Nenad Filipovic's Coronary Calculus

The Serbian engineer's equations model blood flow and could predict heart attacks

The red and blue glow of computer animations fills a lab in the Institute for Basic and Physical Chemistry in Belgrade. It's a warm summer day in Serbia. That we are only 5 minutes from a spot where U.S. cruise missiles once slammed into ministry buildings doesn't much register with Nenad Filipovic. To be sure, like most Serbians, he hopes to put the past behind him and forge a bright future for this rapidly changing heart of the Balkans. But at the moment, he's hunched over his computer, trying to create a new kind of imaging for cardiologists.

The NATO bombs that rained on Serbia during its 1999 conflict with the breakaway province of Kosovo led to the collapse of the Miloševic government, sending Belgrade into chaos and Filipovic to Harvard. Five years ago, he came back as a bioengineering professor at the University of Kragujevac. At the Institute for Basic and Physical Chemistry, he helped start BioIRC, a spin-off that creates predictive software for visualizing the plaques that cause heart attacks and charting their future growth.

"We deal with small plaques," Filipovic says. "The point is, when does this plaque develop, and how much time does a patient need—one month, two months, six months, 10 years?" BioIRC is an interdisciplinary mashup of informatics and hemodynamics. CT and MRI imaging and catheter cameras build a precise geographic image of the individual patient's heart. Blood analysis shows lipids, hemoglobin, albumin, and cholesterol factors. Data-mining tools plumb genetic profiles, patient history, and lifestyle statistics. Vascular shear stress analysis adds precision. The data is crunched on the SEE-GRID (Southeastern European Grid), a computer network consortium of 13 countries. As yet, there is no one tool that measures all of these factors. It's a little bit like writing new software for each heart.

Predictive medicine is a growing but sometimes controversial field. The practice of modeling in itself can be controversial—after all, it didn't always work on Wall Street. But the results are already promising and may eventually affect more than just coronary plaque analysis.

BioIRC is doing clinical trials with 100 patients in Serbia. "We collect data, and we simulate," Filipovic says. "And we compare results over time." He says BioIRC can already predict plaque growth—not position, but growth—with 70 to 80 percent accuracy. Predictive medicine could also greatly improve the placement of medicated stents, drug-delivering mesh tubes that reinforce weak artery walls.

Other centers for predictive medicine include Stanford, the University of Texas, and Worcester Polytechnic. Filipovic is still a Harvard associate, but he has seen his department suffer tough cutbacks. "It's hard to be a scientist now. There's no money," he says. "It's hard here, too. But I see a chance for young people. The best students come to me, and they want to work. It's changed completely here. There's a vision. Not just here. I see changes in Europe."

This article originally appeared in print as "Coronary Calculus."

About the Author

Michael Dumiak is a science and technology reporter based in Berlin. This summer, in Belgrade to interview heart-imaging engineer Nenad Filipovic for the profile "Coronary Calculus", he wondered how this corner of Europe became a source of predictive software that visualizes the plaques that cause heart attacks. But, he says, it all made sense after a few meals. "Serbian cooking is...robust. This guy's going to have a lot of patients."