What is Network Screening?
Network screening is part of any modern road safety management system. It is the practice of systematically reviewing risk data from across the network to identify locations with abnormally high risks and to identify systemic risk factors linked to a significant number of fatalities and injuries. A network screening exercise yields a list of the locations with highest risks and a list of major contributing risk factors. The objective of network screening is to identify emphasis areas for directing investments in location specific or system-wide road safety improvements.
In North America, network screening is typically done using techniques outlined in the highway safety manual (HSM). When following HSM methods, there are several steps involved:
The first step of HSM network screening is create safety performance functions (SPFs). SPFs are equations estimated through generalized linear regression modelling that define the mean number of injuries and fatalities that are typical for various facility types and traffic volumes, and they also define the degree of variance to be expected around that mean.
The second step of HSM network screening is to create an estimate of expected safety for every location in a network. The estimate of expected safety is obtained as a weighted combination of the SPF prediction and the actual crash record observed at the location. This weighted combination uses an Empirical Bayes statistical process with the objective of smoothing out the ‘noise’ in crash counts which can vary randomly.
The third step of HSM network screening is to rank the locations according to their expected safety level. Sometimes a derivative of expected safety is used such as “Excess” crash count (the degree to which “Expected” safety exceeds the SPF).
An emerging trend in North America is to go beyond the HSM’s focus on location specific risks to use the network screening process to identify systemic risk factors. The identification of systemic risk factors is accomplished by conducting a methodical distributional analysis of the collision data.
The next generation of network screening tools are expected to leverage surrogate safety metrics extracted from video analytics. Companies such as MicroTraffic are offering cities new capabilities to extract high resolution, high sample size near-miss risk data from traffic cameras across their network.
Network screening results should be updated every 3-5 years.
Through investments in scalable and repeatable data science tools, Fireseeds North has been able to cut the cost of a typical network screening assignment by a factor of almost 10. Contact Fireseeds North to discuss an update to your network screening, or to set up HSM methods for the first time: email@example.com or 204-996-9550.