The little black boxes in airplanes provide useful information after a crash has taken place. Now researchers have devised a way to use black box info from planes that do not crash—to help prevent accidents from ever happening.

Some airlines already use a program that checks 88 flight parameters in the black box. But a Boeing 787 can record 2,000 flight parameters. So the researchers employed what’s called cluster analysis. They took data from 365 flights of Boeing 777s in a month of operations of a now-defunct European airline. They filtered the data into clusters, groupings of flights with similar patterns. Any data points outside of normal operations appeared as an anomaly.

The information generated identified three problematic flights. In one, for instance, the data showed that the pilot had trouble rotating the plane during takeoff.

The system can identify anomalies that suggest both physical and operational issues—such as those that caused a runway accident in Chicago in December 2005.

The researchers from M.I.T. and Spain will present their paper in October at the 30th Digital Avionics Systems Conference in Seattle. Because to prevent future accidents, you might have to think inside the box.

—Cynthia Graber

[The above text is a transcript of this podcast.]