Design and Development of Machine Learning based Expert System for Epileptic Seizure Diagnosis and Classification Process

Numerous cases of epilepsy indicates that the enthusiasm can be triggered by a range of factors, but it inevitably contributes to a group occurrence - seizures. The diverse competencies in the assessment process of seizure detection algorithms are also overwhelming considerations for cognitive patterns. The quenching of suitable mixing methods highlights the need for a suitable seizure detection system amid multiple advances in the automated detection of epilepsy over the past decade. The method of automatic seizure detection offers a reliable algorithm, using a profound knowledge of the complexities of signals and clinical fields. The overall risk of damage to the patient is limited if challenges and periodic alerts are resolved to start care for seizure control. The following are found: I preprocessing signal (ii) extraction and iii) classification of functions. The presence of artefacts usually blurs the EEG signals, which add signal spikes, which then disrupts the accurate diagnosis and analytics. Due to 50 Hz line noise, that is inherently adaptive and removed through the application of a traditional notch filter, the EEG signal is degraded. The next step is the spatially limited, independent analysis of components and wavelets to enhance EEG signal efficiency. The study was conducted of five types of objects, including motions of the eye, blinking, chewing, mandibular occlusion and the introduction of electrical inconsistencies (due to electrode bursts). It is a major challenge to obtain pertinent knowledge from EEGs by correlating them with pathological events. To do this, we used a back propagation-based neural network classification. The second target after extraction of features is to improve classification precision. Extracting and classifying 100 subjects in each category and breaking the information into preparation, testing and checking the algorithm proposed.