mymunah-07lmtc/Medical_AI_Deploy

Unified offline AI diagnostic device for low-resource settings. Combines lung and heart sound screening into a single Raspberry Pi deployment package. # 🏥 Unified AI Medical Sound Screener: Sahel Sound Triage Platform A unified, offline, AI-powered diagnostic device for low-resource settings. This single application integrates two independent screening modes: - **🫁 Lungs (Respiratory)**: Detects pneumonia, COPD, URTI, and other respiratory abnormalities. - **🫀 Heart (Cardiac)**: Detects heart murmurs for rheumatic heart disease screening. Built by **Maimouna Tougoutcho Coulibaly** in Bamako, Mali. --- ## 📦 What's Inside ``` MEDICAL_AI_DEPLOY/ ├── app.py # Master application (Lung + Heart selection) ├── lung_model.pkl # Trained SVM for lung sounds ├── lung_scaler.pkl # Feature normalizer for lungs ├── lung_label_map.pkl # Label encoder (0=Normal, 1=Abnormal) ├── heart_model.pkl # Trained RandomForest for heart sounds (with SMOTE) ├── heart_scaler.pkl # Feature normalizer for heart ├── heart_label_map.pkl # Label encoder (0=Normal, 1=Abnormal) ├── requirements.txt # Python dependencies ├── start_medical_ai.sh # Auto-start script for Raspberry Pi └── README.md # This file ``` --- ## 🚀 Quick Start (Laptop/Desktop) ### 1. Clone or Copy the Folder ```bash cd Desktop/MEDICAL_AI_DEPLOY ``` ### 2. Install Dependencies ```bash pip install -r requirements.txt ``` ### 3. Run the App ```bash streamlit run app.py ``` Your browser will open at `localhost`. ### 4. Select Mode - Click **"🫁 Lungs (Respiratory)"** for lung sound screening. - Click **"🫀 Heart (Cardiac)"** for heart sound screening. --- ## 🛠️ Hardware Deployment (Raspberry Pi) This app is designed to run offline on a Raspberry Pi 4 (or 3B+) with a USB sound card and lapel microphone. ### Hardware Required - Raspberry Pi 4 (2GB+) - USB Sound Card (CM108 chipset) - Lapel Microphone (3.5mm) - Power Bank (5V, 3A) - MicroSD Card (16GB+) ### Copy to Pi Copy the entire `MEDICAL_AI_DEPLOY` folder to `/home/pi/` on your Raspberry Pi: ```bash # Using a USB …