A crowd-sourced online contest for data scientists has come up with a seizure-predicting model that accurately predicts seizures 82 percent of the time. Seizures impact about one percent of the population, and are described as electrical storms that take place in the brain. Earlier models did little more than guess when a seizure was going to occur. The new approach uses an algorithm to predict when a seizure will take place and could lead the way for an implanted device that would warn epileptics of when a seizure is about to occur. Such a device could take away the need to swallow drugs—more often with unwanted side-effects—for years to prevent seizures.