Wearable Device-Based Physiological Signal Monitoring: An Assessment Study of Cognitive Load Across Tasks

Ling He, Yanxin Chen, Wenqi Wang, Shuting He, Xiaoqiang Hu

Abstract

This study employs cutting-edge wearable monitoring technology to conduct high-precision, high-temporal-resolution (1-second interval) cognitive load assessment on electroencephalogram (EEG) data from the FP1 channel and heart rate variability (HRV) data of secondary vocational students. By jointly analyzing these two critical physiological indicators, the research delves into their application value in assessing cognitive load among secondary vocational students and their utility across various tasks. The study designed two experiments to validate the efficacy of the proposed approach: Initially, a random forest classification model, developed using the N-BACK task, enabled the precise decoding of physiological signal characteristics in secondary vocational students under different levels of cognitive load, achieving a classification accuracy of 97%. Subsequently, this classification model was applied in a cross-task experiment involving the National Computer Rank Examination (Level-1), demonstrating the method's significant applicability and cross-task transferability in diverse learning contexts.

Introduction

In educational settings, particularly vocational training where hands-on skills are paramount, understanding a student's "cognitive load"—the mental effort required to process information—is vital. If the load is too low, the student may become disengaged; if it is too high, they may struggle to retain information. Traditional measurement methods, such as the NASA-TLX subjective survey or bulky laboratory equipment, are often intrusive or only provide retrospective data. This research addresses this gap by developing a non-invasive, portable, and highly granular method to track mental effort in real-time, allowing for potential "just-in-time" instructional adjustments in a classroom environment.

Methodology

The proposed system utilizes a multimodal approach, integrating two distinct physiological streams to capture the state of the nervous system during cognitive activity.

  1. EEG (Electroencephalogram): Data is collected from the Fp1 channel (located on the forehead) using a BrainLink Pro wearable. The analysis covers eight frequency bands, from Delta (deep sleep/repair) to Gamma (high-level processing).
  2. HRV (Heart Rate Variability): Derived from R-R intervals (the time between heartbeats), HRV serves as a proxy for the autonomic nervous system's response to mental stress.

The processing pipeline involves energy-to-voltage transformation, outlier pruning via the K-Nearest Neighbors (KNN) algorithm, and classification using a Random Forest model. This model was selected for its robustness in handling the complex, non-linear relationships inherent in physiological data.

System architecture for cognitive load classification.

Experimental Design and Validation

The study involved 30 secondary vocational students and was structured into two phases to test both accuracy and transferability across different types of mental work.

  • Experiment 1 (The Benchmark): Participants performed the N-BACK task, a standard cognitive psychology test where difficulty is precisely controlled by increasing memory requirements. This phase generated 15,360 data samples used to train and refine the model.
  • Experiment 2 (Real-World Application): To test "cross-task" transferability, the model was applied to students taking the National Computer Rank Examination (Level-1), involving spreadsheet operations and file management. This generated 28,800 samples.

Validation was performed using Leave-One-Group-Out Cross-Validation (LOGO-CV) to ensure the model's generalizability across different individuals rather than just memorizing specific participant patterns.

Experimental workflow illustrating the training on N-BACK tasks and cross-task prediction for the computer exam.

Key Results and Insights

The study yielded several significant findings that bridge the gap between laboratory research and classroom application:

  • High Precision: In controlled settings, the Random Forest model achieved a classification accuracy of up to 97%.
  • Cross-Task Transferability: A model trained on generic memory tasks (N-BACK) successfully predicted mental effort during a real-world computer exam. Under high-difficulty conditions, students were correctly identified as being in a high-load state 80.93% of the time.
  • Single-Channel Efficiency: One of the most significant takeaways is that robust assessment does not require a full EEG cap; a single forehead channel (Fp1) combined with heart rate data is sufficient for high-accuracy classification.
  • Temporal Resolution: The system provides a 1-second resolution, allowing researchers to identify exactly which specific exam questions or operations cause a spike in mental effort.

Conclusion

This research demonstrates the feasibility of using portable wearables for high-resolution cognitive load monitoring in educational settings. The findings suggest that physiological signatures of cognitive load are consistent across different types of mental work, enabling the use of generic training tasks for real-world applications. While future work will explore more diverse vocational tasks and real-time "closed-loop" instructional feedback, the current findings provide a strong foundation for optimizing teaching resource allocation based on objective physiological data.

Summary updated

Abstract

This study employs cutting-edge wearable monitoring technology to conduct high-precision, high-temporal-resolution (1-second interval) cognitive load assessment on electroencephalogram (EEG) data from the FP1 channel and heart rate variability (HRV) data of secondary vocational students. By jointly analyzing these two critical physiological indicators, the research delves into their application value in assessing cognitive load among secondary vocational students and their utility across various tasks. The study designed two experiments to validate the efficacy of the proposed approach: Initially, a random forest classification model, developed using the N-BACK task, enabled the precise decoding of physiological signal characteristics in secondary vocational students under different levels of cognitive load, achieving a classification accuracy of 97%. Subsequently, this classification model was applied in a cross-task experiment involving the National Computer Rank Examination (Level-1), demonstrating the method's significant applicability and cross-task transferability in diverse learning contexts. Conducted with high portability, this research holds substantial theoretical and practical significance for optimizing teaching resource allocation in secondary vocational education, as well as for cognitive load assessment methods and monitoring. Currently, the research findings are undergoing trial implementation in the school.
BibTeX
@misc{he2024wearable,
  author = {Ling He and Yanxin Chen and Wenqi Wang and Shuting He and Xiaoqiang Hu},
  title = {{Wearable Device-Based Physiological Signal Monitoring: An Assessment Study of Cognitive Load Across Tasks}},
  year = {2024},
  eprint = {2406.07147},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2406.07147},
}
Figure