Advanced CNN-LSTM Edge-Computing in Epilepsy Wearables for Real-Time Seizure Detection and Extended Battery Life

Aim: The Project aims to foster a hybrid CNN-LSTM model improved for edge computing to upgrade continuous monitoring, accuracy, latency and battery consumption in wearable epilepsy identification device. Materials and Methods: The device uses edge-registering abilities for real-time handling of EEG information, coordinated with 3-axis accelerometers, Dallas temperature sensors, and EEG sensors. A comparative analysis was led between a CNN-just model and a CNN-LSTM crossover model, each tried with 15 examples. Statistical investigation was performed utilizing SPSS with an significance level of 0.05 and a 95% confidence interval. Results: The CNN-LSTM edge-computing device significantly outperformed the CNN-just model, accomplishing an accuracy range of 81.2 % – 99.9 %, diminished latency to 45.6 ms, and lower battery utilization of 2 %, contrasted with the CNN model's 75.3 % – 90.5 % precision, 59.7 ms latency, and 3.2 % battery use. The edge-computing combination empowered quicker reaction times and limited information transmission overhead, improving real-time execution. Conclusion: The proposed CNN-LSTM edge-figuring framework shows huge upgrades progressively seizure discovery accuracy, latency decrease, and energy productivity, offering a versatile and hearty answer for ceaseless epilepsy checking in wearable clinical gadgets.