Mood Mentor

Inspiration

We live in a hyper-connected world where constant digital engagement, work demands, and unhealthy routines silently affect our emotional and physical well-being.

While wearables and mobile apps collect large amounts of health data, that information is often fragmented and difficult to interpret. Users are left with metrics such as steps walked, hours slept, and screen time used, but no clear understanding of how those numbers relate to their mental health.

We set out to build a system that bridges that gap. Mood Mentor doesn’t just track data. It analyzes it, identifies patterns, and provides personalized recommendations grounded in behavioral science, all in an accessible and actionable format.

What It Does

Mood Mentor is a mobile-first, AI-powered health analytics platform that translates health and behavior data into clear, personalized insights. Built using Swift for the iOS frontend and Python for the backend with Fetch.ai’s uAgents framework, the system:

Collects and normalizes data
Gathers structured inputs from Apple HealthKit, screen time APIs, and journaling entries. Formats them into a unified structure for consistent analysis.
Finds patterns and correlations
Uses time-series modeling, clustering, and sentiment analysis to identify how daily behaviors impact mood and stress levels.
Identifies personal stressors
Detects specific habits or behavioral patterns that align with negative emotional trends such as disrupted sleep or extended screen time.
Delivers personalized recommendations
Matches user behavior with research-backed suggestions tailored to improve emotional well-being.
Presents insights through a mobile interface
Visualizes mood trends, highlights key stressors, allows journaling, and helps track behavioral change over time.

How We Built It

Mood Mentor is structured as a modular, multi-agent system built using Fetch.ai’s uAgents. Each component handles a specific role and interacts with others to produce real-time, user-specific feedback.

System Architecture

DataCollectionAgent
Pulls and formats data from Apple HealthKit, usage trackers, and journaling inputs. Standardizes varied formats for compatibility.
PatternAnalysisAgent
Applies methods such as:
- Time-series analysis to capture daily and weekly behavior-mood relationships
- Threshold detection for key triggers
- Clustering to categorize user states
- Sentiment analysis and topic modeling on journal entries
RecommendationAgent
Maps insights from the analysis to tailored feedback using behavioral science and best practices in mental health.
ClientAgent
Coordinates communication between agents and connects backend logic with the iOS frontend.

Frontend Implementation

Built using Swift for iOS
Provides an interactive dashboard with mood graphs, journaling tools, and recommendation delivery
Communicates securely with backend agents via RESTful APIs
Designed for clarity, usability, and responsiveness on mobile devices

Challenges We Faced

Data integration
Harmonizing different data types and structures across multiple input sources required dynamic and resilient parsing logic.
Pattern detection accuracy
Preventing false correlations and noise interference in real-world datasets demanded precise tuning and layered validation.
Privacy and security
Ensuring that user data remained private while still delivering value was critical. Using Fetch.ai’s agent model allowed data to be processed without centralized storage.
Delivering meaningful recommendations
Avoiding generic advice required building logic that adapts to the user’s unique patterns and conditions.

Accomplishments We're Proud Of

Developed a full-stack solution that turns raw personal data into clear feedback using distributed agents
Created effective behavior analysis tools using a mix of quantitative and qualitative techniques
Built a functional, privacy-conscious architecture that allows decentralized processing
Designed a user-friendly mobile interface that balances simplicity with technical depth
Established a clear roadmap for integrating Mood Mentor into broader digital health and wellness ecosystems

What We Learned

This project helped us gain deep experience in:

Building and scaling agent-based architectures with Fetch.ai’s uAgents in Python
Analyzing multi-source behavioral data using time-series modeling, clustering, and NLP techniques
Developing secure systems that support health data insights without compromising user control
Designing feedback loops that are both automated and personalized
Translating behavioral science into technical logic that drives a better user experience

Final Thoughts

Mood Mentor is a step toward smarter, user-centered health technology. By helping users understand how their behaviors affect their emotional health and giving them feedback they can act on, we hope to make mental wellness more achievable for everyone.

We look forward to continuing development and exploring new opportunities to expand the platform’s impact.