Epileptic Seizure Detection and Prediction Using Deep Learning Technique

Epilepsy is a brain condition that affects people of all ages and is a chronic, non-communicable disease. Epilepsy affects around 50 million individuals worldwide, making it one of the most prevalent neurological illnesses. Epilepsy may happen without any specific reason due to genetic deficiency. The unpredictable nature of epilepsy is very dangerous one. Deep learning is a subclass of machine learning that use many layers of processing to extract higher-level features from input. It has been utilized in a number of applications such as computer vision, natural language processing, and so on. Deep Learning is gaining much popularity due to its supremacy in terms of accuracy when trained with huge amount of data. It learns from examples to automatically discriminate different classes. Even though it merits greater attention from the Deep Learning community in epileptic seizure prediction is a field of study. In this study, we provide an LSTM model for detecting and predicting seizure presents states that take into account the chaotic nature of an EEG dataset. Our model is validated by feeding LSTM with an Epilepsy seizure recognition Data Set, which is available in the UCI Database, which is open to the public in the form of comma separated value released by Kaggle. Purpose: Epilepsy affects about 1% of the world's population. The development of fresh methodologies in the field of epilepsy prediction is in great demand. The requirement for a low-cost wearable monitoring gadget to forecast epilepsy state may establish a fear-free atmosphere for those affected. Observations: Most of the papers were written based on seizure detection by using machine learning techniques and mobile alert to caretakers using smart device. There have been some intriguing new advancements in Internet of Things and Deep learning based strategies that have the potential to create a paradigm change in epileptic seizure prediction [26]. Conclusion: The proposed work is achieved 99% of training and testing accuracy using Long Short-Term Memory networks.