Parkinson's care has a 6-month clinical blind spot. Neurologists rely on infrequent clinic visits to adjust critical medications, while patients face daily risks of falls, freezing of gait, and voice degradation at home without continuous monitoring.
NeuroEdge is an edge-AI ambient monitoring system that tracks patients at home without requiring them to wear any devices.
- Edge Vision Tracking: Uses a localized computer vision pipeline to monitor patient posture and instantly detect physical falls in real-time.
- Generative Voice Triage: Upon detecting an anomaly, the system automatically initiates a conversational assessment using LLMs to determine patient safety, pain levels, and mobility.
- Clinical Reporting: Instantly compiles the acute incident transcript alongside 30-day longitudinal biomarkers (e.g., medication adherence, hypophonia progression, morning gait freezing) into a structured PDF for rapid physician review.
- Computer Vision: OpenCV, MediaPipe Tasks Vision API (Pose Landmarker)
- Generative AI: Google Gemini 3.5 Flash API (
google-genaiSDK) - Speech & Audio:
SpeechRecognition, Google Text-to-Speech (gTTS),pygame - Data Export:
FPDF
Create and activate a Python virtual environment to keep dependencies clean:
python -m venv .venv
source .venv/Scripts/activateInstall the required libraries for vision, LLM integration, and audio routing: pip install opencv-python mediapipe SpeechRecognition google-genai gTTS pygame fpdf
pip install opencv-python mediapipe SpeechRecognition google-genai gTTS pygame fpdf- Configure the API Key Export your Google Gemini API key to your local terminal session:
export GEMINI_API_KEY="your_api_key_here"- Launch the System Run the main tracking script. The system will automatically download the required MediaPipe .task model on its first run.
python vision_tracker.py
To trigger a fall: Drop your head below the red threshold line on the camera feed.
To quit: Press q while focused on the camera window, or Ctrl + C in the terminal