/Parkonix

Unveiling the Tremors, A Reliable Algorithm with 83% Accuracy for Detecting Parkinson's Disease through Spiral/Wave Sketch Images.

Primary LanguagePython

Parkonix

Unveiling the Tremors, A Reliable Algorithm with 83% Accuracy for Detecting Parkinson's Disease through Spiral/Wave Sketch Images.


Parkinson's Disease


Parkonix is an advanced algorithm developed for the accurate detection of Parkinson's disease. Based on the research article "Automated Parkinson's Disease Detection Based on Handwriting Movement" by Rigas et al., this repository contains the implementation of Parkonix using Teachable Machine.


⭐ Overview

Parkinson's disease is a neurodegenerative disorder that affects motor functions, leading to tremors, stiffness, and impaired movement. The research presented in the mentioned article explores the use of spiral and wave sketch images to develop a robust algorithm for Parkinson's disease detection. Parkonix leverages these sketch images to train a machine learning model, achieving an impressive accuracy rate of 83%.


⭐ Features

  • Parkinson's disease detection based on spiral and wave sketch images.
  • Utilizes Teachable Machine, a user-friendly platform for training machine learning models.
  • Achieves an accuracy rate of 83%, providing reliable results.
  • Fast and efficient detection process.
  • Easy to use and integrate into existing applications.

⭐ How to Use

  1. Ensure that you have the necessary dependencies installed.
  2. Obtain a dataset of spiral and wave sketch images.
  3. Train the Parkonix algorithm using Teachable Machine, providing the dataset as input.
  4. Save the trained model and export it in a compatible format.
  5. Integrate the Parkonix algorithm into your application and utilize it for Parkinson's disease detection.

⭐ Model is developed on Teachable Machine

Teachable Machine

Teachable Machine Demo

Teachable Machine is an innovative platform developed by Google that allows users to build custom machine learning models without the need for coding or extensive technical knowledge. The platform utilizes a simple drag-and-drop interface that allows users to input data and train machine learning models quickly and easily.

Teachable Machine has several features that make it an ideal tool for building custom machine learning models. Firstly, it allows users to train models using a variety of data types, such as images, sounds, and sensor data. Secondly, it provides users with the ability to choose from a range of pre-built machine learning models, such as image classification and sound recognition, or to build custom models from scratch.


⭐ To run these scripts, you need the following installed:

  1. Python 3
  2. The python libraries listed in requirements.txt
    • Try running "pip3 install -r requirements.txt"

Step 1: Clone this repository

Run:

git clone https://github.com/SaiJeevanPuchakayala/Parkonix

Step 2: Navigate to the Parkonix directory

Run:

cd Parkonix

Step 3: Install the python libraries

Run:

pip install -r requirements.txt

Step 4: Run the streamlitApp.py file

Run:

streamlit run streamlitApp.py

⭐ Streamlit Deployment Configurations:

[theme]
base="dark"

[browser]
gatherUsageStats = false

⭐ Few images illustrating model performance

Confusion Matrix

Teachable Machine

Accuracy Per Class

Teachable Machine

Accuracy Per Epoch

Teachable Machine

Loss Per Epoch

Teachable Machine


⭐ Deployment References:

  1. https://30days.streamlit.app/
  2. https://docs.streamlit.io/streamlit-community-cloud/get-started/deploy-an-app
  3. https://streamlit-cloud-example-apps-streamlit-app-sw3u0r.streamlit.app/?hsCtaTracking=28f10086-a3a5-4ea8-9403-f3d52bf26184|22470002-acb1-4d93-8286-00ee4f8a46fb
  4. https://docs.streamlit.io/library/advanced-features/configuration

⭐ Note:

If you find my GitHub repository useful, why not give it a star? It's like giving a little virtual high-five that makes my day!