About the Project

Inspiration

Caring for loved ones with Dementia and Parkinson’s can be overwhelming for families and caregivers. One of the biggest challenges in Parkinson’s care is managing medications like Levodopa/Carbidopa (Sinemet), the gold-standard drug that must be taken multiple times per day (3–5x).

This creates a timing-sensitive challenge:

• Patients may forget doses or accidentally double-dose — both of which pose serious safety risks.
  
• Family caregivers feel constant stress about leaving loved ones alone.
  
• Professional caregivers and nurses in facilities often manage multiple patients and need a centralized way to keep track of everyone.
  

This precise-timing medication problem inspired us to build ReMind. With Voice AI agents, we can demo how multiple reminders per day prevent missed or double doses, helping patients stay safe while reducing stress on caregivers.

What We Learned

• How to integrate Voice AI agents (VAPI API) to handle high-frequency, time-critical reminders.
  
• Building with FastAPI (Python) for backend services and React + TypeScript + Vite for the frontend.
  
• Using Tailwind CSS + DaisyUI for an accessible, responsive design.
  
• Working with Postgres to securely manage patient schedules and caregiver data.
  
• The importance of syncing medication adherence tracking across patients and caregivers.
  

How We Built It

• Frontend: React (TypeScript, Vite, Tailwind CSS, DaisyUI) for a clean, responsive interface.
  
• Backend: FastAPI with Uvicorn, using python-dotenv for configuration.
  
• Voice AI: Integrated with VAPI API to provide real-time reminders, confirmations, and caregiver alerts.
  
• Database: Postgres for secure storage of patient schedules, reminders, and caregiver dashboards.
  
• Hosting/Dev: Frontend Deployed with Vercel, Backend via Docker

Challenges We Faced

• Timing precision: Demonstrating multiple reminders in a single day for Parkinson’s patients required carefully scheduled calls.
  
• Dependency issues: Node mismatches (pikaday, react-hook-form) and Python builds (pyaudio needing portaudio).
  
• Syncing front + back: Ensuring reminders made in the UI triggered VAPI voice calls reliably.
  
• Safety concerns: Preventing duplicate reminders or double-dosing required robust backend logic.
  
• Scale: Designing a dashboard that works for families and professional caregiving facilities.
  

What’s Next

• Deploy to cloud with family + facility caregiver accounts.
  
• Integrate with wearables for Parkinson’s symptom tracking.
  
• Expand multilingual voice agents.
  
• Partner with healthcare providers for clinical validation.
  

Built With

• fastapi
• postgresql
• react
• tailwind
• typescript
• vapi