Will I have a seizure today? A new tool may someday help patients predict their seizure risk


For about 30%-40% of the more than 50 million people worldwide with epilepsy, available treatments do not control their seizures. “This staggering number has not changed in decades, despite over 14 new therapies for epilepsy entering the market since the 1990s,” said Sonya B. Dumanis, PhD , director of the Epilepsy Innovation Institute at the Epilepsy Foundation and lead author of a report on a 2017 workshop on the My Seizure Gauge project.

The uncertainty can keep people with epilepsy from partaking in everyday activities, such as driving or engaging in favorite pastimes, because they fear what could happen to them if they were suddenly struck by a seizure, Benjamin H. Brinkmann, PhD , a researcher with the Mayo Clinic in Rochester, Minn., told the Epilepsy Journey in an interview. People might avoid swimming, for example, for fear of drowning if they were to experience loss of consciousness. "Even though you may have one seizure every couple months, you just don't know when that's going to happen,” Dr. Brinkmann said. "It really limits you, more so than you might think."

The My Seizure Gauge project grew out of a need to address this uncertainty, Dr. Brinkmann said. He credits Dr. Dumanis and the Epilepsy Foundation’s close relationship with the community of people who have this condition for homing in on this issue. "Dumanis and the Epilepsy Foundation have a great network of contacts, and they're very good with the media, getting information out to patients, and getting feedback from patients as well,” Dr. Brinkmann said. "That's something that is very valuable. It is sometimes difficult to get without access to that kind of network."

In 2016 the foundation’s Epilepsy Innovation Institute conducted an online survey for feedback about what aspects of this condition most troubled people. More than 1,000 people responded. A majority of
respondents, regardless of seizure frequency and type, selected unpredictability of seizures as a top
issue (Epilepsy Foundation. “Ei2 2016 Community Survey”).

Developing a seizure-gauge device could give people with epilepsy a greater sense of control in managing their disease, Dr. Dumanis said. “You might have seizures less than 1% of your lifetime, but every single day, you don't know if that's the day you're going to have a seizure,” Dr. Dumanis said in an interview.

“Even though you may have one seizure every couple months, you just don’t know when that’s going to happen,” Dr. Brinkmann said. “It really limits you, more so than you might think.”

Seizure forecasting would be different from attempts to categorically predict when people will have seizures, she said. “It's not saying you are 100% likely to have a seizure, but it's ‘today you are at a 70% chance of having a seizure.’ It's like thinking about it in weather terms,” Dr. Dumanis said. “You could imagine a future where someone puts a contingency plan in place for that day just like they would take an umbrella if it might rain.”


The goal of the Epilepsy Foundation's My Seizure Gauge initiative sounds perhaps deceptively easy. The project intends to bring together a global team of experts to harness rapidly growing physiological, behavioral, and environmental data and use them to create a tool to predict seizures. The foundation last year awarded a $3 million grant to a team drawn from King’s College London, the Mayo Clinic, and Seer Medical/The University of Melbourne in Australia. They will collaborate to evaluate biosensors in a range of commercially available devices, with an aim of identifying the types of measurements needed to develop a reliable prototype of a monitor to forecast seizures.

“The timing is perfect, with a convergence of high-speed mobile networks, machine learning, cloud computing, and new wearable devices hitting the market,” Dean Freestone, PhD, chief executive and cofounder of Seer Medical.

“It’s not saying you are 100% likely to have a seizure, but it’s ‘today you are at a 70% chance of having a seizure.’ It’s like thinking about it in weather terms,” Dr. Dumanis said.

Dr. Freestone is one of 3 leaders of the My Seizure Gauge project, along with Dr. Brinkmann and Mark Richardson, BM BCh, MA, PhD, of King’s College in London. The foundation’s approach to this work differs notably in several ways from that taken by major funders of research, according to Dr. Richardson. Many research organizations focus more on what scientists identify as pressing needs and place less emphasis on the patients. They might not back such an ambitious, yet broad, objective, he said. “Conventional funders would often want the goal to be a narrowly defined achievement, with a careful balance between cost and risk,” but the Epilepsy Foundation “seems very keen to keep the goals wide,” Dr. Richardson said about the foundation’s work.

