By Jon Markt, Contributing Author
In December, the National Highway Traffic Safety Administration announced that traffic fatalities had declined for the 10th consecutive quarter. However, more than 120 people die each day in the United States as a result of crashes, according to the Centers for Disease Control and Prevention.
Tens of thousands of people still lose their lives each year from traffic crashes. Even more are seriously injured.
Reducing these numbers requires a Safe System Approach: a multi-pronged plan to safety that focuses not just on infrastructure improvements but also addresses driver behaviors and characteristics.
We can design the safest road and safest vehicles in the world, but an unsafe driver will still be a danger to themselves and others. Effecting this change requires effort in multiple areas, including road design, enforcement, driver behavior and even post-crash care.
One key is to understand the people behind the wheel. States are piloting programs that seek to develop a better understanding of who is driving and their impact on safety. But they’re using that information in different ways, demonstrating that there are alternative methods to address this concern.
In Minnesota, advanced statistical modeling is using driver characteristics and behavior to predict the locations of future fatal and serious injury crashes, giving safety professionals enhanced insights into where to prioritize interventions. This predictive approach seeks to deploy strategic interventions to prevent crashes before they occur.
The effort is still in process, but initial results are encouraging. And it is emblematic of an ongoing shift in approach — targeting changes in driver behavior as a complement to ongoing infrastructure improvements. The ultimate goal remains the same: driving toward zero deaths on our roads.
Predicting Crashes
The Minnesota Department of Transportation (MnDOT) uses crash and citation data to better understand driver behaviors and factors that are associated with fatal and serious injury crashes. Using these associations, MnDOT hopes to use historical information to predict locations where fatal and serious crashes are likely to occur.
Beginning in 2021, a multiphase project has gathered and analyzed a wide range of data to determine where fatal crashes were most likely to occur. A unified database that included typical roadway infrastructure information, census data, weather data, crash data, third-party speed and traffic flow data, and Minnesota State Patrol citation data was combined with a geographic information system base layer that allowed the team to access specific details associated with each crash location.
The initial exploratory phase used unstructured statistical modeling to predict which road segments would be the site of a fatal or serious injury crash in one district. The analysis identified previously overlooked patterns and relationships, and it predicted (with 61% success) which road segments would be the site of a fatal or serious injury crash.
Importantly, it suggested that some road segments were higher risk than previously considered, revealing new opportunities to prevent or eliminate fatal crashes.
In the next phase, the proof-of-concept was expanded and a refined approach predicted fatal and serious injury crashes using two structured and sophisticated models. Ultimately, an approach based upon negative binomial regression was chosen as the final model due to ease of use and interpretability.
In analyzing road segments, the model correctly predicted the exact number of severe and fatal crashes 59% of the time and was within one crash 89% of the time.
In creating and validating the model, the team encountered one somewhat unexpected result: Infrastructure variables played a relatively small role in predicting serious crashes. Instead, driver characteristics — speeding, seatbelt use, age, impaired driving — were the largest predictors of a serious injury or fatal crash.
MnDOT’s strong history of deploying proactive low cost and systematic safety strategies could be one reason for this, but it is also an indication of the importance of prioritizing driver behavior as well as infrastructure improvements.
The advantages of this system over previous methods of determining where to focus safety efforts are twofold: First, now that the model has been created and validated, rolling it out is relatively simple because of the state’s robust data availability. Second, it can prompt a wide range of solutions, not just infrastructure changes.
A recent interaction between stakeholders at a workshop offers one anecdotal example of these solutions. After the model identified a road segment as high-risk, particularly because of a history of dangerous driver behavior, the state highway patrol representative said they would like to do more enforcement, but there was no place for officers to park.
A MnDOT representative attending the workshop responded, saying that if the patrol only needed a pad for a car to park on, that could be done quickly, allowing enforcement to begin soon after.
The model has currently been applied to two of Minnesota’s transportation districts, but with good results may be expanded to cover the entire state. In the meantime, efforts continue to gather more data to provide further refinements and ever more accurate models in the future.
With the model created, the current phase is focused on implementation. An interactive tool that incorporates the analysis is in the final stages of development and being piloted with safety professionals. This will allow agency leaders and others invested in safety in the state to easily see where improvements could be made and explore options for those solutions.
The tool provides a single reference point that safety professionals can use to coordinate efforts. It can take what is a large, complex issue and quickly provide a focus on the highest risk corridors.
When used as part of a highly collaborative safety culture such as MnDOT, the tool can prompt a variety of safety awareness activities and coordination between engineering, law enforcement, emergency responders and other safety professionals.
Importantly, the data is updated regularly, providing new information on high priority corridors and allowing action plans built around the data to measure progress.
The list of actions taken can be tracked as they are implemented, providing ongoing feedback and keeping safety professionals across the state apprised of efforts being undertaken.
From Data to Action
Many DOTs are already collecting a great deal of data on crashes and the drivers involved. But gathering the data is only the first step. It must be analyzed and acted upon.
That action can take multiple forms, but it provides the chance to better understand the drivers involved in serious and fatal crashes. And understanding the drivers then allows agencies to target interventions that will better reach those drivers.
Minnesota’s program will not solve the driver safety conundrum on its own. But it is one piece of addressing a larger problem and building a road system that is safer for all drivers. RB
Jon Markt is HDR’s transportation safety program manager and worked on the MnDOT predictive safety program, in addition to many other safety-focused efforts across the U.S.
