Identification of presence of epilepsy using predictive analytics.

Epilepsy is one of the most prominent neurological diseases, affecting millions of people worldwide. It is a prevalent neurological illness, affecting about 70 million people worldwide. The usual method of seizure detection utilized by neurologists is time-intensive. An abrupt increase in brain activity known as a seizure can happen to anyone at any time. Indians are thought to make for nearly one-sixth of the world's epilepsy population, with estimates ranging as high as 12 million. Due to the health hazards it presents, the condition has always had a significant amount of significance in the biomedical community. It is characterized by recurring, unprovoked seizures, and the electroencephalogram can be used to diagnose it. The epilepsy study evaluates the EEG data to detect epileptic seizures in their early phases. EEG measures the electrical activity in the brain. Building patient-independent models is more challenging than developing patient- specific classifiers, which have received a lot of research. The Bonn University database is utilized for this work's cross- patient perspective since it is more challenging due to EEG variability between various participants. A comparison of the pattern recognition algorithms used for EEG-based epileptic seizure diagnosis was carried out. In this work, we have two methods: logistic regression and random forest. Logistic regression has a little greater accuracy under some circumstances. In pattern recognition problems, proper feature selection is critical. For the best features, we used recursive feature elimination (wrapper method). We used EEG signals in our study to assess the performance of logistic regression, random forest algorithms. In this work, we have compared two methods for discernment of accuracy. The accuracy of Logistic regression and random forest is 99% and 98.3%. [ABSTRACT FROM AUTHOR]