SYSTEMS

Mathematical modeling application expected to save countless lives  

More than two million Americans suffer from a form of a cardiac arrhythmia called fibrillation, the medical term for an irregular electrical activity in the heart that precludes the blood pumping.  While the origins of the disease are yet unclear, its effects are unmistakable.  Certain forms of cardiac fibrillation are the leading cause of stroke and may result in heart failure and sudden death.  Investigators are intensely striving to better understand the causes of cardiac arrhythmias using a variety of experimental and computational approaches.  Now, the University of Michigan’s Center for Arrhythmia Research is using SiCortex high-productivity computing (HPC) systems to conduct pioneering research aimed at predicting and preventing fibrillation, ultimately saving lives.

SiCortex computers, widely recognized for their energy-efficiency and small space requirements, are widely used by leading academic institutions and national laboratories around the world to facilitate important research in a variety of other areas, including engine design, aerodynamics, weather pattern mapping, ecology, astrophysics and complex signal processing, among others.

“This important initiative demonstrates HPC’s far-reaching power to solve a wide array of problems, including those directly linked to medical conditions and treatments,” said Chris Stone, president and CEO of SiCortex. 

The University of Michigan project, led by Dr. Omer Berenfeld and funded in part by grants from the National Institutes of Health, the American Heart Association and the Heart Rhythm Society, runs mathematical modeling applications on two SiCortex systems: one that serves as the mainframe system and the other as an auxiliary development unit.  The numerical simulation studies produced so far provide valuable new insight by presenting data in stunning visual displays.

Historically, fibrillation was thought to be caused by random electrical disturbances in the heart.  New research, however, shows that the electrical activity during fibrillation is not completely random.  In fact, the seemingly chaotic disturbances are found to originate in many cases from organizing centers in the form of spiral vortices, also called rotors.  The University of Michigan team will study computer simulations generated by the SiCortex systems to evaluate the many combinations of variables that can cause rotors and determine their dynamics.

“We are studying the conditions of numerous variables that can interrupt the heart’s regular rhythm and lead to an arrhythmia,” said Dr. Berenfeld, the assistant professor of Internal Medicine and Biomedical Engineering at the Center for Arrhythmia Research.  “Through sophisticated mathematical modeling and parallel computations running on the SiCortex systems, we can reproduce the heart’s electrical activity to pinpoint more precisely where and how abnormal rhythms originate within the heart muscle.  Understanding the origin is an important step in determining the underlying cause of the problem.”

The computer simulations are studied together with electrical data collected from various modalities, including fluorescence imaging and electrodes threaded into the heart’s chambers.  To accurately simulate and analyze the fine characteristics of voltage and motion of ions that penetrate all the way to the molecular level, the researchers needed a system with fast and efficient connectivity.  The SiCortex systems were a natural choice, as they consist of processers linked together by a very fast communication fabric, making them ideally suited to run these complex simulations. 

“In addition to the performance considerations, the system needed to fit in a small computer room since the research lab lacks space big enough to house a large HPC system,” added Dr. Berenfeld.  “As medical researchers first and HPC users second, we needed a system that we could easily configure into an existing space, plug in and start using.  The SiCortex HPC systems also offered the highest performance per dollar compared to competing systems.”