WellHabit

WellHabit logo

Inspiration

Students chase grades. Their bodies pay the price. When focus becomes the only metric, sleep, hydration, posture, eye rest, movement, and emotional reflection quietly disappear.

Most student tools count tasks, deadlines, and minutes studied. Few ask the question that actually matters:

Is the student's body keeping up with the student's ambition? That question is why I built WellHabit — a tool for students who study hard but forget themselves. It helps them stay focused while remembering water, sleep, movement, eye breaks, emotional reflection, medication, and recovery. Students should never have to choose between their grades and their bodies.

What it does

WellHabit is a responsible AI wellness companion for students. The loop is simple:

Students log what they did → AI surfaces study-body patterns → WellHabit suggests one small, safe next action.

It catches the behaviors that quietly accumulate during intense study:

studying too long without breaks sleeping too little forgetting water skipping movement creeping fatigue emotional overload missed medication screen-induced eye strain

WellHabit converts these signals into small, practical actions that fit inside a student's existing workflow — not on top of it.

How we built it

We built WellHabit as a Flask web app around one main loop:

daily input → AI-assisted pattern analysis → small wellness action → todo/reminder → user feedback

The backend uses Python, Flask, SQLAlchemy, Flask-Login, Flask-WTF, CSRF protection, SQLite, and Werkzeug password hashing. The frontend uses HTML, CSS, JavaScript, and Jinja templates with a calm green wellness-style UI.

WellHabit includes a focus timer, hydration reminders, sleep logs, mood check-ins, journaling, guided breaks, Care AI, AI-generated micro-intervention todos, pattern recognition, medication reminders, calendar/history tracking, voice/TTS support, and optional camera-based fatigue weak signals.

For AI features, we used the OpenAI API when available, with fallback behavior for local use. For optional camera feedback, we used browser-based vision tools such as MediaPipe Tasks Vision. Camera signals are treated only as weak signals, not emotion diagnosis.

Challenges we ran into

One major challenge was that this was a solo project, and I joined the hackathon relatively late. Because of that, I had to prioritize the most important parts first: a working student wellness workflow, clear SDG 3 alignment, responsible AI boundaries, and a demo that shows real functionality.

The hardest part was turning many features into one connected system instead of a collection of separate tools. WellHabit includes focus timing, hydration reminders, sleep logs, guided breaks, Care AI, AI-generated todos, pattern recognition, medication reminders, calendar/history tracking, voice/TTS support, and optional camera-based fatigue weak signals.

I also had to be careful with responsible AI. Instead of claiming that AI can detect emotions, diagnose burnout, or provide therapy, WellHabit uses safer ideas:

weak signals
behavioral patterns
user confirmation
habit support
human-in-the-loop medication reminders
clear “not therapy” and “not medical advice” boundaries

Technically, I handled browser camera permissions, MediaPipe loading, AI fallback behavior, CSRF protection, notification behavior, and keeping the UI simple even with many features.

Accomplishments that we're proud of

We are proud that WellHabit is more than a habit tracker. It connects studying, recovery, and responsible AI into one student-centered workflow.

Our favorite part is the Care AI micro-intervention loop:

Signal → Suggestion → Todo → Completion → User Rating Logged for Future Personalization

This turns AI support into action. Instead of only chatting, WellHabit can suggest a small task such as drinking water, doing a breathing reset, taking a 20-20-20 eye break, or stretching.

We are also proud that WellHabit detects behavioral trends, not diagnoses. It can recognize patterns such as hydration lag, overfocus, fatigue pattern, and reduced recovery without labeling students with medical conditions.

What we learned

We learned that health-related AI needs more than features. It needs boundaries.

AI should not pretend to be a doctor, therapist, or perfect emotion detector. It should help users notice patterns, take small actions, and stay in control.

We also learned that small interventions are often more useful than long advice. A tired student may not need a lecture; they need one clear next step: drink water, rest their eyes, breathe for one minute, stretch, or sleep earlier.

The biggest lesson was:

Responsible AI is not less innovative. Responsible AI is better innovation.

What's next for WellHabit

Next, we want to make WellHabit more reliable, scalable, and evidence-informed.

Future improvements include:

testing with real students
improving wellness score weights using research
adding Docker for easier deployment
moving from SQLite to PostgreSQL
improving mobile responsiveness
strengthening privacy controls and data deletion
expanding region-matched crisis support resources
improving medication confidence labels
personalizing suggestions using user feedback ratings

Long-term, WellHabit could support schools, student clubs, tutoring programs, and wellness initiatives as a low-cost browser-based tool for healthier study routines.

WellHabit: Study hard. Recover smarter.