In-vehicle Sensing and Data Analysis for Older Drivers with Mild Cognitive Impairment.

Area of Science:

• Neuroscience
• Gerontology
• Transportation Safety

Background:

• Driving is a complex cognitive task sensitive to age-related changes.
• Early detection of cognitive decline, such as mild cognitive impairment (MCI), is crucial for maintaining independence and safety.
• Unobtrusive monitoring of daily driving can reveal subtle cognitive shifts.

Purpose of the Study:

• To design low-cost in-vehicle sensing hardware for high-precision positioning and telematics data.
• To identify key indicators of early cognitive changes in daily driving.
• To apply machine learning for detecting early warning signs of cognitive impairment during normal driving.

Main Methods:

• Development of affordable in-vehicle sensors for data collection.
• Statistical comparison of driving patterns between drivers with and without MCI.
• Implementation of machine learning models (Random Forest) to identify influential factors.

Main Results:

• Drivers with MCI demonstrated smoother and safer driving patterns compared to cognitively unimpaired individuals.
• This suggests compensatory strategies by drivers aware of their cognitive status.
• Random Forest models highlighted the number of night trips, total trips, and education level as significant predictors.

Conclusions:

• Daily driving behavior analysis, using unobtrusive sensing and machine learning, can aid in the early detection of cognitive impairment.
• Individuals with MCI may consciously modify driving to mitigate risks.
• Driving data, trip frequency, and educational background are important factors in cognitive change assessment.