Washington, May 13 : Scientists have proposed a new approach to warn of an impending pandemic by detecting subtle signals in human behaviour.

"The goal is a public information and awareness system for pandemic with the same level of credibility, timeliness and visibility as storm-warning icons presented on television screens," said Barrett Caldwell, a Purdue University associate professor of industrial engineering.

The system works by monitoring `event phases' of human behaviour leading up to a pandemic, such as an increase in people purchasing flu-related medications or "foraging" on the Internet for certain types of information related to the flu.

"If you can recognize the triggers, the signals suggesting an event is likely to occur, you can start responding to it, gathering resources, preparing and mobilizing people. Our basic research idea could be used for any pandemic, or even other types of disasters," said Sandra K. Garrett, an assistant professor of industrial engineering at Clemson University.

The study shows how pre-pandemic events are separated into four categories of "human factors," or social behaviour: a period during which it is first possible to detect signals of an emerging pandemic; a time when it is possible to begin early efforts to prevent or mitigate spread; a time when it is critical to implement such measures; and a period when it is time to complete mitigation steps.

The method is an elaboration of "signal-detection theory," conceived decades ago.

"Normally, when psychologists study signal detection, they are looking at very rapid changes, like whether a tone changes, whether a light changes colour or turns on and off," Caldwell said.

The new approach proposes to make signal detection sensitive to more gradual events that are slower to develop.

"This is important because a pandemic is not a single point in time but a scenario that may take place in several waves over a period of months," he said.

"One of the challenges is that the way influenza spreads, you don''t know that someone''s sick until several days later, and by then they have had the opportunity to infect other people. At that point you have to project backward to see where people have first been sick and where certain flu-related events have happened. You are reactive, rather than proactive," he added.

The researchers envision a system that uses icons similar to those used to alert the public about an impending blizzard, hurricane or tornado.

The findings will be presented June 2 at the Industrial Engineering Research Conference in Miami. (ANI)