.png){kind=icon}
.png){kind=icon}
Return
Drunk Driver Detection Using Multiple Non-Invasive Biosignals
.png){kind=icon}
OA
AI
DOI：10.3390/s25051281.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This study aims to decrease the number of drunk drivers, a significant social problem. Traditional methods to measure alcohol intake include blood alcohol concentration (BAC) and breath alcohol concentration (BrAC) tests. While BAC testing requires blood samples and is impractical, BrAC testing is commonly used in drunk driving enforcement. In this study, the multiple biological signals of electrocardiogram (ECG), photoplethysmogram (PPG), and electrodermal activity (EDA) were collected non-invasively and with minimal driver restraint in a driving simulator. Data were collected from 10 participants for approximately 10 min at BrAC levels of 0.00%, 0.03%, and 0.08%, which align with the latest Korean drunk driving standards. The collected data underwent frequency filtering and were segmented into 30 s intervals with a 10 s overlap to extract heart rate variability (HRV) and pulse arrival time (PAT). Using more than 10 machine learning algorithms, the classification accuracy reached 88%. The results indicate that it is possible to classify a driver's level of intoxication using only non-invasive biological signals within a short period of about 30 s, potentially aiding in the prevention of drunk driving.
Keywords:
drunk driving detection
non-invasive biological signal monitoring
machine learning
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.