Talking to James Stewart of Saint John is like waking up on the set of Minority Report, the Tom Cruise movie about technology that prevents a crime before it happens.

Only the crime, in this case, is drunk driving.

Stewart is the founder and chief executive officer of RTV, which stands for Repeat Target Vehicles, a startup developing predictive analytics software that can identify for police departments people who are likely to drink and drive.

The idea is to let the drivers know they are being watched so they will refrain from driving drunk. If the company succeeds, it will save money and, more importantly, lives.

 “The biggest factor in the decision to drink and drive is the belief that you won’t get caught,” said Stewart, a former crime analyst now completing a PhD in information technology. RTV aims to give police the tools to notify people who frequently drive drunk (or their families) so they will decide not to get behind the wheel after drinking.

The company is worth noting this week because it is showing up at two big events this month.

It has entered the New Brunswick Innovation Foundation’s Breakthru competition and was at the event’s boot camp Saturday; and on Thursday, Stewart will attend the Big Data Congress in Saint John, a new event to encourage data analytics in the region.

RTV exemplifies the efforts behind the Big Data segment in the region because it aims to mine data to help public bodies improve efficiency and cut costs.

Police departments frequently receive and record calls from people who have seen a car they are sure is being driven by someone who is drunk, but the departments don’t always have the resources to follow up.

What Stewart’s technology does is take each reported vehicle and compare it with other data on the owner, such as his or her criminal record and other information.

The logarithms can identify which people are likely to drink and drive again.

 “It always starts with the vehicle,” said Stewart. “We look at the number of calls and the nature of the calls, and then we look at the characteristics of the owner.”

This means police officers can visit these people and present them with a written notice saying that local police are aware that their vehicle is high-risk and that they will be looking for it.

Stewart stressed the goal is not to target anyone for a police hunt so much as to persuade individuals not to drive drunk in the first place.

Of course, the pain being addressed is profound. Mothers Against Drunk Driving estimates that in 2009 in Canada, crashes involving drunk drivers killed 1,074 and injured 303,000 people, and cost the country more than $20 billion.

Stewart has teamed up with serial entrepreneurs Stephen Goddard and Keith Dunphy. They have validated the system. In a New Brunswick test run, it identified seven car owners who were likely to drive drunk, and four in the ensuing months were involved in crashes.

The company is looking for $250,000 to $750,000 in funding, with which it hopes to roll out and test its prototype.