The new AI can analyze medical data and predict epileptic seizures. While it works only for individual patients, but in the future, it can be used in large quantities.
Researchers at the University of Washington combined Artificial Intelligence (AI) with a theory of systems theory to develop a more effective way to detect and accurately determine an epileptic seizure in real-time.
They explained that most of these seizures occur when brain activity is interrupted by abnormal activity of nerve cells in the brain, which is caused by excess charges in the nerve cells of the brain. These charges cause a state of uncoordinated arousal in the nervous system, therefore, for a short time period, the brain cannot function normally and a so-called epileptic seizure occurs.
At the same time, the accuracy of seizure detection is low, especially when temporal EEG signals are used. Therefore, the team developed a network output technique to facilitate the detection of a seizure.
“Our method allows us to obtain raw data, then process it manually or use the machine learning model to get the information we need. The main advantage of our approach is that the signal from 23 electrodes is analyzed at the same time, and unnecessary information is immediately cut off, so we do not need a large amount of computing resources. Especially in order to see the general picture of the patient’s condition, we introduced a parameter that evaluates all sources of information about him”.
In the event that AI detects a suspicious or dangerous brain activity, he will inform the doctor about it. While this method needs to be developed for each patient individually. The next step is the integration of machine learning to summarize the various types of seizures in patients.