Official implementation of the paper "Bag of Samplings for Parkinson's Disease Diagnosis based on Recurrent Neural Networks"
- Install the requirements with
pip install -r requirements.txt(requires Python 3.6) - Run
./data/get_handpd2.shto download the data from the Recogna Laboratory servers - The training protocols are based in the work by Afonso et al.
[1], in which two training regimes are considered: a) 50% of the data is used for training and 50% for testing b) 75% of the data is used for training and 25% for testing. In our case we need to validate the hyperparameters, hence we employ the data partitions in the table below.
| Training regime | Train | Validation | Test |
|---|---|---|---|
| a | 50% | 0% | 50% |
| b | 75% | 0% | 25% |
| a (ours) | 40% | 10% | 50% |
| b (ours) | 65% | 10% | 25% |
These splits can be generated with the script split.py in data/. Since the random seed is fixed, the output of the script should always be the same. Nevertheless, the splits used in our experiments are detailed in the file data_splits.py.
- You can reproduce the paper results using the script
models/train_models.py. Running it with the-hoption shows how to set the hyperparameters. - If you wish to run the sampling interval powerset selection, use
models/poewrset_selection.pythe output of this script is persisted in the disk.
To reproduce the experiment depicted in Figure 6 in our paper, run experiments/models/learning_curve.py and to generate the plots, use experiments/models/plot_learning_curves.py.
If you use our work in your research, please cite the following paper:
Ribeiro, L.C., Afonso, L.C., Papa, J.P.. Bag of samplings for computer-assisted parkinson’s disease diagnosis based on recurrent neural networks.
@article{ribeiro2019bos,
title = "Bag of Samplings for computer-assisted Parkinson's disease diagnosis based on Recurrent Neural Networks",
journal = "Computers in Biology and Medicine",
volume = "115",
pages = "103477",
year = "2019",
issn = "0010-4825",
doi = "https://doi.org/10.1016/j.compbiomed.2019.103477",
url = "http://www.sciencedirect.com/science/article/pii/S0010482519303476",
author = "Luiz C.F. Ribeiro and Luis C.S. Afonso and João P. Papa",
}
[1] L. C. Afonso, G. H. Rosa, C. R. Pereira, S. A. Weber, C. Hook, V. H. C. Albuquerque, and J. P. Papa, “A recurrence plot-based approach for parkinson’s disease identification,” Future Generation Computer Systems, vol. 94, pp. 282 – 292, 2019.