Date of creation: 15. 7. 2016
It is possible to download software here.
Publication to be cited
GALÁŽ, Z.; MEKYSKA, J.; MŽOUREK, Z.; KISKA, T.; SMÉKAL, Z.; REKTOROVÁ, I. Degree of Parkinson’ s Disease Severity Estimation Based on Speech Signal Processing. In Proceedings of the 39th International Conference on Telecommunication and Signal Processing, TSP 2016. 1. 2016. s. 1-4. ISBN: 978-1-5090-1287-9.
GALÁŽ, Z. Potential of Prosodic Features to Estimate Degree of Parkinson’ s disease severity. In Proceedings of the 22nd Conference STUDENT EEICT 2016. Brno: 2016. s. 533-537. ISBN: 978-80-214-5350- 0.
In the field of biomedical signal processing (e.g. pathological speech signal processing), often it is necessary to use regression analysis to evaluate the sufficiency of selected objective parametrization techniques to quantify and predict the health state of the analysed subject on selected subjective rating scales assessing severity of the disease. For this purpose, the researchers nowadays use multiple sophisticated regression techniques and metrices to evaluate regression models. Regression analysis tool (RAT) provides an easy and fast way to train and test variety of regression models.
Software RAT provides 5 possible regression techniques: Classification and Regression Trees (CART), Ordinal Regression (OR), Linear Regression (LR), Support Vector Machines Regression (SVMR), Gaussian Process Regression (GPR). It also provides 10 metrics evaluating quality of regression process: Gini Index (GI), Absolute Error (AE), Mean Absolute Error (MAE), Squared Error (SE), Mean Squared Error (MSE), etc.). The package also provides the function for the cross-validation process (k-fold, leave-one-out).
The RAT software is written in the MATLAB programming environment. The testing scripts demo.m can be used to test functionality of the software. The script loads data from the test_reg.mat file, which includes the parametrization matrix “feat_matrix”: rows are determined for the observations; columns are determined for the parameters and the vector of labels “labels” (for the regression processes it is the continuous numeric scale).
This work was supported by projects AZV 16-30805A, LO1401, COST IC1206 a FEKT-S-14-2335. The described research was performed in laboratories supported by the SIX project; the registration number CZ.1.05/2.1.00/03.0072, the operational program Research and Development for Innovation.