Epileptic seizures may be predictable.

People with epilepsy typically lead peaceful lives in the days and hours between their seizures. However, the calm they experience before the mental storms may not be so tranquil after all.

Patterns of mild electrical disturbance in the brains of epilepsy patients seem to foreshadow a seizure hours before its onset, reports a group of neurobiologists and electrical engineers. If the findings are confirmed, they may lead to devices that would theoretically predict seizures and even halt them before they begin, suggest the researchers in the April NEURON.

Fifty million people around the world--roughly 1 in 100--live with epilepsy. It's characterized by erratic discharges of electrical activity in the brain. These seizures can set off violent spasms and cause a person to lose consciousness (SN: 6/3/00, p. 384).

"One of the worst aspects of epilepsy is that seizures come without warning," says Brian Litt of the Hospital of the University of Pennsylvania in Philadelphia, who co-authored the study with colleagues from the Georgia Institute of Technology and Emory University, both in Atlanta. Although not often directly harmful, a seizure can endanger a person by unexpectedly occurring during an activity, such as driving, that requires control.

Currently, epilepsy patients must take seizure-quelling medications every day. Some complain that these drugs dull the senses and have other side effects.

One potential advantage of prediction is that patients would need medication only in the few hours surrounding an epileptic episode, says Robert Lowenstein, an epilepsy researcher at Harvard University Medical School in Boston.

Accurate prediction of seizures, he adds, also offers the possibility that an implanted electronic device could stimulate the brain and defuse seizures before they occur. Lowenstein and others are developing such devices.

Past research focused on detecting a seizure in the few minutes before it occurs (SN: 5/23/98, p. 326). Litt decided to take a longer view of the process by using a standard measurement of brain activity, the electroencephalogram (EEG).

Attracted by the quantitative rigor of industrial engineering, Litt and his colleagues teamed up with electrical engineer George Vachtsevanos and his group, all at Georgia Tech. They usually design failure-detection systems in equipment such as aircraft.

Rosana Esteller of Vachtsevanos' group had designed a technique for handling large data sets. The researchers examined 3 days' data from EEG analyses of five epilepsy patients being evaluated for surgery to remove the region of their brain responsible for their seizures.

The researchers found that a series of electrical events anticipated the seizures. Mild energy bursts occurred 7 hours before the seizure, followed 5 hours later by frequent, symptomless seizures too small for the patients to recognize. "It's like a match that keeps lighting a fuse until finally it catches," says Litt.

"This is some of the strongest evidence to date that [a seizure] can be predicted several hours beforehand," contends Lowenstein.

Litt expects that within only a few years, epilepsy patients will carry brain-implant devices that will both predict and preempt seizures.