MediKiosk 🏥 — AI-Powered Multilingual Smart Hospital Kiosk & Emergency Dispatch Platform

Inspiration

Government hospitals face long OPD queues, repetitive doctor intake, language barriers, and fragmented emergency ambulance dispatch. MediKiosk addresses these challenges by automating intake before consultation.

What it does

MediKiosk is a full-stack healthcare platform with five interconnected modules:

🤖 AI Clinical Intake Engine

A 15-step clinical interview covers symptoms, medical history, medications, and allergies. It supports 10 Indian languages through an offline medical dictionary and optional Bhashini API. Follow-ups never affect the main progress bar.

🚨 Real-Time Red Flag Detection

Continuously detects emergency indicators, escalates cases to EMERGENCY, alerts doctors via WebSockets, and pre-populates ambulance dispatch.

👨‍⚕️ Doctor Triage Dashboard

Provides AI-generated summaries, vitals, red flags, medical history, documents, and a live queue. Doctors can edit summaries and use clinical chat.

🚑 Smart Ambulance Dispatch & Live Tracking

Uses OSRM road routing, nearest-vehicle matching, Leaflet GPS simulation, and synchronized WebSocket tracking across Patient, Driver, and Hospital portals.

🔐 Role-Based Architecture

Five portals — Patient, Doctor, Driver, Hospital Staff, and Admin — use JWT authentication and facility-level RBAC.

How we built it

Frontend: React, TypeScript, TailwindCSS, Vite, Leaflet, Recharts Backend: FastAPI, SQLAlchemy, SQLite/PostgreSQL, WebSockets Services: AI conversation engine, red-flag detector, clinical summaries, translation, OSRM routing, ambulance matching Deployment: Vercel, Render, Docker Compose

Challenges we ran into

Progress drift, multi-portal synchronization, medical translation accuracy, unreliable Indian road routing, and ambulance acceptance race conditions were solved using authoritative state, curated dictionaries, fallback routing, and database locking.

Accomplishments that we're proud of

• Multilingual medical intake across 10 languages

• Real-time ambulance tracking

• Offline-capable clinical intake

• Production-grade JWT/RBAC architecture

• Faster doctor consultations through AI summaries

What we learned

Healthcare UX requires empathy, medical translation needs curation, real-time systems need one source of truth, state machines improve reliability, and offline-first design is essential.

What's next for MediKiosk

ABHA integration, voice intake, kiosk hardware mode, AI-assisted diagnosis, HIS/EMR integration, and multi-hospital ambulance coordination.

Built With

react · typescript · tailwindcss · vite · fastapi · python · sqlalchemy · sqlite · websockets · leaflet · osrm · jwt · docker · vercel · render

Built With

• docker
• fastapi
• jwt
• leaflet.js
• osrm
• python
• react
• render
• sqlalchemy
• sqlite
• tailwind
• typescript
• vercel
• vite
• websokets