A New Neuromorphic Computing Approach for Epileptic Seizure Prediction

Several high specificity and sensitivity seizure prediction methods with convolutional neural networks (CNNs) are reported. However, CNNs are computationally expensive and power hungry. These inconveniences make CNN-based methods hard to be implemented on wearable devices. Motivated by the energy-efficient spiking neural networks (SNNs), a neuromorphic computing approach for seizure prediction is proposed in this work. This approach uses a designed gaussian random discrete encoder to generate spike sequences from the EEG samples and make predictions in a spiking convolutional neural network (Spiking-CNN) which combines the advantages of CNNs and SNNs. The experimental results show that the sensitivity, specificity and AUC can remain 95.1%, 99.2% and 0.912 respectively while the computation complexity is reduced by 98.58% compared to CNN, indicating that the proposed Spiking-CNN is hardware friendly and of high precision.

💡 Research Summary

This paper introduces a new neuromorphic computing approach for predicting epileptic seizures, addressing the limitations of traditional convolutional neural networks (CNNs). While CNN-based methods offer high specificity and sensitivity in seizure prediction, they are computationally expensive and power-intensive, making them challenging to implement on wearable devices. The proposed method leverages energy-efficient spiking neural networks (SNNs) by using a Gaussian random discrete encoder to convert EEG samples into spike sequences for processing in a Spiking-CNN.

The Spiking-CNN combines the advantages of both CNNs and SNNs, achieving high prediction accuracy while significantly reducing computational complexity. Experimental results demonstrate that the proposed approach maintains sensitivity at 95.1%, specificity at 99.2%, and an AUC of 0.912, with a reduction in computation complexity by 98.58% compared to CNNs.

This research highlights the potential for Spiking-CNNs to be implemented effectively on wearable devices due to their low power consumption and high precision. The study underscores the importance of developing energy-efficient models that can operate in real-time environments, such as those required for continuous health monitoring applications. By integrating the strengths of both CNNs and SNNs, this approach not only enhances prediction accuracy but also ensures practical applicability across various platforms, including wearable technology.