'Chaos' Theory Empowers VA And University Of Florida Researchers To Predict Epileptic Seizures

December 8, 1999

GAINESVILLE—Inspired by an intriguing mathematical concept known as chaos theory, researchers at the University of Florida Brain Institute and the Malcom Randall Veterans Affairs Medical Center in Gainesville have developed a technique for predicting some types of epileptic seizures minutes to hours before they begin.

Their work, under development for more than a decade and now the basis of a U.S. patent application, opens the door to the creation of implantable devices that can detect signs a seizure is approaching and deliver medication, or electrical or magnetic stimulation to try to prevent it.

“We had determined some years ago that there was a theoretical potential for predicting seizures,” said Dr. J. Chris Sackellares, a VA neurologist and a UF professor of neurology, neuroscience and biomedical engineering who has been researching temporal lobe epilepsy with Leonidas D. Iasemidis, a VA research engineer and UF research assistant professor of electrical and computer engineering, and neuroscience. “But it has only been in the past year that we have really been able to demonstrate that we can do so reliably. And it has only been recently that we realized that the state of transition to a seizure could last for many hours.”

Sackellares and Iasemidis, whose studies have been supported by grants from the National Institutes of Health and the U.S. Department of Veterans Affairs, are scheduled to present related research today (12/8) at the national meeting of the American Epilepsy Society in Orlando.

The researchers began to suspect in 1988 that the emerging field of chaos science would be able to shed light on epilepsy. In a nutshell, chaos theory offers a mathematical approach for seeing a kind of order in events that previously had appeared to be random. Early in the 1990s, the approach enabled Sackellares and Iasemidis to be the first to identify the existence of a pre-seizure transition period.

In their early research the scientists were looking for such transitions to occur seconds to minutes before a seizure began. But in the past year, by analyzing electrical activity in the brain recorded for a 10-day period, they have identified a warning stage developing anywhere from minutes to many hours ahead of time.

Their technique involves using sophisticated mathematical formulas to sort through the brain’s complex electrical signals, which can be recorded by electroencephalograms, or EEGs. The scientists theorize that a seizure’s function is to correct a neural system gone awry. Though it may sound counterintuitive, a buildup of organized, harmonious signals is what apparently needs to be fixed to return the brain to its naturally chaotic state.

To predict seizures, Iasemidis, who directs the Gainesville VA’s Brain Dynamics Laboratory, and Sackellares look for signs of communication between the site where a patient’s seizure begins and elsewhere in the brain. When an increasing number of electrode pairs begin oscillating together during an EEG, it signals a seizure is on its way.

“We may not be able to pinpoint the exact time, but we can determine whether you are in danger,” Sackellares said.

Iasemidis noted that their goal is to be able to identify a window of opportunity for preventing seizures. “Predicting exactly when a seizure will occur is not the main question,” he said. “We’re interested to see if we can knock the system out of its route to the seizure. We’d like to see if we can intervene with either electricity or medication to try to get the system to reset itself right at the beginning of the buildup of the pre-seizure transition.

“The sooner we can identify a warning flag and intervene, the more likely we will be successful by using a very low-intensity stimulus,” he said. “Our technique also may be useful in evaluating the effect of anti-epileptic drugs on the recovery of the system.”

That information could prove helpful during a crisis, such as during a rapid succession of seizures known as status epilepticus, a condition that can cause brain damage, Iasemidis said.

Epilepsy is the name given to a variety of seizure disorders, which are estimated to afflict more than 2.3 million Americans, according to the Epilepsy Foundation of America. In 70 percent of the cases, there is no known cause; head trauma, tumors, strokes, infections and poisons are implicated in the others.

The research data were gathered from patients in the Epilepsy Monitoring Unit at Shands at UF medical center. All research participants, who were undergoing a presurgical evaluation, had temporal lobe epilepsy that could not be controlled by medication. The data were analyzed after the patients were discharged, so the prediction techniques have not yet been used to influence treatment.

But Sackellares and Iasemidis say it is realistic to think that implantable devices can be developed to detect the preseizure state and automatically act to thwart it. They noted that such devices have been developed for other conditions, including diabetes.

A scientist who collaborated with Sackellares and Iasemidis in the early 1990s said he had been skeptical back then that the computational difficulties of the line of research could be overcome.

“It seems that they have been able to press ahead and realize their long time-dream of making the methods associated with chaos theory practical,” said William J. Williams, a professor of electrical and biomedical engineering and computer science at the University of Michigan. “No one else has pursued this direction of research so persistently in order to achieve such an advanced understanding of epilepsy. The practical applications of their work will likely have a lasting beneficial effect on many people who suffer from epilepsy. “