MedicAId
A WhatsApp health-record demo we put together in five hours at a hackathon.
Built in <5 hours · hackathon · healthtech
The idea
For a hackathon, we built a WhatsApp bot to handle medication reminders, appointments, prescriptions, and test results. We also wanted someone to be able to help a child or an older parent with these tasks from the same chat.
What we made
MedicAId connects WhatsApp to a simulated FSE. A user writes in Italian, and the bot can look up family records, create reminders, list appointments, or return a document. We used OpenAI and Gemini APIs to route natural-language requests. The health-record integration is mocked for the demo; it is not a claim of access to the real FSE.
In under five hours, we got the Django backend, Celery workers, WhatsApp webhook, model routing, caregiver permissions, and a deployed demo working together.
What the demo could do
- Medication reminders — automatic alerts, snooze support ("remind me in an hour")
- Appointment management — weekly/monthly overview, first available slot with a specific doctor
- Prescription expiry alerts — proactive notifications before expiry
- Documents & test results — notifies new results, sends PDF on request
- Caregiver mode — secure access to family members' data with role verification via phone number
Demo
Architecture
Incoming WhatsApp messages go to Celery so the webhook can answer quickly while a worker handles the slower model call. Cloudflare Tunnel exposed the demo without spending hackathon time on firewall configuration.
The caller's phone number is checked against the simulated FSE's authorized list. Owners, children, and caregivers have different permissions, and a number without access does not get health data back.
Tech Stack
| Layer | Technology |
|---|---|
| Chatbot | WhatsApp Business API (Meta) |
| Backend | Python, Django, Celery, Redis |
| AI | OpenAI GPT + Google Gemini |
| Infrastructure | Docker, Cloudflare Tunnel |
| Security | Role-based access via FSE phone list |