Related Experiment Video
Updated: Aug 9, 2025
09:32
Cortical Source Analysis of High-Density EEG Recordings in Children
Published on: June 30, 2014
21.4K
On the Influence of Aging on Classification Performance in the Visual EEG Oddball Paradigm Using Statistical and
Nina Omejc1,2, Manca Peskar3,4, Aleksandar Miladinović5
1Department of Knowledge Technologies, Jožef Stefan Institute, 1000 Ljubljana, Slovenia.
Life (Basel, Switzerland)
|February 25, 2023
Summary
Brain-computer interfaces (BCIs) using electroencephalogram (EEG) data face challenges with age-related variability. Temporal EEG features show better classification performance and are less impacted by age than statistical event-related potential (ERP) features.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Signal Processing
Background:
- Electroencephalogram (EEG) is a common input for brain-computer interfaces (BCIs).
- Age-related variability in event-related potentials (ERPs) presents a challenge for EEG-based BCIs.
- ERPs are frequently used as primary signal features in EEG BCI applications.
Purpose of the Study:
- To investigate the impact of aging on EEG signal classification for BCIs.
- To compare the effectiveness of temporal versus time-independent statistical ERP features for BCI classification across age groups.
- To evaluate how different classifiers are affected by age-related differences in EEG features.
Main Methods:
- A visual oddball study was conducted with 27 young and 43 older healthy adults.
- 32-channel EEG data were collected while participants passively viewed frequent and rare stimuli.
- Two types of datasets were created: one with amplitude and spectral features, and another with statistical ERP features.
Main Results:
- Linear classifiers demonstrated the best performance among nine tested classifiers.
- Classification performance varied significantly between the temporal and statistical ERP feature datasets.
- Temporal features yielded higher maximum individual performance scores, lower variance, and were less affected by age-related differences.
Conclusions:
- Feature extraction and selection are critical for robust BCI performance, especially considering age-related variability.
- Temporal EEG features are more resilient to age-related performance degradation in BCIs compared to statistical ERP features.
- The choice of classifier and its feature ranking mechanism influence the impact of aging on BCI performance.
