Epilepsy forecasts could help people manage their daily life and medication around their chance of seizures

Geneva, Switzerland – A new study published today in The Lancet Neurology shows that it is possible to forecast seizures in people with epilepsy from one to several days in advance using long-term brain data. 

People with epilepsy are plagued by constant uncertainty about the looming threat of seizures. Now, a Swiss-US team has detected individual cycles of epileptic activity that can show the probability of a future seizure. The approach is remarkably reliable, and clinical trials are now being planned.

The team built a computer model and trained it on data from people with drug-resistant epilepsy who were implanted with a long-term brain monitoring device. Using similar approaches to those in weather forecasting, the model can correctly forecast phases of higher seizure risk.

The study demonstrates that long-term monitoring data is key to reliable forecasting and highlights the need for minimally invasive monitoring devices and intelligent software.

The researchers are also working in collaboration with the Wyss Center for Bio and Neuroengineering, based at Campus Biotech in Geneva, to deve­lop a minimally invasive brain monitoring device, Epios, to record the long-term data needed to forecast seizures. The device, which slips under the skin of the scalp, could give people with epilepsy the power to plan their lives according to the likelihood of having a seizure.

A third of epilepsy sufferers are resistant to treatment for this neurological disease that affects 1% of the population. The onset of seizures is unpredictable and has been the subject of research since the 1970s. The unpredictable nature of the disease means patients must often take medication and adjust their lifestyles. Neuroscientists from the University of Geneva (UNIGE) and the University Hospital of Bern (Inselspital) – working with the University of California in San Francisco (UCSF) and Brown University in Providence – have developed a technique that can predict seizures between one and several days in advance. By recording neuronal activity over at least six months using a device implanted directly in the brain, it is possible to detect individual cycles of epileptic activity and provide information about the probability of a future seizure.

Epilepsy is a chronic condition in which the brain can switch suddenly to a pathological state where excessive neuronal activity can result in seizures, which may cause convulsions. How and why the brain swaps from one state to another is still poorly un­derstood.

“Specialists worldwide have been trying for over 50 years to predict seizures a few minutes in advance, but with limited success,” explains Timothée Proix, a researcher in the Depart­ment of Fundamental Neurosciences in UNIGE’s Faculty of Medicine.

Seizures do not appear to be preceded by any obvious warning signs that would make prediction easier. The frequency, depending on the individual, varies from once a year to once a day. 

“It’s a huge problem for patients”, begins Maxime Baud, Wyss Center Staff Neurologist, Clinical Neurologist at Inselspital, and senior author on the paper “This unpredictability is associated with a permanent threat that obliges patients to take medication on a daily basis. And in many cases, it prevents them from participating in certain sports. Living with this hanging over you can also affect your mental health”.

Existing treatments are often difficult to bear: they depend on drugs with numerous potential side effects to reduce neuronal excitability and sometimes involve neurosurgery to remove the epileptic focus, i.e. the starting point of the brain seizures. Moreover, a quarter of patients do not respond to these treatments, meaning they have to learn to manage the chronicity of their disease.

Epileptic activity can be measured using cerebral electrical activity data recorded by electroencephalography. This can be used to identify interictal discharges – evanescent discharges that appear in between seizures without directly causing them.

“We observe clinically that epileptic seizures recur in clusters and cyclically. To ascertain whether the interictal discharges can explain these cycles and forecast the onset of a seizure, we analysed the data in greater detail”, continues Dr Baud.