Last month, the Federal Motor Carrier Safety Administration (FMCSA) published a long-awaited notice expanding the popular Crash Preventability Determination Program (CPDP). Though this is good news for the industry, there is also frustration about how long it is taking FMCSA to adjudicate these requests and raised concerns that the new crash scenarios may further exacerbate the problem.

Many have expressed hope that FMCSA research into using artificial intelligence (AI) to make these determinations will improve the process. Unsurprisingly, newly published research says, “not so fast.”

The research, conducted by the Virginia Tech Transportation Institute (VTTI) and published in early December, expresses optimism that a model can be built to reduce total time spent by DataQ analysts by approximately 50 percent. The VTTI team demonstrated this by building a human-in-the-loop AI model that reads Texas police accident reports (PARs) and makes preliminary determinations within 30 seconds. These are later reviewed and confirmed by human DataQ analysts. The problem? It only works in Texas.

That’s because every state has its own version of the police accident report. This despite federal efforts to construct a Model Minimum Uniform Crash Criteria designed to ensure States are collecting a “standardized data set describing motor vehicle crashes.” While States may be collecting similar or identical data elements, they’re doing so in different forms and media, complicating FMCSA’s efforts to improve program throughput.

The research concludes that if a model like that built in Texas was replicated in the top 15 states, approximately 70 percent of all determination requests would benefit from the efficiency improvements. This is not good enough and potentially not worth the effort.

In STC’s mind, the solution is simple: standardize the PAR across all states. We understand there are administrative, and potentially legal, barriers to this in many, if not all, states.  However, in our opinion, the juice would definitely be worth the squeeze. The benefit would be greater than improvements to the CPDP too. Crash data would be easier to extrapolate and understand on the federal level, unlocking new insights into crash causation and avoidance, thereby improving safety. While getting all the states to sing from the same book of Christmas carols could be a monumental task, the benefits are obvious and worth the effort.