Crash Risk Predictors in Older Drivers: A Cross-Sectional Study Based on a Driving Simulator and Machine Learning...

Area of Science:

• Gerontology
• Traffic Safety
• Neuroscience

Background:

• Driving ability relies on integrated motor, visual, and cognitive functions.
• Safe driving requires timely information processing and appropriate responses to traffic scenarios.
• Assessing older drivers is crucial for understanding age-related driving risks.

Purpose of the Study:

• To evaluate older drivers using a driving simulator.
• To identify motor, visual, and cognitive variables affecting safe driving.
• To determine the primary predictors of traffic crashes in older adults.

Main Methods:

• Cluster analysis (K-Means) to group drivers by characteristics.
• Random Forest algorithm to predict traffic crashes.
• Analysis of motor, visual, and cognitive functions in 100 older drivers (mean age 72.5 years).

Main Results:

• Two clusters of drivers were identified, with no significant differences in crash or infraction rates between them.
• Cluster 1 drivers had higher age, driving time, and braking time compared to Cluster 2.
• The Random Forest model accurately predicted road crashes (r=0.98, R²=0.81), with advanced age and functional reach test results being major risk factors.

Conclusions:

• Advanced age and impaired functional reach are significant predictors of traffic crashes in older drivers.
• Driving simulation and predictive modeling can effectively identify risks associated with aging and driving.
• While clusters showed some differences in driving behavior metrics, advanced age and functional reach were the most critical crash predictors.