Compliance Framework for Seizure Detection via Gaussian Deep Boltzmann Machine Using EEG Data Signal

Epilepsy is a neurological condition in which there is abnormal electrical discharge of brain cells all at once, causing neurological disturbance. Epilepsy can be caused by many factors such as traumatic injury, vascular malformation, brain abscesses, tumors, calcium metabolism disorders, and pyridoxine deficiency. The electroencephalogram (EEG) is one of the tests used to diagnose this disease and EEG-based classification is used to analyze the real-time signal to predefine the data into a set of classes. Different variety of algorithms were previously used to develop and explorer the EEG signal features. The deep learning algorithm was rarely used investigate to explore and extract the features from EEG signals. In this paper, several algorithms were proposed to decompose the EEG signals and extract features, and Hilbert-Huang Transform (HHT) was used to extract the data is an adaptive technique that have flexibly in stoppage criteria. After feature extraction, the Gaussian Deep Boltzmann machine (GDBM) was used to determine the classification accuracy, memory requirement, and computational complexity, in order to clarify which algorithm provided better results. The proposed methods were compared with three other logistic regression algorithms, Naïve Bayes, and support vector machine (SVM). The results showed that the DBM outperformed the over other selected state-of-the-art methods is statistically significant in individuals with epilepsy versus healthy people.