The detection of an epileptiform activity on EEG signals by using data mining process

A lot of theoretical and practical studies on biomedical and signal processing have been carried out in the world and Turkey. The studies generally include electronic system designs, mathematical models, statistical analysis, fuzzy logic, neural network and genetic algorithm based models, mobile medical applications and related software designs and intelligent equipment automations. Increases in the number of such studies after 2000?s are very significant. A very limited and insufficient number of studies on data mining in the world and Turkey clearly show the need for such studies. Using data mining process in the area of medical data shortens the time span of analysis and increases the accuracy of the result. Such findings indicate that this study has high implementation value and an up-to-date subject.In this study; the goal is to develop a tool and additional software that would help diagnose whether or not there are epileptic activities based on electroencephalograph (EEG) data. Firstly, eight attribute parameters are selected in order to be able to gather data by using grouping and classification algorithm in the process of data mining. Then, the values of selected attribute parameters are determined. Thirdly, the values of selected attribute parameters are analyzed by eleven different data mining algorithms and it is observed that some selected algorithms help detect epileptic activity with high accuracy levels. Accordingly, it is found that different data mining algorithms allow for a comparison of calculation time and accuracy levels of the results. Through the algorithms that give the highest accuracy level, the findings of the study provide health care professionals with the data to diagnose epileptic activity, to ease the decision making process in the process of diagnosis, and to contribute to neurological studies and computer science literature in Turkey by increasing the level of accuracy in diagnosis.