Bayer Professor of Chemical Engineering and Biotechnology
MIT

He’s a good professor," says Gregory Stephanopoulos’s wife, Maria Flytzani-Stephanopoulos, who, like her husband, is a chemical engineer, though in a different area of specialization and at a different university, Tufts, down the road from Cambridge, Mass.-based MIT, where her husband teaches. "It is important to think and be creative, and he tries to instill that in his students. They’re free to run and create."

That’s probably because Stephanopoulos, 52, likes to do the same thing. He got creative when he saw how difficult it was for pharmaceutical companies to get a handle on antibiotic production. Biological compounds aren’t predictable. You know how much steel goes into a Ford, but there are always wild variations in what emerges from the slurry of raw materials used to grow antibiotics. There are simply too many variables. And that meant uncertain supply for the pharmaceutical companies.

So in 1994 Stephanopoulos applied IT to the problem. Along with some of his grad students, he developed a data mining program, called Dbminer, which sifts the information kept about each batch of slurry for behavior patterns inside those dark tanks. The program makes it much easier to identify problems?or highlight things that work well?significantly reducing the unpredictability of the process. He also built a version of the program for some major pharmaceutical companies, such as Eli Lilly and Merck, and is in the process of examining commercial versions of the software to sell to other companies.

It’s a lot more complicated than sifting through cash register receipts to discover that men buy diapers and beer at the same time on Thursdays, but the mathematical underpinnings of his solution are widely applicable and could produce drastic increases in speed in everything from medical diagnostics to drug research.