This project aims to build multiple classifiers for the identification of Parkinson Disease symptoms using Machine Learning and provide those results to medical staff.
In this project we are using IMU (Inertial Measurement Unit) data to facilitate the Parkinson Disease diagnosis.
Here you could find the paper with all the research and design process: Link.
Project Implemented as Bachelor's thesis of:
- Jose García
- Julian Bolaños
Mentors:
- Andrés Navarro
- Nicolás Salazar
This project followed a custom version of CRISP-DM methodology, and the folder structure is explained below:
notebooks
: Jupyter notebooks used for experimentation, analysis,
modeling and development.
data
: Store all the raw, processed and intermediate data from IMUs
docs
: Documentation from src functions and project in general
results
: Stores graphics, metrics, weights and model's hyperparameters.
src
: Python scripts for data download, formating, preprocessing, model building and metrics calculation
enviroment
: Configuration files, API keys and environment variables.
To run any resource from this project we strongly recommend create an enviroment using
our .yml
file located in enviroment/
and install all the dependencies on it.