Innovative ML-driven Recognition of EEG Patterns in Cognitive Training

Electroencephalogram (EEG) patterns depict electrical activity in the brain. They reveal insights into neurological functions, aiding in diagnosing conditions like epilepsy, sleep disorders, and brain injuries. The purpose of this research is to establish an innovative machine learning (ML)-driven recognition of EEG patterns in cognitive training. In this study, we propose an innovative Dynamic Artificial Rabbit Search-driven Advanced Bidirectional Long Short-Term Memory (DAR-ABLSTM) for robust classification of EEG patterns in cognitive training tasks. EEG was employed to investigate the impact of various forms of cognitive training on brain activity. We obtained EEG recordings from 50 healthy individuals during cognitive training and after a five-week programme. A signal processing procedure is employed to preprocess the obtained raw signal data. Our proposed model employs a novel approach stimulated by the foraging behavior of rabbits to enhance the classification of EEG patterns. We also conducted a t-test using SPSS analytical software to evaluate the pre- and post-cognitive training measures. The proposed recognition model is implemented in Python software. In the findings assessment phase, we effectively assess the performance of our proposed DAR-ABLSTM in classifying EEG patterns across multiple evaluation metrics, such as sensitivity (94.53%), accuracy (97.01%), F1-score (95.72%) and specificity (96.62%). Our experimental results demonstrate the capability and reliability of the proposed recognition in dynamic scenarios. The results of the analysis showed that both the negative and positive moods had significantly changed. The study suggests varying responses to different cognitive training methods.