Built for busy OPDs
Structured pre-consultation intake gives clinicians a reviewable history while preserving consultation time for examination and medical reasoning.
About MediKiosk
Patient History Before the Consultation Begins
MediKiosk is an independent, open-source prototype that helps patients describe symptoms and share prior records before they meet a doctor.
Designed for Indian patients
Large touch targets, spoken prompts and 10 interface languages support patients across literacy levels, ages and preferred modes of interaction.
Clinician in control
AI-generated summaries remain drafts. A physician can verify, edit or reject every output; the software does not provide an autonomous diagnosis.
Technology
Open, Self-Hostable Technology
The modular prototype supports local components and configurable AI providers, helping teams evaluate deployments with tighter control over health-data processing.
AI & Voice
Sarvam AI
TTS · STT · Translation
Local LLMs
Ollama · Llama · Gemma
OCR Engine
Sarvam · Tesseract · Cloud Vision
Whisper
Local Speech Recognition
Backend
FastAPI
Python REST API
PostgreSQL
pgvector · Embeddings
Redis
Session Cache
FHIR R4
Interoperable Output
Frontend & Infra
React + TypeScript
Vite · Tailwind CSS
i18n · 10 Languages
Hindi · Bengali · Tamil · more
Docker Compose
One-command deploy
Nginx
Reverse Proxy · SPA
Local-first deployment path
Core services can be self-hosted, and speech, OCR and language-model providers are configurable. Production deployments still require a security, privacy and compliance review.
Patient Journey
From Walk-In to a Reviewable History
A voice-first workflow designed for patients of all literacy levels and ages.
1
Choose Language
Patient selects from 10 Indian languages. All subsequent interactions happen in their chosen language.
2
ABHA Login / Register
The prototype demonstrates ABHA-style identification or new-patient registration. The public demo uses a clearly labelled mock identity flow.
3
Audio Consent
Consent is read aloud by the kiosk in the patient's language. One tap to agree — no forms to read.
4
Voice Interview
AI asks structured medical questions via voice. Patient speaks naturally — STT captures, LLM extracts symptoms, vitals, and history.
5
Document Scan
Upload prescriptions, lab reports, or discharge summaries. OCR extracts text, LLM structures it into a timeline.
6
Doctor Dashboard
Physician reviews an editable draft summary with red-flag alerts, a document timeline and FHIR R4 data before using it clinically.
Frequently Asked Questions
Clear Answers About MediKiosk
What is MediKiosk?
MediKiosk is an independent, open-source patient intake prototype for Indian OPDs. It collects structured history by voice or touch, digitizes prior medical documents and prepares a draft summary for physician review.
Who is MediKiosk designed for?
It is designed for high-footfall outpatient departments, including AYUSH and allopathic settings, and for patients who benefit from spoken prompts, regional-language interfaces or large touch controls.
Does MediKiosk diagnose patients?
No. It does not diagnose or replace a clinician. Summaries and red-flag alerts are decision-support outputs that a qualified physician must review, edit or reject.
Is MediKiosk an official ABDM or government service?
No. It is an independent Smart India Hackathon prototype inspired by Ministry of Ayush problem statement 26047. It generates FHIR R4 data, while the current public demo uses mocked or sandboxed identity and health-record connections.
Which languages are available?
The interface includes English, Hindi, Bengali, Gujarati, Kannada, Malayalam, Marathi, Punjabi, Tamil and Telugu. Voice-service availability can vary by configured provider.
Should real patient information be entered in the public demo?
No. The public deployment is a demonstration, not a production clinical service. Use fictional test information only. A real deployment requires a separate security, privacy and clinical-governance review.