Inspiration College burnout builds silently through deadlines, poor sleep, and chronic stress. We wanted a tool that flags risk early and helps students rebalance before they hit a wall.

What it does Burnout Radar imports calendar events, lets users paste assignments, and takes a short voice check-in (Deepgram). It predicts a 7-day burnout risk score and highlights peak stress times. It also includes a Recovery Planner that recommends nap windows + nap length and meal timing using calendar free blocks and sleep debt.

How we built it

Frontend: Next.js / React UI

Backend: API server + database

Integrations: Google Calendar OAuth + Deepgram voice transcription/stress signals

Data: ATUS used for baseline behavioral context (optional / supporting)

Challenges:

Designing a stress model that is explainable and safe (no medical claims)

Making recommendations that respect calendar constraints

Translating voice signals into a useful “stress input” without overreaching

Accomplishments:

End-to-end demo: calendar → assignments → voice → 7-day forecast

Shippable Recovery Planner that feels smart and actionable

Clear ethical framing: not diagnosis, not therapy, no medical data

What’s next:

Better personalization over time using user feedback after deadlines

More robust workload estimation from syllabi/Canvas exports

Share this project:

Updates