The foundation also is striking a balance between taking a risk on innovative approaches and wanting to ground this project in realistic steps, Dr. Richardson said. “The goal was deliberately a 'moonshot,' but at the same time lower-level goals were seen as highly valuable – for example, just getting a big dataset from patients with wearables is seen as a highly worthwhile goal,” Dr. Richardson said.

The My Seizure Gauge project kicks off with a broad review of commercially available devices that people can wear to track physiological signals like heart rate and blood oxygenation, Dr. Brinkmann said. The researchers then will decide on two to three wearables devices that they will send home with patients included in the study.

“In order to predict seizure, you really need long stretches of data,” Dr. Brinkmann said. "You really need multiple seizures both to train an algorithm and then also to test an algorithm and to have enough statistical power to really tell whether you're doing a good job predicting seizures or not."

Dr. Brinkmann said researchers already anticipate varied quality in the data that they will see over a few years. But there should be a good body of research to advance this project in its third year to the point of holding a machine-learning competition online. As this stage, the My Seizure Gauge project expects researchers to work with the data gathered in the previous year. “People will try to take those biosignals, and we'll provide them some training data and then some unlabeled test data, and they will have to develop algorithms that do a good job on the test data predicting seizures,” Dr. Brinkmann explained.

Hurdles Expected

Enthusiastic as he is about the My Seizure Gauge project, Dr. Brinkmann anticipates that some potential pitfalls may lie ahead. One potential problem may be choosing the wrong technology for gathering data on biological patterns. The field of so-called wearables is advancing quickly, with natural fallout to be expected among competing device companies. Dr. Brinkmann had hoped to include a particular armband monitor in the My Seizure Gauge project, but the maker of the device appears to have gone out of business, he said. “We keep emailing and calling, and nobody ever responds” from the company that made the armband, Dr. Brinkmann said. "It looked really good. I think the patients might have found it very comfortable, but that's kind of off the table."

The researchers in the My Seizure Gauge project, therefore, are focused on the biosignals they intend to track more than the products they will use to gather the data. “Any particular device can disappear from the market or not work well,” Dr. Brinkmann said. “But if we know what biosignals to measure, we can always go out and build a device that works the way we need it to.”

Maintaining Bluetooth connectivity also may be a challenge for developing a seizure gauge, he said. "We'll have to have an algorithm running on a smartphone or a tablet," Dr. Brinkmann said. "If the Bluetooth connectivity drops and we get a signal that a seizure is coming, no algorithm, no matter how good, can help with that."

Still, Dr. Brinkmann remains optimistic and seems clearly enthusiastic about My Seizure Gauge. He cited the strong ties among investigators as one of the strengths of the project. There are highly collaborative relationships among the investigators involved, including his two coleaders of the My Seizure Gauge project, Dr. Richardson and Dr. Freestone, as well as with other colleagues working on the initiative: Mark Cook, MD, MBBS, of the University of Melbourne; Greg Worrell, MD, PhD, of the Mayo Clinic; and Andreas Schulze-Bonhage, MD, of the University of Freiburg in Breisgau, Germany.

“It’s complicated. We can certainly do it,” Dr. Brinkmann said of the effort to identify data patterns needed to advance the project toward development of a seizure gauge. “It’s just a lot for a mobile system to manage."

