Voice Analysis for Parkinson's Disease

Codebase for the papers "An Analysis of Features for Machine Learning Approaches to Parkinson's Disease Detection" and "Cross-Lingual Transferability of Voice Analysis Models: a Parkinson's Disease Case Study" (extended abstract). For all the references, contributions and credits, please refer to the papers.

This code was initially developed as part of the M.Sc. Thesis in Computer Science and Engineering "Cross-Lingual Transferability of Voice Analysis Models: a Parkinson's Disease Case Study" (executive summary). The M.Sc. degree was released by the Dipartimento di Elettronica, Informazione e Bioingengeria (DEIB) of the Politecnico di Milano University (PoliMI). The Thesis was supervised at PoliMI by the staff of the ARCSlab.

Repository structure

This repository is organised into four main directories:

• experiments/ contains the directories to host:
  • results of the experiments;
  • experiment configuration dumps.
• notebooks/ contains the Jupyter notebooks to do:
  • data exploration;
  • results analysis.
• resources/ contains:
  • directories to host the dialogue corpora used in the experiments, and the references to download them;
  • directory to host the YAML configuration files to run the experiments.
• src/ contains modules and scripts to:
  • fit and evaluate models;
  • extract audio features;
  • preprocess data.

For further details, refer to the README.md within each directory.

Installation

To install all the required packages, instead, run the following commands:

# Create anaconda environment
conda create -n prkns python=3.10
# Activate anaconda environment
conda activate prkns
# Install packages
pip install -r requirements.txt

Finally, download and initialise the RNNoise submodule

# Install RNNoise submodule 
git submodule init; git submodule update

Once the RNNoise submodule is initialised, follow the instructions in rnnoise/README.md to install it.

To add the source code directory to the Python path, you can add this line to the file ~/.bashrc

export PYTHONPATH=$PYTHONPATH:/path/to/voice_analysis_parkinson/src

Finally, make sure that FFMpeg is installed and available on $PATH.

Data preprocessing

There are two scripts for data preprocessing in ./src/bin/utils/:

1. ./src/bin/utils/preprocess_data.py can be used to denoise the file within a directory.
2. ./src/bin/utils/prepare_data.py can be used to apply segments splitting (given the split timings).

Run experiments

There is a script to run the experiments, it expects to have ./src in the Python path and all data sets to be downloaded and placed in the ./resources/data/raw/ directory.

To run the script in foreground:

python ./src/bin/main.py --configs_file_path ./resources/configs/path/to/config.yaml

To run the script in background:

nohup python ./src/bin/main.py --configs_file_path ./resources/configs/path/to/config.yaml > experiment_"$(date '+%Y_%m_%d_%H_%M_%S')".out &

Contributors

• Claudio Ferrante ([email protected])
• Vincenzo Scotti ([email protected])

References

If you are willing to use our code or our models, please cite our work through the following BibTeX entries:

@inbook{ferrante-etal-2023-analysis,
	address = {Boca Raton},
	author = {Ferrante, Claudio and Menon, Bindu and Pillai, Anitha S. and Sbattella, Licia and Scotti, Vincenzo},
	booktitle = {Machine Learning and Deep Learning in Natural Language Processing},
	doi = {10.1201/9781003296126-13},
	editor = {Pillai, Anitha S. and Tedesco, Roberto},
	isbn = {978-1-003-29612-6},
	pages = {169--183},
	publisher = {CRC Press},
	title = {An Analysis of Features for Machine Learning Approaches to Parkinson's Disease Detection},
	url = {https://doi.org/10.1201/9781003296126-13},
	year = {2023},
}

@conference{ferrante-scotti-2023-cross,
	author = {Ferrante, Claudio and Scotti, Vincenzo},
	booktitle = {Booklet of abstracts -- SPOKEN LANGUAGE IN THE MEDICAL FIELD: Linguistic analysis, technological applications and clinical tools},
	pages = {40--42},
	title = {Cross-lingual transferability of voice analysis models: a Parkinson's Disease case study},
	url = {https://www.aisv.it/lecce2023/BookOfAbstract_Lecce2023.pdf#page=40},
	year = {2023}
}