Wearable biosensor monitoring of driver fatigue in intelligent transport systems: A systematic review and IoT framework

Vovk, Yaroslav; Vovk, Oleh; Holotenko, Oleksandr; Vovk, Yuriy; Pientak, Mariia · 2026 · Crossref

DOI: 10.14254/jsdtl.2026.11-1.03

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This systematic review addresses the critical safety gap in intelligent transport systems regarding driver fatigue, a factor implicated in up to 50% of severe collisions in Europe. Despite existing research, no unified framework previously integrated clinical physiological thresholds, wearable biosensor specifications, machine learning architectures, and Internet of Things (IoT) deployment within a regulatory-compliant system. The study aims to bridge fragmented engineering, clinical, and policy streams by proposing the Physiologically-Grounded Driver Monitoring (PGDM) conceptual framework. The authors conducted a systematic literature review adhering to PRISMA 2020 guidelines, searching five databases (IEEE Xplore, ScienceDirect, PubMed/PMC, Web of Science, and arXiv) for publications from 2019 to May 2026. From an initial 2,847 records, 43 peer-reviewed studies met inclusion criteria after quality assessment using the adapted Mixed Methods Appraisal Tool. The review empirically benchmarked methods using four open datasets: OpenDriver (81 drivers, ~4,600 hours of open-road ECG/IMU data), WACHSens (62 participants), DD-Database (10 participants, EEG/EOG/ECG), and AdVitam (346 participants). Key findings identify specific physiological biomarkers and technological solutions. EEG frontal theta power (4–8 Hz) anticipates behavioral drowsiness by 2–7 minutes, offering a crucial warning window. Validated heart rate variability thresholds include RMSSD below 20 ms and LF/HF ratios above 2.0, while PERCLOS exceeding 70% indicates severe drowsiness with 91% sensitivity. The review highlights obstructive sleep apnea as a dominant uncontrolled confound in current algorithms. Technologically, a hybrid CNN-LSTM-Attention architecture achieved 97.3% accuracy with 15–18 ms edge inference latency, while Transformer-based multimodal fusion demonstrated the lowest cross-subject performance degradation (5.2 percentage points). The significance of this work lies in the PGDM framework, the first published architecture to derive alert levels directly from validated clinical thresholds while ensuring compliance with EU regulations, including GSR 2019/2144, the AI Act 2024/1689, MDR 2017/745, and GDPR. This provides transport engineers and fleet operators with a standards-compliant blueprint for real-time monitoring. The authors conclude by identifying seven priority research gaps, most notably the need for prospective cohort studies that stratify fatigue algorithms by polysomnography-confirmed obstructive sleep apnea status to improve clinical validity.

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.

Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified_with_issues.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• drowsiness detection algorithms
• truck driver fatigue
• dms validation
• dms regulation
• drowsiness

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
• Methodological Resource: validation psychometrics, tool software