Setbacks and stumbles are more the norm than the exception in any field of medical research. The My Seizure Gauge project itself is built partly from a close examination of the work of a failed medical device company. In a summary note from the 2017 My Seizure Gauge workshop, Dr. Dumanis noted the early progress made by Neurovista, including its funding of the first trial in people of an implantable seizure warning system. Neurovista developed an ambulatory intracranial electroencephalogram (EEG) system to monitor people with drug-resistant epilepsy. It was tested in Australia, with 15 patients initially implanted with the device. Of this group, 11 patients completed the data collection period, with reported sensitivities of the device ranging from 65% to 100%, Dr. Cook and colleagues reported in the Lancet Neurology in 2013. The following year, an article in the Lancet Neurology quoted Dr. Cook, saying that the study was done in Australia partly because the U.S. Food and Drug Administration (FDA) wouldn’t approve a trial for a device that didn’t have a therapeutic component. Dr. Cook took exception to that “short-sighted,” comment “because clearly the prediction is a therapy of a sort or might lead to a better therapy.”

According to Dr. Dumanis’s 2017 report on the My Seizure Gauge project, the Neurovista trial “showed that seizure forecasting could be done, which led some to argue that seizure forecasting is now an engineering problem.” But Neurovista could not persuade U.S. regulators to let it advance testing. “The investors pulled out, and Neurovista went defunct, but that idea lived on,” Dr. Dumanis said in an interview with Epilepsy Journey.

The My Seizure Gauge project is meant to define the parameters of measures that would be needed for a prototype seizure forecasting device, Dr. Dumanis said. If it succeeds, the Epilepsy Foundation can then try to attract firms to build on this knowledge and seek to develop seizure gauge tools that could someday reach the market. "If we can figure that out, and share that with the community, then companies can come in. That's our role,” to create the research collaboration that can tease out the clues for a successful device, Dr. Dumanis said. “Our role is not to do an end-to-end product but to accelerate the field forward.”

credit: gorodenkoff/Getty Images

Epilepsy Foundation seeks Google Map for brain

Nonprofit group is seeking to spur broad collaboration to better understand why seizures start and stop.


The Epilepsy Foundation would like to someday provide researchers with the same kind of snapshot of the activity happening within people’s brains that Google Maps now gives drivers of traffic on roads.

“You could say ‘Oh, there are red lines here. There's a lot of activity. Can we reroute to prevent the seizure? Can we reroute to stop the seizure?’” said Sonya Dumanis, PhD, director of the Epilepsy Innovation Institute at the Epilepsy Foundation. “That really was the idea of the Brain Map in its very high-level concept.”

At this time, though, the foundation’s My Brain Map project is only in a very early stage, Dr. Dumanis told the Epilepsy Journey in an interview. There’s a June 2019 deadline for submission for an initial round of letters of intent through which researchers will present ideas.

The challenges for the Brain Map project are formidable, a point that the Epilepsy Foundation makes clear. Its own report of an initial 2018 workshop that it held on the Brain Map observes that it “may seem like an insurmountable challenge” given the complexity of an organ with more than 100 hundred billion neurons, “not to mention glial cells, ion channels, dendrites, and other neural components.”

Even with more than a century’s worth of research on the brain, researchers still are unclear about why seizures start, how they spread in the brain, and what causes them to stop, the foundation’s report said.

One of the goals of the Brain Map project is to bring together researchers from a wide variety of disciplines, Dr. Dumanis said. Collaborating at the foundation’s 2018 workshop on the Brain Map were neuroscientists, clinicians, mathematicians, engineers, and representatives of medical firms, along with people who have epilepsy.

“When you think about the modeling, it's way more complicated than someone who has a PhD in neuroscience or an MD necessarily would actually understand,” Dr. Dumanis said, referring to the need for collaboration.

In its early stages, this Epilepsy Foundation project will support exploratory pilot projects that look for ways to depict brain activity across different modalities, time, and spatial scales.

The foundation said it is seeking proof-of-concept studies to provide insights into the mechanisms that spark and stop seizures. The first grants awarded through the Brain Map project will be for as much as $200,000. These grants are intended to be seed funding to generate preliminary data, which researchers could then use to secure further funding from other sources.

The foundation might provide additional funding through its Epilepsy Innovation Institute (Ei2) in cases where a proof-of-concept project proves successful. The foundation said it will have a priority of funding investigators who propose employing new technologies and analytical methods that draw from existing data sets.

Dr. Sonya Dumanis