2.
L. Vijaya Kumar
– Malla Reddy Engineering College for Women, Hyderabad, Telangana, India.
3.
Srivani Putta
– Malla Reddy Engineering College for Women, Hyderabad, Telangana, India.
4.
Siddamsetty Saritha
– Sreyas Institute of Engineering and Technology, Hyderabad, Telangana, India.
5.
Devi Uma M.
– Malla Reddy Engineering College for Women, Hyderabad, Telangana, India.
Received -
Accepted -
Published 20-Jun-2026
Abstract
Patients’ quality of life is significantly affected by epilepsy, a long-term brain condition characterised by frequent and unpredictable seizures. Early detection of these seizures can lower the risks associated with them and allow for prompt preventive actions. Although EEG readings provide insights into brain activity, predicting seizures is challenging because the patterns can be chaotic and unpredictable. Recently, machine learning has helped identify subtle patterns in these recordings by linking complex information over time. This method uses deep learning to alert users about upcoming seizures by analysing EEGs. It moves away from traditional methods and utilises CNNs to capture location-based features from the data while uncovering hidden patterns. To manage the timing relationships within the EEG data, it employs LSTM, a type of recurrent network that excels at following sequences. To enhance data quality and the model’s performance, the EEG signals go through preprocessing steps like segmentation, normalisation, and noise reduction. The trained model can predict the preictal state before a seizure occurs, enabling early warning generation. Experimental evaluations on a publicly available EEG dataset show that the proposed method performs better than traditional machine learning techniques when it comes to prediction accuracy, sensitivity, and false alarm rates.
Keywords Convolutional Neural Network (CNN), Deep learning, Electroencephalogram (EEG)
