Sign in Subscribe
Machine learning

Machine Learning Research Reveals Effective Driver Drowsiness Detection Method

Strategic Insights

— 1 min read

A new study on machine learning has yielded promising results in detecting driver drowsiness, a leading cause of traffic accidents. Researchers from the American University of the Middle East have proposed a system that uses heart rate variability (HRV) analysis to assess driver drowsiness, leveraging wearable devices equipped with photoplethysmography (PPG) sensors. The study demonstrates the effectiveness of the Random Forest (RF) classifier in detecting drowsiness, achieving a testing accuracy of 86.05%, precision of 87.16%, recall of 93.61%, and F1-score of 89.02%.

Key Takeaways:

  • The researchers identified driver drowsiness as a critical issue in transportation systems, resulting from factors such as intoxicated driving, fatigue, and sleep deprivation.
  • The proposed system integrates wearable devices with PPG sensors, transmitting data to a smartphone and then to a cloud server for analysis.
  • Two novel algorithms were developed to segment and label features, predicting drowsiness levels based on HRV derived from PPG signals.
  • The Random Forest (RF) classifier achieved the highest testing accuracy (86.05%), precision (87.16%), recall (93.61%), and F1-score (89.02%) among six classification algorithms applied.
  • The Support Vector Machine with Radial Basis Function (SVM-RBF) also showed strong generalization performance, with a testing F1-score of 87.15%.
  • The study's findings suggest that HRV-based drowsiness detection systems can be integrated into Advanced Driver Assistance Systems (ADAS) to enhance driver safety.
  • The research includes contributions from Zakwan AlArnaout, Chamseddine Zaki, Yehia Kotb, Mouhammad AlAkkoumi, and Nour Mostafa.

Statistics:

  • Testing accuracy: 86.05% (Random Forest (RF) classifier)
  • Precision: 87.16% (Random Forest (RF) classifier)
  • Recall: 93.61% (Random Forest (RF) classifier)
  • F1-score: 89.02% (Random Forest (RF) classifier)
  • Mean change between training and testing datasets: -4.30% (Random Forest (RF) classifier)
  • Testing F1-score: 87.15% (Support Vector Machine with Radial Basis Function (SVM-RBF))

Sources:

  • Exploiting heart rate variability for driver drowsiness detection using wearable sensors and machine learning. Scientific Reports, 2025;15(1):24898.
  • Scientific Reports (www.nature.com/srep/)
  • American University of the Middle East
  • Nature Portfolio (www.nature.com/)

Read more