Enhancing driving safety with customized music: A simulator study using entropy weight and Bayesian Networks

Introduction: Listening to music while driving is a common practice. Extensive research has explored its effects on driving performance, with a growing consensus suggesting that the optimal complexity of music varies depending on different driving scenarios to maintain drivers’ arousal levels. However, these optimal levels can vary significantly among individuals. This study investigates how drivers with different characteristics respond to sudden events under various musical conditions. Methods: We integrated eight driving-related ability traits into a comprehensive construct titled “driving characteristics.” Twenty-eight drivers were categorized into three groups (qualified, good, and excellent) and completed driving tasks under various musical conditions defined by tempo (fast/slow) and lyrics (with/without). Driving performance was assessed using multiple indicators and synthesized into a composite score using the Entropy Weight Method. Bayesian Networks were utilized to analyze collision probabilities and identify critical influencing factors. Results: Significant main effects were found for both lyrics (p = 0.01) and driving characteristics (p < 0.01). Music without lyrics (0.73 ± 0.02) demonstrated superior performance compared to music with lyrics (0.68 ± 0.02). The excellent group (0.79 ± 0.03) significantly outperformed both the good (0.66 ± 0.02) and qualified (0.66 ± 0.02) groups. An interaction effect showed that slow-tempo music outperformed fast-tempo music when lyrics were present. Bayesian Networks indicated that the impact of music on overspeed and collision probabilities varied based on driving characteristics. Conclusions: While drivers with superior driving characteristics can better manage complex music, individual differences in driving strategies can lead to performance variability even within the same characteristic group. Drivers should prioritize safe driving and adapt their music choices to align with their capabilities. Practical Applications: These findings may help vehicle manufacturers and music platforms enhance safety and performance by recommending personalized music and improve drivers’ responses to sudden incidents.