SomniBot: AI-Powered Personalized Sleep Assistant
A one-month hackathon project that achieved a Top 1% result. My role was UI/UX and prototype design.
Background
Sleep is essential, yet 30-40% of adults struggle with insomnia. Our initial research revealed a significant gap in the market: 80% of users desired a private, trustworthy sleep companion. Existing solutions like generic searches or AI chatbots failed to build this trust due to privacy issues and unreliable information. This defined our core design challenge: How can we create a clinically-informed, personalized experience that users feel safe confiding in?
Goal
To design and prototype an AI-powered sleep assistant that delivers personalized, clinically informed recommendations using FHIR and RAG, addressing the user's need for privacy, trustworthiness, and personalization.
Process & Design Strategy
Recognizing that a 'one-size-fits-all' solution would fail for a complex issue like insomnia, our design strategy centered on "progressive personalization." We mapped a user journey from initial uncertainty to actionable insights. After collaborative brainstorming, my design process focused on three core features to build this journey:
- Quick Self-Assessment: To provide immediate value and establish a baseline without overwhelming the user.
- Symptom Tracking Portal: To empower users to see patterns in their habits for more accurate insights.
- Common Causes Selection: To simplify a complex input process through a structured, visual interface.
In-Depth Architecture
User Flow Diagram
Step 1: Start & Home
User begins their journey and selects "Quick Self-Assessment".
Step 2: Questionnaire
User answers a series of 5 questions about their sleep habits.
Step 3: Result Summary
The app shows an initial result summary and suggestions.
Step 4: User Choice
User can choose to track symptoms or explore common causes.
Step 5: End Session
User returns to home or ends their session with actionable insights.
Information Hierarchy
- Level 1 – Main Functions
- Self-Assessment
- Symptom Tracking
- Common Causes
- Level 2 – Module Actions
- Q&A flow
- Data input & history
- Explore causes
- Level 3 – Micro Interactions
- Button states
- “Suggest” interactions
- Dynamic prompts
Early Ideation Sketch
Early ideation sketch for the 'Common Causes' feature, focusing on a simple, visual selection process.
Accomplishments that we're proud of
- Successfully implementing RAG to improve response accuracy, which directly builds user trust in our product.
- Seamlessly integrating complex FHIR data into a user-friendly, calming chatbot interface that I designed.
- Designing an intuitive self-assessment and EHR-driven analysis system that simplifies a complex medical process into an accessible experience.
The final UI focused on a clean, intuitive, and calming interface to guide users through their sleep analysis journey.
Takeaway
- From a design perspective, this project reinforced the critical role of UX/UI in translating complex medical data into a simple, calming, and effective user experience that guides users toward better sleep habits.