AI‑Powered Mental Health Journal, Sentiment Analyzer & Report Generator
Horizon is a mental health journaling platform built with Next.js that helps users reflect on their daily life, track emotional and behavioral patterns, and generate meaningful insights over time.
AI in Horizon is an enhancement layer, not the product itself. The core of Horizon is structured journaling, analytics, and reflection, with AI used carefully to support understanding, summarization, and long‑term pattern recognition.
- Write and save daily journal entries
- Encourages self‑reflection and emotional awareness
- Helps users build consistency through daily tracking
- Calendar‑based journal history
AI helps interpret data — it does not replace human judgment.
- Sentiment analysis for journal entries
- Non‑clinical condition signals based on trends
- Time‑aware insights (recent days vs long‑term patterns)
- Better long‑term emotional and behavioral trend tracking
AI explanations are grounded in structured analytics, not raw text guesses.
- Search past journal entries using keywords
- Quickly find specific events, triggers, or emotional states
- Useful for reflection and therapist consultations
Each journal entry can optionally include:
- Mood / happiness score
- Confidence level
- Sleep hours and sleep quality
- Stress and overthinking level
- Productivity
- Exercise, diet, and screen time
- Physical problems or symptoms
- Exception notes
- Location (optional)
These structured signals power analytics and AI‑assisted insights.
When a user opens Horizon for the first time, they complete a short onboarding survey to capture a personal baseline, including:
- Typical happiness level
- Normal sleep duration and quality
- Common problems or stressors
- Known mental conditions (optional)
- Location (optional)
-
Personal Baseline Setup Helps Horizon understand what is normal for the user.
-
Improved Insights Baselines are used for comparisons, trends, and AI explanations.
-
Anonymous Community Insights (Optional) Aggregated trends may be shown without exposing personal identity.
Horizon can generate structured mental health reports that users may choose to share with professionals.
Reports may include:
- Weekly / monthly mood trends
- Sentiment distribution
- Sleep‑mood correlations
- Journaling consistency scores
- Highlighted improvement or deterioration periods
- Clear summary insights
These reports help professionals:
- Understand patterns quickly
- Reduce reliance on memory‑based recall
- Make data‑backed decisions
AI is one component of Horizon, not the entire system.
-
Raw journal data is never sent directly to AI
-
Data is normalized and analyzed in code first
-
Trends, correlations, and risk signals are computed deterministically
-
AI is primarily used for:
- explanation
- reflection
- summarization
This design keeps Horizon safe, explainable, and reliable.
Each journal entry may store:
- Journal text
- Sentiment label and score
- Condition signal (non‑clinical)
- Mood / happiness score
- Confidence
- Sleep hours and sleep quality
- Stress and overthinking level
- Productivity
- Exercise, diet, and social interaction
- Physical problems
- Exception notes
- Location
- Created timestamp
AI works on derived features, not raw journal text.
- Next.js (App Router)
- React
- TypeScript
- Tailwind CSS
- Data visualizations & animations
- Next.js API routes
- TypeScript
- Redis (caching & performance optimization)
- Structured analytics pipelines
- Supabase (PostgreSQL)
- Supabase Auth
- Normalized journal schema
- Derived analytics layer
- Sentiment analysis
- Time‑series trend analysis
- Statistical metrics
- Rule‑based intelligence
- Controlled LLM usage (explanation only)
- Alerts when mental health deteriorates over time
- Weekly / monthly reflection summaries
- Downloadable report export (PDF)
- Nearby doctor suggestions (opt‑in, location‑based)
- Advanced analytics dashboard
- Long‑term progress visualization
Horizon is not a medical device and does not provide diagnosis or treatment. It is a self‑reflection and insight tool intended to support mental well‑being.
git clone https://github.com/your-username/horizon.git
cd horizon
npm install
npm run devHorizon is built as a product first:
- journaling
- reflection
- structure
- analytics
AI exists to support these goals, not replace them.