEO, 2 PM, 2 Dev, 1 UX Designer (me)
Timeline
June - Dec 2023
Project Type
Mobile Native
Overview
Project Background
Have you ever had a sore throat but were too busy to book a doctors appointment?
MyAdvocate is a new app that uses AI-driven analysis to listen to sound of your cough and provide an instant wellness score. As the sole UX designer at this startup, I transformed a confusing user experience into an approachable human-centered health tool.
Process
User-Centered Design
1
Discover
2
Definition
3
Ideation
4
Design & Test
Impact
An user-friendly, accessible health tool with a refreshed look and feel
App Usability
Likeliness to Return
New WCAG Compliant Colours & Font
Reusable Design System Components
Problem
Advanced technology, low engagement
What once required a doctor’s visit and days of waiting for resutls can now be done at home in under a minute. MyAdvocate's AI sound-analysis technolgoy is revolutionary step forward in the wellness space. So why was user adoption so low? The team brought me in as the sole UX designer to uncover usability issues and design an improved solution.
Images of original user experience
Research Insights
After conducting 5 user interviews and usability tests, it was clear that users found MyAdvocate difficult to use. Not only was there too much text, but it felt cold and clincial. Users felt left in the dark with data they didn't know how to interpret or connect with.
Low Task Success Rate
Majority of users received an error during their recordings.
Low Understanding of Score
Majority of users incorrectly interpreted their score.
Low Likeliness to Return
Majority of users were unlikely to return to the app.
Pain Point #1
Overwhelming instructions lead to low adoption
User research showed that adoption was being hurt by a long, instruction-heavy onboarding experience. Users were asked to read and remember too much information upfront, increasing cognitive load and making it harder to move confidently through the next screens. This led to a consistent drop-off early in the experience.
Before
Users overwhelmed by lengthy instructions and low readibility.
After
Engaging cards with digestable content, reducing congitive load.
Key Design Decision
From Lists -> Content Cards
To make the onboarding experience more lightweight and engaging, I proposed to break down the long list of instructions into muliple screens. We tested multiple ideas and eveutally landed on swipable content cards due to their fast-paced nature and usage of natural swiping gestures. After approval from the CEO, we decided to move forward with content cards using a refreshed color palette and to improve contrast and accessibility and a modern feel.
Pain Point #2
Lack of visual cues lead to high error rates
Once users got past the instructions, many felt unsure about what to do next on the recording screen. Questions like “Did it start recording?”, “What do I press now?”, and “Am I doing this right?” came up, along with many error messages. It was clear that users needed more guidance and reassurance while completing the recording.
Before
User confusion and frustration about their progress, no clear next steps or instructions.
After
Clear understanding of progress and ideal strength of cough thorugh icons, colour and helper copy.
Key Design Decision
Introducing the new cough strength indicator
One of the key issues the data science team faced with user cough submissions was that they were either too loud or too quiet for processing, causing an error message and another cough submission. Meanwhile, users were left in the dark had no idea what went wrong. My initial idea was to show the cough wave on screen while recording for visual feedback.
However, after collaborating with the data science team we came to a compormise to show a dynamic scale that change based on the decible level of the cough. By using helper text to guide users to the next step, and colour to illustrate where the ideal strength of cough is, clarity and trust was added to user experience.
Pain Point #3
Confusing scores lead to low retention
After receiving their respiratory score, most users didn’t understand what it meant for their health. The value of the score wasn’t clear, leaving users to figure out its meaning on their own and adding unnecessary cognitive load. To address this, I focused on making the score immediately understandable, actionable, and easy to interpret.
Before
Users unable to interpret score, confused by jargon and cluttered UI.
After
Visually prioritized content with new UI layout using size, labels and colours. FAQ section added for deeper inquiries.
Key Design Decision
New UI Layout with clear visual hierarchy
To simplify the score screen by making the focal point the score and the scale, I worked with the CEO and Product Managers to identify which content was essential to keep on screen, and which could be deprioritized. I proposed a tab structure, one for the main score information and the other a FAQ tab for users who had a deeper curiousity. When we tested this idea it worked well! However, we noticed users were still a bit on the fence with how to interpret their score, so on the second iteration I proposed adding in a AI summary card with a quick score summary.
Solution
Making complex health insights simple & accessible
MyAdvocate now tells a story, connecting the dots and filling in the gaps in users’ knowledge. Key information is presented in bite-sized pieces, real time guidance is provided to meet the user right when they need it, and the score is easy to understand at a glance. Together, these improvements help users discover health insights with confidence.
Reflection
Effectively prioritizing tasks as a solo designer
As the only designer on the team, I learned how to break down visionary work into tangible action items, communicate effectively with my team about timelines and delvierables and gained insight into all aspects of the user experience from research, to UX writing, to design.
Collaborating cross-functionally in a fast paced environment
Working with this startup team challegned me to adapt to their fast-paced sprints and MVP mindset. I learned to focus on making incremental changes that could build toward the ideal state over time.
If there was more time, I would love to explore the profile section of the app and evaluate how the AI's predictive trends and insights could be shared to further personalize the experince and increase retention even further.