Mental Workload Assessment Using Deep Learning Models From EEG Signals: A Systematic Review
DOI: 10.1109/tcds.2024.3460750
archive: archived pipeline: cataloged
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This systematic review and meta-analysis addresses the growing application of deep learning models for assessing mental workload (MWL) using electroencephalography (EEG) signals. The study is motivated by the critical role of MWL in information systems, where accurate assessment is necessary to optimize human performance, prevent cognitive overload or underload, and enhance user experience. While EEG is preferred for its non-invasive nature and high temporal resolution, the field lacks standardized protocols for data preprocessing and model evaluation. The paper aims to identify opportunities, challenges, and best practices in this domain by investigating three specific research questions: the input formulations used for training deep neural networks, appropriate cross-validation procedures for EEG signals, and the types of MWL classification problems addressed. The authors conducted a comprehensive literature search across six major databases (ACM Digital Library, IEEE Xplore, ScienceDirect, Scopus, Springer Link, and Wiley Online Library) using customized search strategies tailored to each platform’s syntax. Following a PRISMA-guided screening process, the initial pool of 3,220 articles was reduced to 108 relevant studies that reported original research on deep learning for EEG-based MWL classification. The review categorizes these studies into five distinct problem types: within-subject, cross-subject, cross-session, cross-task, and combined cross-task and -subject. The analysis focuses on signal preprocessing, feature engineering, and model training methodologies, with a particular emphasis on the integrity of temporal sequences in EEG data. Key findings indicate that while deep learning offers promising potential for dynamic MWL classification, real-world applications remain limited due to small sample sizes, lack of population diversity, and inconsistent preprocessing standards. A critical methodological flaw identified is the shuffling of data before splitting into training and test sets, which disrupts the temporal sequence of EEG signals and inflates accuracy inaccuracies. The review highlights that time-series cross-validation or leave-session-out approaches better preserve temporal integrity, leading to more reliable model performance evaluations. Furthermore, the study notes that self-reporting methods, such as the NASA-TLX, are often used to label data but are subject to subjective biases and can themselves increase participant workload. The significance of this work lies in its provision of a comprehensive framework for evaluating EEG-based MWL assessment. By identifying the specific challenges associated with different cross-validation strategies and classification problem types, the review paves the way for more accurate and reliable research in information systems. The authors conclude that enhancing universal preprocessing standards and adopting rigorous temporal validation techniques are essential for developing adaptive information systems that can dynamically align with users’ cognitive states, thereby improving productivity, precision, and user satisfaction in safety-critical and interactive environments.
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | Crossref | — | — | 1 | 2026-08-09 |
| archive | success | unpaywall | — | — | 2 | 2026-08-09 |
| extract | success | cached | — | — | 4 | 2026-08-23 |
| clean | success | clean | — | — | 1 | 2026-08-09 |
| chunk | success | chunk | — | — | 1 | 2026-08-09 |
| enrich | failed | — | — | — | 1 | 2026-08-09 |
| promote | success | — | — | — | 1 | 2026-08-09 |
| summarize | success | llm | qwen3.8-27b-gittensor | summ-v5 | 3 | 2026-08-23 |
| tag | success | vector_similarity | — | — | 10 | 2026-08-11 |
| verify | success | — | — | — | 2 | 2026-08-09 |
Summary generated by qwen3.8-27b-gittensor on 2026-08-23; verification: pending re-verification.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Empirical Findings: physiological data