Exploiting heart rate variability for driver drowsiness detection using wearable sensors and machine learning

AlArnaout, Zakwan; Zaki, Chamseddine; Kotb, Yehia; AlAkkoumi, Mouhammad; Mostafa, Nour · 2025 · Crossref

DOI: 10.1038/s41598-025-08582-2

archive: archived pipeline: cataloged verified

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 driver drowsiness, a leading cause of traffic accidents that impairs response time, awareness, and judgment. Motivated by the limitations of existing detection methods—such as the impracticality of subjective self-reports, the environmental sensitivity of vehicle-based metrics, and the invasiveness of traditional physiological sensors like EEG—the authors propose a non-invasive, real-time drowsiness detection system. The study focuses on exploiting Heart Rate Variability (HRV) derived from Photoplethysmography (PPG) signals collected via wearable sensors. The research aims to validate the feasibility of using HRV as a physiological marker for drowsiness and to develop a deployable framework that integrates wearable data acquisition, cloud-based processing, and machine learning for timely driver alerts. The methodology involves an end-to-end system model where wearable devices equipped with PPG sensors transmit data via Bluetooth to a smartphone and subsequently to a cloud server. The authors developed two novel algorithms: one for dynamic, segment-wise annotation of real-time input and another for windowed HRV feature extraction from PPG signals. This pipeline allows for continuous drowsiness estimation. To evaluate the system, the researchers utilized real-driving data and applied six supervised machine learning classification algorithms to labeled datasets. The performance of these models was assessed using standard metrics, including accuracy, precision, recall, F1-score, and runtime, to determine the most effective algorithm for real-world deployment. The results demonstrate that the Random Forest (RF) classifier achieved the highest performance among the tested algorithms. Specifically, RF attained a testing accuracy of 86.05%, precision of 87.16%, recall of 93.61%, and an F1-score of 89.02%. It also exhibited the smallest mean change between training and testing datasets (-4.30%), indicating strong robustness and generalization capabilities. The Support Vector Machine with Radial Basis Function (SVM-RBF) also showed strong generalization, with a testing F1-score of 87.15% and a mean change of -3.97%. These findings confirm that HRV features extracted from PPG signals are effective indicators of driver drowsiness when processed through appropriate machine learning classifiers. The significance of this work lies in its practical application for enhancing Advanced Driver Assistance Systems (ADAS). By leveraging commercially available wearable technology and cloud computing, the proposed system offers a scalable, non-intrusive solution for monitoring driver fatigue. The high performance of the Random Forest classifier suggests that HRV-based detection can provide reliable, timely alerts to drivers, potentially reducing accidents caused by drowsiness. The study contributes a validated, deployable framework that bridges the gap between physiological monitoring and real-time safety interventions, offering a more accessible alternative to invasive or environment-dependent detection methods.

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.

Topics

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

• drowsiness detection algorithms
• workload measurement
• 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