Combining STFT and random forest algorithm for epileptic detection

In the automatic detection of epileptic seizures, time varying electroencephalography (EEG) signals monitoring of critically ill patients is an essential procedure in intensive care units. There is increasing interest in using seizure detection algorithms, such as random forest, for seizures EEG analysis, but a better understanding of how to design and train random forest for EEG decoding and how to visualize the informative EEG time and frequency features the dimensionality reduction of PCA is still needed. Here, we studied seizure detection algorithms designed for recognizing diseased signals from raw seizures EEG. Our results show the recognizing performance of random forest algorithm reaching at mean recognizing accuracies 96%. It can exploit and might help doctors better diagnose the extent of epilepsy.