🌿 Serenity-AI: Personalized Mental Wellness Companion

📖 Project Overview

Serenity AI is a comprehensive mental health application designed to support users' emotional well-being through journaling, mood tracking, personalized activities, and AI-powered insights.

🚀 Key Features

Personalized mood tracking
Sentiment-aware journaling
AI-powered emotional insights
Activity and exercise recommendations
Gamification of mental wellness
Emergency support features

Frontend Repository

https://github.com/Serenity-AI-Chatbot/serenity-ai-frontend

Backend Repository

https://github.com/Serenity-AI-Chatbot/serenity-ai-ml

📸 Project Media

Demo Video

Architecture Diagram

)

System Flow

)

Screenshots

Home Page

-Interative 3D model of a robot

Dashboard

Activity tracking
Mood trends
Journal insights
Recommended activities

Journal Interface

AI-powered journaling
Emotion detection
Real-time insights
Tag suggestions

Activity Recommendations

Personalized suggestions
Difficulty levels
Progress tracking
Achievement system

Character Interface

Interactive Chat Bot which can be used to talk to the user
with all the context of the user's mood and journal entries

🛠 Tech Stack

Frontend

Next.js 14
Tailwind CSS
Shadcn/UI Components
Vercel Deployment

Backend

FastAPI Service
Supabase Authentication & Database
Google Gemini AI (Embeddings & Language Model)

ML Infrastructure

Custom ML Model (AWS EC2)
Spacy & NLTK Models
Vector Embeddings
Sentiment Analysis Pipeline

Database

Supabase Postgres
Vector Embeddings
Row Level Security

🔧 Prerequisites

Python 3.9+
Node.js 18+
Supabase Account
Google AI Studio API Key
AWS Account (for EC2)
Vercel Account (optional)

📦 Project Structure

mental-health-app/
│
├── frontend/                # Next.js Application
│   ├── app/
│   ├── components/
│   ├── lib/
│   └── styles/
│
├── ml-service/             # FastAPI ML Service
│   ├── app/
│   │   ├── main.py
│   │   ├── models/
│   │   │   ├── custom_model.py
│   │   │   └── nlp_models.py
│   │   ├── routers/
│   │   └── utils/
│   └── requirements.txt
│
├── database/                # Supabase Migrations & Functions
│   ├── migrations/
│   └── functions/
│
└── README.md

🔐 Environment Setup

1. Supabase Configuration

Create a new Supabase project
Enable the vector extension
Run database migrations
Set up Row Level Security (RLS)

Required Supabase Environment Variables

NEXT_PUBLIC_SUPABASE_URL=your_supabase_project_url
NEXT_PUBLIC_SUPABASE_ANON_KEY=your_supabase_anon_key
SUPABASE_SERVICE_ROLE_KEY=your_supabase_service_role_key
GEMINI_API_KEY=your_gemini_api_key
ELEVENLABS_API_KEY=your_elevenlabs_api_key
FASTAPI_API_URL=your_fastapi_api_url

2. Google Gemini AI Setup

Get API key from Google AI Studio
Set environment variable

GEMINI_API_KEY=your_gemini_api_key

3. Frontend Next.js Setup

# Install dependencies
cd frontend
npm install

# Run development server
npm run dev

4. ML Service Setup (AWS EC2)

# Clone repository and navigate to ml-service
git clone <repository-url>
cd mental-health-app/ml-service

# Create virtual environment
python -m venv venv
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Run FastAPI server with PM2
pm2 start "uvicorn app.main:app --host 0.0.0.0 --port 8000"

📊 Database Schema

Core Tables

1. Journals Table

Stores user journal entries with AI enhancements
Fields:
- Basic: id, user_id, title, content
- AI Insights: summary, mood_tags, embedding, keywords
- Metadata: created_at, tags
- Location: nearby_places (JSONB)
- Related Content: latest_articles (JSONB)

2. Activities Table

Catalog of wellness activities
Fields:
- Basic: id, title, description, category
- Difficulty: difficulty_level (beginner/intermediate/advanced)
- Mood Matching: recommended_moods
- Planning: estimated_duration
- Organization: tags

3. User Activities Table

Tracks user engagement with activities
Fields:
- References: user_id, activity_id
- Status: planned_at, completed_at, status
- Feedback: difficulty_rating, reflection

🔍 Database Functions

1. get_mood_trends()

sql Function: get_mood_trends(p_user_id UUID) Returns: TABLE (entry_date DATE, mood_categories JSONB, total_entries BIGINT)

Analyzes mood patterns over time
Groups mood tags by date
Aggregates mood frequencies
Provides total entries per day

2. get_recommended_activities()

sql Function: get_recommended_activities(p_user_id UUID, p_current_mood_tags TEXT[]) Returns: TABLE (activity_id UUID, title TEXT, description TEXT, match_score double precision)

Recommends activities based on current mood
Calculates mood-activity match scores
Uses intersection algorithm for matching
Returns top 5 matched activities

