Evaluation of Intoxication Level with EOG Analysis and Machine Learning: A Study on Driving Simulator.

Abstract

Driving after consuming alcohol can be dangerous, as it negatively affects judgement, reaction time, coordination, and decision-making abilities, increasing the risk of accidents and putting oneself and other road users in danger. Therefore, it is critical to establish reliable and accurate methods to detect and assess intoxication levels. One such approach is electrooculography (EOG), a non-invasive technique that measures eye movements, which has been linked to intoxication levels and holds promise as a method of estimating them. In recent years, machine learning algorithms have been utilized to analyze EOG signals to estimate various physiological and behavioural states. The purpose of this study was to investigate the viability of using EOG analysis and machine learning to estimate intoxication levels in a simulated driving scenario. EOG signals were measured using JINS MEME_R smart glasses and the level of intoxication was simulated using drunk vision goggles. We employed traditional signal processing techniques and feature engineering strategies. For classification, we used boosted decision trees, obtaining a prediction accuracy of over 94% for a four-class classification problem. Our results indicate that EOG analysis and machine learning can be utilized to accurately estimate intoxication levels in a simulated driving scenario.