Can science predict a traffic accident will occur on a certain stretch of highway just before it happens?

Mohamed Abdel-Aty thinks so.

Abdel-Aty, associate professor of transportation engineering at the University of Central Florida in Orlando, is working on a study he says could save lives on highways across the United States. In September, he began collecting and analyzing traffic speed and volume along a 40-mile stretch of Interstate 4 from Walt Disney World to Lake Mary. He uses sensors buried in the highway every half-mile which transmit readings every half-minute.

That data will be compared with the locations and times of actual traffic accidents. By analyzing the speed-volume data an hour before an accident occurred, Abdel-Aty hopes to design a computer program that can spot trends that foretell an impending crash.

``If we know the ingredients for an accident, then we can prevent the accident,'' he said.

Breaking The Cycle

The sensors could be connected to variable message signs on the highway. Instead of the traditional warnings like ``Expect Delays,'' the signs would flash ``Slow Down'' or ``Exit Here'' when the computer spotted conditions that were ripe for a traffic crash.

``We can break the cycle that leads up to an accident,'' Abdel- Aty said.

Funded by the Department of Transportation in September, the study will continue for two years. Imagine the data stream from sensors every half-mile, along a 40-mile stretch of highway, sending readings every half-minute, and you'll have an idea of the challenge Abdel-Aty faces.

``We've been looking at an enormous amount of data,'' he said.

Drunks And Seat Belts

The study has the potential to save a lot of lives and injuries in Florida, where motorists rack up 250,000 traffic accidents every year. The University of Central Florida, with its Center for Advanced Transportation Simulation Systems, is among a handful of universities designated as federal transportation research centers, receiving $1 million a year in grants.

One of the leading traffic-safety analysts in Florida, Abdel-Aty also did an analysis on driving under the influence. Among his findings: Contrary to conventional wisdom, middle-aged drivers were more likely than teenage drivers to drive drunk. Teens, although frequently speeding when involved in accidents, typically were not driving drunk, according to the study.

In an earlier study on seat-belt use, Abdel-Aty showed that drivers with more education and higher incomes are more likely to use seat belts.

But it's his study on how to avoid accidents before they happen that really has Abdel-Aty excited. ``If we can come up with a system that would flag a certain location when the ingredients for an accident comes up,'' he said, ``it can be applied on highways all over the country.''