3. match_journals()

sql Function: match_journals(query_embedding VECTOR(768), match_threshold FLOAT, match_count INT, user_id UUID) Returns: TABLE (id UUID, content TEXT, summary TEXT, created_at TIMESTAMP, similarity FLOAT)

Performs semantic search on journal entries
Uses vector embeddings for similarity
Supports configurable matching threshold
Returns matched entries with similarity scores

4. get_dashboard_insights()

sql Function: get_dashboard_insights(p_user_id UUID, p_days_back INT DEFAULT 90) Returns: JSONB Comprehensive analytics function providing:

Journal trends:
- Weekly journal counts
- Mood distribution
- Keyword analysis
Activity trends:
- Completion rates
- Difficulty ratings
- Category distribution
Summary insights:
- Total counts
- Most common moods
- Activity engagement

5. get_journals_by_date()

Function: get_journals_by_date

Description: Retrieves journal entries for a specific user, optionally filtered by year and month. This function allows users to fetch their journal entries within a particular time frame, making it easier to review and analyze past entries.

Parameters:

p_user_id UUID: The unique identifier of the user.
p_year INT DEFAULT NULL: (Optional) The year to filter journal entries.
p_month INT DEFAULT NULL: (Optional) The month to filter journal entries.
p_limit INT DEFAULT 50: (Optional) The maximum number of journal entries to return.

Returns: A table containing the following fields:

id UUID: The unique identifier of the journal entry.
title TEXT: The title of the journal entry.
content TEXT: The main content of the journal entry.
summary TEXT: An AI-generated summary of the journal entry.
mood_tags TEXT[]: Array of mood tags associated with the entry.
tags TEXT[]: Additional tags for categorization.
keywords TEXT[]: Extracted keywords from the entry.
song TEXT: Link to a related song.
created_at TIMESTAMP WITH TIME ZONE: The timestamp when the entry was created.

6. get_journal_stats_by_period()

Function: get_journal_stats_by_period

Description: Fetches journal entries for a user within a specified date range. This function is useful for generating reports and analytics based on user activity over a particular period.

Parameters:

p_user_id UUID: The unique identifier of the user.
p_start_date DATE: The start date of the period.
p_end_date DATE: The end date of the period.

Returns: A table containing the following fields:

id UUID: The unique identifier of the journal entry.
title TEXT: The title of the journal entry.
content TEXT: The main content of the journal entry.
summary TEXT: An AI-generated summary of the journal entry.
mood_tags TEXT[]: Array of mood tags associated with the entry.
tags TEXT[]: Additional tags for categorization.
keywords TEXT[]: Extracted keywords from the entry.
song TEXT: Link to a related song.
created_at TIMESTAMP WITH TIME ZONE: The timestamp when the entry was created.

🔒 Security Features

Row Level Security (RLS)

Enabled on all main tables:
- journals
- activities
- user_activities

Performance Optimizations

sql Indexes: idx_journals_user_id idx_user_activities_user_id idx_journals_mood_tags (GIN) idx_journals_keywords (GIN) idx_journals_embedding (IVFFLAT)

🔧 Vector Support

Enabled vector extension for semantic search
Uses 768-dimensional embeddings
Optimized for similarity searches

📈 Data Analysis Features

Mood Analysis

Daily and weekly aggregations
Mood category distribution
Trend analysis over time

Activity Tracking

Completion rates
Difficulty progression
Category distribution
User engagement metrics

Journal Analytics

Semantic search capabilities
Keyword extraction
Mood pattern recognition
Content summarization

🔄 Integration Points

AI Services

Vector embeddings for semantic search
Mood prediction from journal content
Content summarization
Keyword extraction

External Services

Location-based recommendations
Related article suggestions
Activity recommendations

🚀 Performance Considerations

Optimized Queries

Uses CTEs for complex analytics
Efficient date-based grouping
Indexed text search capabilities
Vector similarity optimization

Data Aggregation

Weekly rollups for trends
Efficient JSON aggregation
Optimized mood analysis
Smart activity matching

🤝 Contributing

Fork the repository
Create your feature branch (git checkout -b feature/AmazingFeature)
Commit changes (git commit -m 'Add some AmazingFeature')
Push to branch (git push origin feature/AmazingFeature)
Open a Pull Request

📜 License

Distributed under the MIT License. See LICENSE for more information.

📞 Support

For issues, please open a GitHub issue or contact [email protected] or [email protected]

🌈 Future Roadmap

Multi-language support
Advanced machine learning models
Integration with health tracking devices

Name		Name	Last commit message	Last commit date
Latest commit History 89 Commits
migrations		migrations
public		public
seed_script		seed_script
src		src
.gitignore		.gitignore
README.md		README.md
components.json		components.json
eslint.config.mjs		eslint.config.mjs
middleware.ts		middleware.ts
next.config.ts		next.config.ts
package.json		package.json
pnpm-lock.yaml		pnpm-lock.yaml
postcss.config.mjs		postcss.config.mjs
tailwind.config.ts		tailwind.config.ts
tsconfig.json		tsconfig.json
vercel.json		vercel.json

Him-anshuSharma/serenity-ai-frontend

Folders and files

Latest commit

History

Repository files navigation