We don’t need to drive for long before it becomes second nature. This conceals the fact that it’s a highly intricate activity that utilizes both sides of the brain and all four brain lobes.
This means tiny changes in our driving ability we wouldn’t be aware of could reflect future cognitive decline. If you can detect these changes, then treatment strategies can be employed early on. Thanks to artificial intelligence (AI) and machine learning this may now be possible.
Driving is an extremely complex activity and, as we reported in January, makes use of all four brain lobes - frontal, parietal, temporal, and occipital. These are all utilized to cope with the many cognitive challenges that come with driving.
Alzheimer’s Or Aging?
A weakening in the function of any one of these lobes could simply reflect aging. But then again, it could point to the very earliest signs of Alzheimer’s, well before drivers first notice a problem, which typically revolves around navigation; finding themselves getting lost even on familiar routes.
Scientists from Columbia University have been working on a reliable method to predict future cognitive problems in drivers using AI. Two years ago, using the model they created, they succeeded in predicting mild cognitive impairment (MCI) and dementia with high accuracy.
Data Is Gathered While Driving
For their 2021 study the Columbia team used information gathered from 2,977 active drivers aged 65 to 79 with no significant cognitive impairment or degenerative medical conditions when they were enrolled.
The data was captured through in-vehicle recording devices. This is called “naturalistic” because it reflects data captured in a real-world setting. This can be processed to measure driving performance in great detail.
The study ran from August 2015 through March 2019. During this time 33 people were diagnosed with MCI and 31 with dementia.
After developing and applying a machine-learning model called Random Forests, which is widely used in AI for classifying disease status, lead author, Professor Sharon Di summed up the findings.
“Based on [29] variables derived from the naturalistic driving data and basic demographic characteristics, such as age, sex, race/ethnicity and education level, we could predict mild cognitive impairment and dementia with 88 percent accuracy.”
This was an improvement on a previous model tested called Logistic Regression.
Predicts MCI And Dementia With 96 Percent Accuracy
Over the last two years the researchers have improved their game even further to produce a model that has even greater accuracy.
For the latest study they used the same group of 2,977 drivers but instead of 29 variables they constructed 200 variables, called digital markers, using data on the driver, the vehicle, and the environment captured by the in-vehicle recording devices.
After a reevaluation of the medical records, 36 participants were now diagnosed with MCI, eight with Alzheimer’s, and 17 with other or unspecified dementia.
This time they used a new machine-learning tool which they call interaction-based, building on a statistic named Influence Score. Fortunately, we don’t need to understand the incredible complexities involved in this method. What matters is what it can achieve.
The new learning model outperformed Random Forests and Logistic Regression models, with an accuracy of 96 percent in predicting MCI and dementia.
The Two Most Important Driving Markers
The team found that the two most influential driving variables were the right-to-left turn ratio and the number of hard braking events. “With advancing age, drivers make relatively fewer left turns and more right turns because left turns are riskier,” noted Professor Di.
Her colleague, Professor Guohua Li, explained, saying, “Our study indicates that digital markers embedded in routinely collected driving data can be used through innovative machine learning techniques as valid and reliable artificial intelligence for predicting mild cognitive impairment and dementia.”
These findings don’t surprise us. As we reported in January, after memory loss takes hold and progresses, driving safely becomes far more challenging.
Memory Loss Makes It Hard To Drive Safely
One study found that folks with Alzheimer’s had more than double the car wrecks per year as healthy people of the same age – 0.09 crashes per year compared with 0.04 crashes.
Another study showed that people with mild cognitive impairment and very mild dementia were impaired to the same level as 16- to 20-year-old drivers.
Not every person with Alzheimer’s disease or MCI will struggle to drive safely, however. Alzheimer’s has stages that can last for different lengths of time in different people, and a lot depends on the severity of the disease and which cognitive abilities are affected.
