Predicting Crashes

By February 25, 2019 No Comments

CMT’s mission is to make roads safer by making drivers better. Many of the things we talk about are intuitive, like distracted driving leads to crashes. The big questions come after this. How much does distraction impact crash risk? Can you predict if someone is likely to have a crash based on how often they use their phone? What about other factors like speeding, combined that with aggressive driving?

Intuitively it’s easy to understand the more someone engages in risky behavior, the worse the outcome could be for them. At CMT, our data scientists put that hypothesis to the test by building a sensor-based model that can predict the outcome of a crash.

We built this model in May 2018, and then began an effort to engage with insurance regulators in the US to approve the model’s usage by our insurance company partners. Today we announced CMT’s efforts to file our telematics scoring model has yielded an approval in 29 states. We undertook this effort to help our customers to deploy new rating models based on predictive factors of automobile crashes.

Our data comes from millions of trips and billions of miles collected from the DriveWell platform, which allows us to interpret data from sensors, smartphones, and our IoT Tag. We collect data on speeding, phone distraction, acceleration, harsh cornering, and hard braking. When combined with our Tag, we can detect crashes. The crashes we use in our scoring model are further confirmed to be losses.

The results from our model are significant -and they aren’t limited to insurance pricing.  As noted in today’s release, implementing telematics-based scoring models has public policy benefits as well. “Distracted driving is one of the most significant variables – with over 13x lift alone,” said Chief Technology Officer and Cofounder Hari Balakrishnan. “But we know telematics can change long-term driving behaviors; there’s an average 35 percent reduction in distracted driving after enrolling in DriveWell. Most people own smartphones now – meaning most people can participate in mobile telematics programs.”

Best-in-Class Scoring Models

Leaders in modeling and reducing phone distraction

It’s important to work with regulators to provide access to information that ultimately can lead to lower crash rates. We are excited about the progress we have made since we started this process in May 2018, and look forward to continuing our efforts to make roads safer by making drivers better.

The bottom line: we have seen the direct results our DriveWell platform can have on driving behavior. This model is one proof point. We are excited to deploy solutions that improve driver behavior and reduce risk; and will continue to build solutions that focus on our mission.