According to Statistics Korea, by 2025, South Korea will enter a "super-aged society," with over 20% of the population aged 65 or older. Consequently, traffic accidents caused by elderly drivers are becoming a significant social problem, transcending ...
According to Statistics Korea, by 2025, South Korea will enter a "super-aged society," with over 20% of the population aged 65 or older. Consequently, traffic accidents caused by elderly drivers are becoming a significant social problem, transcending the simple personal injury issue. However, the current driver's license aptitude test system for elderly drivers applies a uniform renewal cycle based on a physical age threshold of 75 years and focuses on assessing fragmented physical functions such as vision. This system presents fundamental limitations in accurately assessing the complex driving abilities of elderly drivers.
This study aimed to develop a scientific "aptitude test assessment model" that comprehensively reflects not only the physical function of elderly drivers but also their cognitive abilities and behavioral characteristics. Based on this model, we propose a practical "conditional driver's license system" that can ensure public traffic safety without infringing on the mobility rights of older drivers. To this end, empirical data from 1,356 drivers aged 75 and older who visited the Ansan Driver's License Examination Center in the Seoul metropolitan area and renewed their aptitude test in 2025 were collected and analyzed in-depth.
This study's analytical methodology attempted a multifaceted approach to overcome the limitations of a single method. Logistic regression, a traditional statistical technique, was used to explain causal relationships between variables and linear trends in risk. Random Forest, a cutting-edge machine learning technique, was adopted to maximize accident prediction accuracy by learning nonlinear interactions. Furthermore, to reflect normative values and field expertise not revealed in the data, Analytic Hierarchy Process (AHP) was conducted with a group of transportation experts, and a hybrid modeling approach was applied to integrate the results of these three methods.
The main findings of this study are as follows: First, we found that the accident risk of older drivers is closely related to cognitive decline (dementia screening test scores) rather than simply age due to physical aging. Random forest analysis, a machine learning technique, yielded a predictive accuracy of 89.34%, with dementia screening test scores (significance 55.2%) and hearing ability (27.7%) identified as the most critical variables in determining accident occurrence. Furthermore, logistic regression analysis statistically demonstrated (p<.001) that each unit lower in education completion grade (worse grade) increased the accident risk by approximately 1.33 times.
Second, by combining statistical explanatory power, machine learning predictive power, and expert insights, we developed the Elderly Driver Aptitude Score (EDAS) model, which is rated on a 100-point scale. The final scoring criteria were set at 45 points for the dementia screening score, 25 points for education completion, 20 points for hearing, and 10 points for vision. This significantly strengthened the discrimination power of cognitive risk factors (dementia and hearing) while also adjusting for the importance of vision, an essential prerequisite for driving.
Third, based on the EDAS model derived in this study, the elderly driver population was categorized into three groups: normal (80 points or higher), conditional license (60-79 points), and failing (less than 60 points).
This study is academically unique in that it combines expert opinions with objective data, such as individual driver characteristics and accident-causing factors, to ensure both objectivity and rationality in the evaluation criteria. Furthermore, the conditional license system proposed in this study is expected to be a practical policy that harmoniously addresses the dual values of "guaranteeing seniors' right to mobility" and "ensuring traffic safety."