decodershq/project-eval

The rural areas healthworkers copilot in Nigeria # eval. — AI Clinical Triage Assistant for Community Health Workers `eval.` is a mobile-responsive clinical triage assistant and decision-support copilot designed for Community Health Extension Workers (CHEWs) and rural health practitioners in Nigeria and resource-constrained settings. --- ## 🌟 Key Features - **Guided Step-by-Step Assistant**: 4-step interactive clinical intake stepper (`Demographics` → `Vital Signs` → `Symptoms & Voice Dictation` → `Targeted Focus`). - **Paper Health Record Photo Scanner**: Optical Character Recognition (OCR) and Gemini Multimodal Vision API scanning to automatically extract vitals and patient notes from paper notes and physical chart photos. - **Voice Dictation**: Hands-free symptom input using browser speech recognition. - **Live AI Reasoning Console**: Interactive 5-step clinical reasoning visualizer demonstrating WHO IMCI & Nigeria Standard Treatment Guidelines (NSTG) evaluation. - **Standardized SBAR Handover**: Auto-generated Situation, Background, Assessment, Recommendation briefs formatted for rapid referral to secondary care facilities. - **Offline Protocol Fallback**: Integrated local clinical rule engine guaranteeing zero-downtime operation when internet connectivity drops in remote locations. --- ## 🛠️ Architecture & Tech Stack - **Frontend**: React 19, TypeScript, Vite, TailwindCSS (Vanilla CSS tokens) - **Icons**: Lucide React - **Backend API**: Node.js, Express, Local `llama.cpp` OpenAI-compatible API (`localhost`) running Gemma - **Theme**: Warm Amber & Espresso Brown Palette (`#fef7f0`, `#292524`, `#d97706`, `#b45309`) --- ## 🚀 Getting Started ### Prerequisites - Node.js (v18+ recommended) - `llama.cpp` running locally on `localhost` (with Gemma model) - npm / yarn / pnpm ### Installation 1. **Clone the repository**: ```bash git clone github.com cd project-eval ``` 2. **Install Frontend Dependencies**: ```bash npm install …