Clues to Controlling Seizures
The same type of modeling used by meteorologists to forecast the weather could help scientists design better electrical-stimulation therapies for the brain. These therapies, which involve sending small jolts of electricity to specific neural targets, are currently in use for both Parkinson’s and epilepsy, two neurological diseases in which drugs have had limited success.
As neurosurgical technologies improve and medical devices become smaller and more precise, interest in stimulation therapies has blossomed: different therapies are now being tested in a range of disorders, including brain injury, obsessive-compulsive disorder, and depression. Scientists theorize that electrical stimulation blocks abnormal electrical patterns that arise with different diseases, but little is known about how these devices actually work. As use of this technology grows, it is becoming increasingly important for scientists to develop more precise ways to target aberrant brain activity while leaving normal neural communication intact.
“In some sense, we have no idea what electrical stimulation is doing to the brain,” says Robert Duckrow, a neurologist at Yale University, in New Haven, CT, who tests electrical therapies. “It’s almost as if we need to take a step back and say, What is the right way to stimulate the brain to achieve a specific end?”
Steven Schiff, a neurosurgeon and engineer at Pennsylvania State University, is trying to do just that. Schiff and his collaborators are borrowing an engineering technique, known as control theory, to model the networks of neurons that produce the abnormal electrical activity that is characteristic of both seizures and movement disorders such as Parkinson’s. The results should allow scientists to more precisely design stimulation therapies, improving their effectiveness. “We would like to get to the point where we can minimize the energy used and minimize the effect on normal [cognitive] processes,” says Schiff.
See neural activity recorded from a slice of rat cortex and simulated neural activity from three versions of a model calculated from that activity.
With epilepsy, 30 to 40 percent of patients fail to find relief from anticonvulsant medications, and not all patients are eligible for surgery to remove the part of the brain that generates seizures. The vagus nerve stimulator, which stimulates a nerve leading to the brain, was approved for epilepsy treatment more than a decade ago. But it has limited success: only about a third of patients who undergo the procedure report a reduction in seizure rates by 50 percent or more.
Deep brain stimulation, which involves surgically implanting electrodes directly into the brain, has become routine for treating Parkinson’s disease: nearly 40,000 Parkinson’s patients have undergone the procedure to date. While for many patients it’s a welcome alternative to drugs, the treatment needs to be effective for many symptoms, not just the tremors which are the most obvious visible signs of the disease. “It’s the inability to start a movement which is the most disabling to many patients. The harder thing to do is to be more sophisticated in how you maximize patients’ ability to move, and by using models, we hope to create more effective algorithms to interact with the brain’s activity in such patients,” says Schiff
One drawback to the existing stimulation systems is that the pattern and amplitude of electrical signals that doctors use to stimulate the brain have been chosen largely by trial and error. “It’s like the early days of cochlear implants,” says Rahul Sarpeshkar, an electrical engineer at MIT who studies neural prosthetics and computational modeling of disease. “At first, people just tried things. But once they approached the problem more rationally and tried to more closely mimic the biology, people could hear better.”
Building models that mimic brain activity has been a challenge because communication in networks of neurons is so complex. “Scientists are just now beginning to investigate how nerve cells interact to produce the phenomenon of seizure,” says Schiff. His team uses an approach that has shown success in modeling weather and other complex nonlinear systems, and it has just recently been applied to neurons.
The idea is that if scientists can accurately model the activity leading up to and during a seizure, they can use that model to test in real time the type of stimulation parameters that are most effective at preventing abnormal activity before it evolves into a full-blown seizure.
So far, Schiff’s team has been able to build models that replicate oscillating neural activity recorded from the cortex of rodents. And in a paper published earlier this year in the Journal of Neural Engineering, the researchers showed that they could control these virtual wave patterns, outlining a potential approach to controlling electrical activity via neural implants. The scientists are now trying to repeat the feat using dynamics of actual seizures recorded in the lab.
Schiff’s team has also built a neural model of Parkinson’s and soon hopes to test his modeling approach in patients. “We hope to sort out over the next year how to proceed with the [Parkinson’s disease] control in order to make it suitable for safe human study.”
Geoffrey Hinton tells us why he’s now scared of the tech he helped build
“I have suddenly switched my views on whether these things are going to be more intelligent than us.”
ChatGPT is going to change education, not destroy it
The narrative around cheating students doesn’t tell the whole story. Meet the teachers who think generative AI could actually make learning better.
Meet the people who use Notion to plan their whole lives
The workplace tool’s appeal extends far beyond organizing work projects. Many users find it’s just as useful for managing their free time.
Learning to code isn’t enough
Historically, learn-to-code efforts have provided opportunities for the few, but new efforts are aiming to be inclusive.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.