EEG-Based Closed-Loop Neurofeedback for Attention Monitoring and Training in Young Adults

Bingbing Wang, Zeju Xu, Tong Luo, Jiahui PanJournal of Healthcare Engineering 2021 · DOI

TL;DR

This paper introduces an improved Random Forest (IRF) algorithm for more accurate, five-level EEG-based attention monitoring. They combine this with a closed-loop neurofeedback system and three novel serious games to train sustained, selective, and focus attention in young adults, demonstrating significant improvements in attention control.

What Problem Does This Paper Address?

Young adults today face increasing challenges with attention due to constant media interaction and multitasking. Existing methods for monitoring and training attention often suffer from low accuracy, poor attention levels, and inefficiency. This paper aims to bridge this gap by developing a more robust EEG-based system that can accurately monitor attention across multiple levels and provide effective, engaging training through neurofeedback and serious games.

Methods

The core of this work lies in two main modules: attention monitoring and attention training.

For attention monitoring, the authors propose an Improved Random Forest (IRF) algorithm. This IRF classifier is enhanced through:

  • Grid Search Optimization: To fine-tune hyperparameters of the Random Forest model.
  • Multiple Cross-Validation (10-fold): To ensure robust performance and prevent overfitting.
  • Five Attention Levels: Moving beyond simpler classifications to a more nuanced understanding of attention states.

The IRF algorithm itself is a standard ensemble method, but its application here is tailored for EEG signal analysis. The process involves:

  1. EEG Data Preprocessing: Using a wavelet transform (specifically Daubechies 4) to decompose EEG signals into subbands and extract power spectral densities (PSDs) from specific frequency bands (delta, theta, alpha, beta, gamma).
  2. Feature Extraction: Selecting 20 EEG features from these subbands.
  3. IRF Classification: Training the IRF model on these features to classify attention into five levels.

For attention training, the paper utilizes a closed-loop neurofeedback system integrated with three serious games:

  • Tug of War: For sustained attention.
  • Adventures of Birds: For selective attention.
  • Greedy Jelly: For focus attention.

The games are designed to be engaging for young adults, with real-time feedback provided via a progress bar reflecting their attention levels. The neurofeedback loop uses the IRF algorithm's output to dynamically adjust game difficulty or character actions, creating an interactive training environment.

Datasets

  • Personal EEG Concentration Tasks dataset: Used for offline analysis to verify the effectiveness of the proposed attention monitoring method. This dataset involved 80 subjects.
  • Newly collected EEG data: From 10 young adults during online experiments for both attention monitoring and training.

Key Insights and Ideas

  • Nuanced Attention Monitoring: Classifying attention into five levels, rather than just a few, provides a more detailed picture of an individual's cognitive state, which is crucial for personalized training.
  • IRF for EEG: The IRF algorithm, with its grid search and cross-validation, proves to be a powerful tool for extracting meaningful patterns from noisy EEG data, outperforming other standard algorithms.
  • Serious Games as Training Tools: Integrating neurofeedback into engaging game mechanics is a promising avenue for making attention training more effective and enjoyable, especially for young adults.
  • Synergy of Monitoring and Training: Using the attention monitoring output directly as a control parameter within the games creates a truly closed-loop system, enhancing the training feedback loop.
  • Multi-faceted Attention Training: Addressing sustained, selective, and focus attention within a single framework offers a more holistic approach to improving overall cognitive control.

Key Results

  • Attention Monitoring Accuracy: The proposed IRF-based attention monitoring method achieved an average accuracy of 79.34% in offline analysis, outperforming state-of-the-art algorithms like SVM, KNN, AdaBoost, ET, and RF. In online experiments, the IRF algorithm achieved an average accuracy of 79.06%.
  • Attention Training Effectiveness: After attention training using the serious games, young adults showed statistically significant improvements (p < 0.05) in four key indicators:
  • Schulte times: Decreased significantly, indicating faster task completion.
  • Win times: Decreased significantly, showing improved performance in sustained attention tasks.
  • Game scores: Increased significantly, reflecting better selective attention.
  • Skill times: Increased significantly, demonstrating enhanced focus attention.
  • Neurofeedback Impact: A comparison experiment without neurofeedback showed that while familiarity with games improved performance, the rates of change in indicators were significantly lower compared to sessions with neurofeedback.

Limitations and Future Work

  • Small Sample Size: The authors acknowledge that the primary limitation is the small number of subjects examined during the online experiments.
  • External Factors: Habits, motivations, and mental statuses of the subjects were not fully considered, which could influence results.
  • Data Volatility: The inherent volatility of conductive media in BCIs and the challenge of collecting sufficient datasets for young adults are noted.

Future Work: * Data Fusion: Incorporate EEG with other physiological signals (e.g., facial expressions, verbal speech) to further improve attention monitoring accuracy. * 3D Serious Games: Develop more advanced 3D serious games for attention training. * Larger Datasets: Collect more attention data from young adults to build more comprehensive datasets.

Summary updated

Abstract

Attention is an important mechanism for young adults, whose lives largely involve interacting with media and performing technology multitasking. Nevertheless, the existing studies related to attention are characterized by low accuracy and poor attention levels in terms of attention monitoring and inefficiency during attention training. In this paper, we propose an improved random forest- (IRF-) algorithm-based attention monitoring and training method with closed-loop neurofeedback. For attention monitoring, an IRF classifier that uses grid search optimization and multiple cross-validation to improve monitoring accuracy and performance is utilized, and five attention levels are proposed. For attention training, we develop three training modes with neurofeedback corresponding to sustained attention, selective attention, and focus attention and apply a self-control method with four indicators to validate the resulting training effect. An offline experiment based on the Personal EEG Concentration Tasks dataset and an online experiment involving 10 young adults are conducted. The results show that our proposed IRF-algorithm-based attention monitoring approach achieves an average accuracy of 79.34%, thereby outperforming the current state-of-the-art algorithms. Furthermore, when excluding familiarity with the game environment, statistically significant performance improvements (<i>p</i> < 0.05) are achieved by the 10 young adults after attention training, which demonstrates the effectiveness of the proposed serious games. Our work involving the proposed method of attention monitoring and training proves to be reliable and efficient.
BibTeX
@article{wang2021eeg,
  author = {Bingbing Wang and Zeju Xu and Tong Luo and Jiahui Pan},
  title = {{EEG-Based Closed-Loop Neurofeedback for Attention Monitoring and Training in Young Adults}},
  journal = {Journal of Healthcare Engineering},
  year = {2021},
  doi = {10.1155/2021/5535810},
}
Figure