Epicast / README.md
janeodum
Docker + CUDA: live LoRA inference + HeAR cough analysis
6ddecee
metadata
title: EpiCast  AI Disease Surveillance
emoji: 🌍
colorFrom: green
colorTo: blue
sdk: docker
app_file: app.py
pinned: true
hardware: t4-small
license: mit
short_description: WHO IDSR surveillance · MedGemma 4B + 27B · West Africa

EpiCast — AI-Powered Disease Surveillance

EpiCast is a mobile-first AI platform for community health workers in sub-Saharan Africa, built on the WHO Integrated Disease Surveillance and Response (IDSR) framework.

Demo Tabs

Tab Description
🔬 Syndromic Extraction MedGemma 4B extracts structured signals from clinical narratives
📋 Situation Reports MedGemma 27B generates WHO IDSR-compliant district reports
🚨 Live ECOWAS Alerts Real-time syndromic alerts across ECOWAS member states
ℹ️ About Architecture, models, and technology stack

Optional: RunPod Integration

Set these secrets in your Space settings to connect to the live RunPod backend:

  • RUNPOD_ENDPOINT_ID — your RunPod serverless endpoint ID
  • RUNPOD_API_KEY — your RunPod API key

Without these the Space runs fully on demo data.

Mobile App

The full EpiCast mobile app (React Native / Expo) runs MedGemma 4B Q4_K_M on-device for privacy-preserving offline syndromic extraction, cough audio analysis via HeAR, and clinical photo triage via MedSigLIP.