Non-intrusive Drowsiness Detection Techniques and Their Application in Detecting Early Dementia in Older Drivers
DOI: 10.1007/978-3-031-18458-1_53
archive: archived pipeline: cataloged
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper addresses the critical safety issue of drowsy driving, which accounts for a significant portion of road accidents, particularly among older drivers who are at higher risk for fatigue and early-stage dementia. The study aims to identify practical, low-cost, and non-intrusive methods for detecting driver drowsiness to alert drivers promptly and reduce accident rates. The research specifically explores the application of these detection techniques in monitoring the driving behaviors of older adults to aid in the early identification of dementia and Alzheimer’s disease. The methodology involves a comprehensive review of 567 papers published after 2015, narrowed down to 15 highly relevant studies selected from major academic databases. The authors categorize existing drowsiness detection techniques into two primary groups: behavioral-based and vehicular-based methods. Behavioral techniques utilize non-intrusive visual data, such as eye tracking, Eye Aspect Ratio (EAR), PERCLOS (Percentage of Eye Closure), yawning detection via Convolutional Neural Networks (CNN), and head pose estimation using depth cameras. Vehicular techniques rely on semi-intrusive data, including lane detection, steering wheel angle variability, and yaw angle time-series analysis. The paper compares the accuracy and limitations of these methods, noting that while behavioral methods generally offer higher accuracy, they are susceptible to environmental factors like lighting conditions. Key findings indicate that behavioral-based detection is more effective than vehicular-based methods for non-intrusive monitoring. Specific results from reviewed studies include an 88.75% average accuracy for fuzzy K-nearest neighbor eye tracking, 93% accuracy for SVM-based Eye Aspect Ratio classification, and up to 98.81% accuracy for multi-task CNNs detecting yawning. Vehicular methods, such as steering wheel velocity analysis and lane heading difference, show promise but are often affected by external variables like weather and traffic. The authors also present a preliminary design for a video-based system intended for older drivers, which integrates telemetry and dual-camera video systems (Driver State Monitoring and front-facing) to collect behavioral indices such as eye closure duration, head turning, and lane crossing. Initial data from ten senior drivers over several months revealed that indices like prolonged eye closure and head distraction were the most significant markers, while other metrics like smoking or phone usage were negligible. The significance of this work lies in its bridge between general drowsiness detection and specific geriatric health monitoring. By establishing that behavioral cues are superior for non-intrusive applications, the paper supports the development of robust, low-cost in-vehicle systems. The proposed architecture, which uses AI algorithms to analyze log files from in-car cameras, offers a scalable solution for tracking driving behavior over time. This approach not only enhances road safety by alerting drowsy drivers but also provides a non-invasive tool for healthcare professionals to monitor cognitive decline in older populations, potentially facilitating earlier intervention for dementia.
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. Discovered via author_sweep_intake on 2026-05-28.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | author_sweep | — | — | 2 | 2026-05-28 |
| archive | success | manual_pmc_pow_fetch | — | — | 35 | 2026-08-22 |
| extract | success | cached | — | — | 4 | 2026-08-23 |
| clean | success | clean | — | — | 1 | 2026-06-25 |
| chunk | success | chunk | — | — | 1 | 2026-06-25 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-25 |
| enrich | success | semantic_scholar | — | — | 9 | 2026-06-21 |
| promote | success | — | — | — | 1 | 2026-06-04 |
| summarize | success | llm | qwen3.8-27b-gittensor | summ-v5 | 2 | 2026-08-23 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-25 |
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).
- Methodological Resource: tool software