Skip to content

Latest commit

 

History

History
188 lines (128 loc) · 5.39 KB

README.md

File metadata and controls

188 lines (128 loc) · 5.39 KB
Raw

RtDC-LLM-IoT-Health

Real-time Data Capture for Integration with LLMs to Enhance User Experience in IoT-based Personal Healthcare: Monitoring using Pulse and Temperature Sensors

Note: This is a Final Year Project for De Montfort University and Asia Pacific University of Technology and Innovation

GithubCoverPage

Whole Version

Overview

This project leverages real-time data capture from heart pulse and temperature sensors to integrate with Large Language Models (LLMs) to enhance the user experience in IoT-based personal healthcare. The system monitors environmental temperature and heart pulse, processes the data, and provides insightful feedback using LLMs.

Features

  • Real-time data capture from IoT sensors
  • Integration with OpenAI's Large Language Models
  • User-friendly interface using Soft UI Dashboard
  • Continuous monitoring and logging of health data
  • Insights and advice generation based on health data

Requirements

Software Requirements

  • Python 3.8+
  • Flask and additional Flask-related libraries
  • Paho-MQTT
  • OpenAI API
  • Bootstrap 5.1.3 (for spinner)
  • Arduino IDE for ESP32 setup

Hardware Requirements

  • ESP32 Development Board
  • Breadboard (for quick testing and non-soldered setups)
  • LM35 Temperature Sensor
  • Pulse Sensor Heart Rate (generic or labelled as XD-58C)
  • Connecting wires

Setup Instructions

Python and Flask Setup

  1. Clone the repository:

    git clone https://github.com/Mattjesc/RtDC-LLM-IoT-Health.git
cd RtDC-LLM-IoT-Health

  2. Create and activate a virtual environment:

    python -m venv venv
source venv/bin/activate   # On Windows: venv\Scripts\activate

  3. Install dependencies:

    pip install -r requirements.txt

  4. Set up OpenAI API:

    Replace 'YOUR_API_KEY_HERE' in app.py with your actual OpenAI API key.

    openai.api_key = 'YOUR_API_KEY_HERE'

    We are using GPT-4o mini in this case to save costs for testing, but you are free to use any LLM AI models, providers, or host locally. This FYP uses GPT-4o mini to prove the concept works.

  5. Run the application:

    python app.py

  6. Access the dashboard:

    Open your web browser and go to http://127.0.0.1:5000/.

Arduino Setup

  1. Install the ESP32 board in Arduino IDE:

    • Open Arduino IDE and go to File > Preferences.
    • In the Additional Boards Manager URLs field, add the following URL: 
      https://dl.espressif.com/dl/package_esp32_index.json
    • Go to Tools > Board > Boards Manager.
    • Search for ESP32 and install the esp32 package by Espressif Systems.

  2. Select the correct board:

    • Go to Tools > Board and select ESP32 Dev Module.

  3. Install required libraries in Arduino IDE:

    • PubSubClient: For MQTT communication.
    • PulseSensor Playground: For handling the pulse sensor.
    • WiFi.h Library: For Wi-Fi connectivity.

    You can install these libraries via the Arduino Library Manager.

  4. Open the main.ino file in Arduino IDE:

    cd RtDC-LLM-IoT-Health/main
open main.ino

  5. Upload the code to your ESP32:

    Ensure your ESP32 is connected to your computer, select the correct board and port, then click on the upload button in the Arduino IDE.

Dependencies

  • Flask
  • Paho-MQTT
  • OpenAI
  • Other Flask-related libraries

requirements.txt:

flask==2.2.5
Werkzeug==2.3.7
jinja2==3.1.2
flask-login==0.6.2
flask_migrate==4.0.4
WTForms==3.0.1
flask_wtf==1.2.1
flask-sqlalchemy==3.0.5
sqlalchemy==2.0.21
email_validator==2.0.0
python-dotenv==0.19.2
blinker==1.4
WTForms-Alchemy==0.18.0
flask-restx==1.0.3
gunicorn==20.1.0
Flask-Minify==0.37
Flask-Dance==7.0.0
PyJWT==2.4.0
Flask==2.2.5
paho-mqtt==1.6.1
openai==0.10.2

Usage

  • Real-time Data Capture: The system captures real-time data from heart pulse and temperature sensors using MQTT protocol.
  • Data Logging: Logs the data and displays it on the dashboard.
  • LLM Integration: Sends the captured data to OpenAI's LLMs to get insightful advice.
  • User Interface: Uses Soft UI Dashboard for an intuitive and visually appealing interface.

Soft UI Dashboard

The user interface is built using Soft UI Dashboard, which includes:

  • soft-ui-dashboard.min.css: Core CSS file for Soft UI Dashboard.
  • nucleo-icons.css and nucleo-svg.css: Icon libraries.
  • soft-ui-dashboard.min.js: Core JavaScript file for Soft UI Dashboard.

For more details, visit the Soft UI Dashboard repository.

Disclaimer

This project is limited by financial constraints, and the results may vary with higher quality sensors and equipment. In this implementation, we are using the GPT-4o mini model to save costs for testing purposes.

License

This project is licensed under the MIT License. See the LICENSE file for more details.

Contributing

Contributions are welcome! Please open an issue or submit a pull request for any improvements or bug fixes.

Acknowledgements