Jan Kalina

Dept. of Artificial Intelligence

Institute of Computer Science of the Czech Academy of Sciences

Location: Institute of Computer Science, Pod Vodárenskou věží 2, 180 00 Prague, Czechia

   Room 318

Date:   Tuesday 15 October 2024

Time: 13:00 CET

Abstract:

EEG signals are frequently utilized in neuroscience research and in corresponding biomedical applications to facilitate comparative analyses between two cohorts: patients and control subjects. However, numerous instances of incorrect application of multiple comparison procedures to EEG signals have been observed, such as the absence of prior utilization of multivariate tests or the selection of inappropriate methods. This paper offers an overview and discussion of multivariate two-sample tests and multiple comparison procedures, highlighting some nonparametric multivariate tests that have not been previously applied to EEG data. A particular study involving EEG signals obtained from patients with Alzheimer's disease and control individuals is analyzed. This seems to be the first EEG analysis systematically comparing different types of multivariate tests. Multivariate tests turn out to be significant and we also investigate the tests only for a selected channel or for a selected frequency. The best results are obtained with rank tests based on interpoint distances. 

Keywords: Two-sample test, inter-point distances, rank tests, multiple comparisons, exchangeability.

References: 

1. Kalina J., Janáček P. (2023): Testing exchangeability of multivariate distributions. Journal of Applied Statistics 50, 3142-3156.

2. Kalina J., Kukal J., Vyšata O. (2024+): Two-Sample hypothesis tests for the multivariate analysis of EEG signals. Submitted.

3. Marozzi M., Mukherjee A., Kalina J. (2020): Interpoint distance tests for high-dimensional comparison studies. Journal of Applied Statistics 47, 653-665.

4. Mukherjee A., Kössler W., Marozzi M. (2022): A distribution-free procedure for testing versatile alternative in medical multisample comparison studies. Statistics in Medicine 41, 2978-3002.