How do you design for emotions that users themselves struggle to identify?
Project overview & My role
Pilot testing wearable-AI assistance for emotion learning
ITI Technologies is a healthcare startup focusing on building self-care and routine-building mobile solutions primarily for neurodivergent users. My team was tasked with the challenge to understand the capabilities of wearable (such as smart watch) to record and predict significant stress signals.
Talked to 10+ neurodivergent users about their everyday struggles and effective coping strategies for feeling good
Led 3 rounds of pilot testing over 3 weeks to Iterated a wearable–mobile system
Key challenges
1) Resolving differences between client needs and user wants
What we assumed - The early conversation within the team focused on the technical difficulties of getting a wearable device to accurately predict the exact user feeling. We researched a wide range of wearable offerings (Garmin, Oura Ring, Emotion Sticker), aiming to find a technology that allows us to quickly experiment with emotion sensing.
What research revealed - Users are more interested in interpreting what they feel first. Moreover, they want to be resilient. Thus, instead of doubling down on assessing the accuracy of biosensors, I pitched the team to shift focus to creating user flows to uncover emotion patterns and build coping strategies step-by-step.
From interviews with 10 neurodivergent users
"I have a lot of trouble identifying emotions other than sad or overwhelmed."
"racing thoughts, sweating, panicking, stomach pains, looping memories over and over in my head when I experience a meltdown"
2) Operating proof-of-concept testing in 3 weeks
We had 3 weeks and a complex system that could break at three different points. Rather than testing everything loosely, I made the call to test only two flows in-depth, the crucial ones forming the backbone of the concept. Testing the machine learning algorithm live for pattern discovery was de-prioritized for future improvements once the flows are established right.
Problem statement
Current tools lack support to address neurodivergent users’ struggle with recognizing emotions.
Wellness market is divided in two distinct directions
Personal triggers, generic solutions.
The wellness industry splits into 2 categories: wearables that track biometrics, and apps that guide mindfulness. Both ends put the heavy lifting on users to self-evaluate their physical and mental conditions. Health services act like trackers than support systems.
User story #1
As an autistic user, I want a wearable made of familiar materials to avoid sensory overload.
User story #2
As a user with ADHD, I want a tool to recap scenarios and explain triggers to understand my feelings.
Wellness features aren't as useful as soothing techniques.
While I gained more insights on users' needs for emotional learning through interviews, users also mentioned the contrast between coping strategies and gamified approaches by wellness apps. Features like streaks or daily reflections may feel fresh at first, but they rarely alleviate the emotional weight, as indicated by the high drop-off rate.
Journaling vs. Journal features
60% of users choose journal as the go-to reflection tool
Constant efforts and little feedback often lead to dropoff
Coping strategy vs. Daily streaks
7 out of 10 users uses grounding or listens to music for coping
May create reliance on rewards without actual relief
Our opportunity
How might we help users recognize what they feel before it becomes a crisis?
Elaborating on the critical pain point that user struggled to interpret feelings, I designed a framework that combines physical and psychological cues to identify potential patterns of overwhelming signals.
Each step of the cycle means to build on each other towards the goal of pattern recognition.
Early concept testing feedback
More steps creates more friction
Users expressed interests in the quick tap action but were hesitant about using smart watches especially when they are in meetings and presentations.
User prefers a more subtle way to record.
Time-based journal creates emotional clutter where sometimes user prefer to have reflections as something optional. Emoji representation of emotion also feels limiting as user frequently mentioned the difficulty of recognizing emotions in the moment.
My contributions at a glance
A feedback loop supported by gestures and constant learning
01
A usability re-imagination: optimizing comfort and privacy with a soft wearable
User testing wearable 2.0
It started with the question asked during user testing. "How do I log my thoughts in the middle of a meeting?" We explored different tactile methods, eventually consolidating on a wrist wearable where user can wear it like a bracelet and tap or squeeze during moments of unease.
01
Tap now, view later: read relative data and analysis about tapped emotions anytime
User navigates between three tabs to view their entries, insights regarding their logs, and AI chats spaces to practice calming techniques.
01
Mood forecast: an AI-empowered review of logged emotions to provide actionable feedback
Based on the data received, the app utilizes AI to analyze and then suggest personalized regulation strategies. The suggestions aim to serve as gentle nudges rather than milestones for users to achieve.
The outcome
Reduced logging interaction time from 12 seconds to 4 seconds
Both the user and the client expressed great interest in investing in a knitted wearable for capturing their emotions. The user tested the wearable and gave it an overall score of 4 out of 5 for its comfort, ease of use, and overall willingness to use.
>
snapshots from user testing
Press to log emotion: being sensory-friendly
"This method feels natural and unobtrusive. A simple press lets me log an emotion effortlessly without disrupting my daily activities."
The versatility of the proof-of-concept
"The data collected can be valuable for me to be aware of how I feel but also potentially be insighful to show to my therapists."
