It is true, IBM researcher Stefan Harrer is developing a new and novel way of monitoring a person's brain in order to predict epileptic seizures. Working with neurologists from the University of Melbourne and the companies Australian branch Harrer is designing a computer system that he says is like an artificial brain, this computer will be dedicated to analyzing a real human brain. They have designed a software application that will interpret human brain waves by using a neural network. A patient's brain waves are fed into the neural network and analyzed by the IBM proprietary hardware. The hardware in question is an experimental IBM chip called TrueNorth, this chip has been built with an architecture that is designed to mirror the human brain which is why it makes an efficient neural network.
The team of researchers in the genes using the computer chip with an external computer and eventually on a wearable device that will work cooperatively alongside a brain implant, the implant would send EEG information to the wearable device or computer which would then use the data to predict the onset of an epileptic seizure. The team expressed the desire for the system to be on a wearable device as quickly as possible as this would then allow them to do an analysis in real-time. They also said that having the wearable device working efficiently is the only real way for this technology to have a great impact beyond looking cool on research papers. They envisage the device on detecting an oncoming seizure would alert the patient and the patient's doctors using a connected smart phone.
Using the data attained from earlier implant studies the scientists involved expect they will be able to see deeper into the underlying structure of seizure brain activity. Harrer and his team hope to use this information to further refine the device so it can then be used not only to predict seizures but possibly be used to even stop them before they occur. Harrer said that to be unable to stop the seizure the artificial brain would have to compensate for the real brain which was malfunctioning.
The difficulty the team has faced from the start is trying to extract the really meaningful information from all the background noise generated by the patient's brain, they have to be able to detect a specific seizure pattern for each individual patient and so far this has eluded them but they all say they are getting closer everyday. This whole thing might sound like science fiction but it isn't, neurologists at the University of Melbourne have already been running a study in which less complex implants gathered EEG readings from epilepsy patients for the last three years, which is why Harrer joined forces with them with his study. The ultimate aim of both of these projects is to one day be able to attach an artificial neural network to a human body, this is many years in the future but all of the team believe it not only can be done but has to be done.