uthy4r/mamacord-ai

AI-powered maternal triage platform for frontline health workers in low-resource African healthcare settings. Built with multimodal LLMs to support early risk detection and clinical decision-making at the point of care. # 🤱 Mamacord AI ### *Connecting 30% of the World's Maternal Deaths to the Care That Can Stop Them.* Mamacord AI is an AI-powered maternal triage and referral coordination tool designed for frontline health workers (Traditional Birth Attendants, Community Health Workers, and PHC nurses) operating in low-resource Nigerian settings without access to specialist care or electronic health records. Given a set of structured clinical inputs, Mamacord AI returns an evidence-based **Green / Yellow / Red** risk classification grounded in the Nigerian National Maternal Health Guidelines and WHO Pregnancy Protocols, and automatically generates a structured clinical handover note for Red-flag cases. --- ## 🩺 The Problem Nigeria accounts for nearly **30% of all global maternal deaths**: 1,047 per 100,000 live births. The three leading causes (pre-eclampsia, obstetric haemorrhage, and sepsis) are all detectable with basic clinical assessment. The gap is not medical knowledge. **It is the absence of objective triage tools at the point of first contact.** --- ## ✅ What Mamacord AI Does - Accepts structured manual input of patient vitals, point-of-care lab results, and USS findings - Runs a RAG pipeline grounded in WHO and Nigerian national maternal health guidelines - Returns a **Green / Yellow / Red** risk classification with cited clinical rationale - Automatically generates a structured referral handover note for Red-flag cases - Delivers targeted health literacy content to frontline workers and patients --- ## 🏗️ Architecture ``` 🔢 Structured Clinical Inputs (vitals, labs, USS findings) ↓ ⚙️ FastAPI Backend ↓ 🔍 Hybrid Retrieval: ChromaDB (semantic) + BM25 Re-ranking ↓ 📚 Evidence Retrieval: WHO & Nigerian National Maternal Health Guidelines ↓ ✍️ Evidence-Grounded Prompt Construction ↓ 🤖 GPT-4o-mini: Risk Classification + Clinical Rationale ↓ 🟢🟡🔴 Triage Output + Auto-generated Referral Handover Note ``` --- ## 🛠️ Tech Stack | Layer | Technology | |---|---| | Fronte …