# SBTM: epileptic seizure prediction from EEG signal using deep learning in blockchain-enabled smart healthcare monitoring with IoT networking

## TL;DR

This paper introduces a new deep learning model for predicting epileptic seizures using EEG signals, integrated with blockchain and IoT for secure remote healthcare monitoring.

## Contribution

The novel contribution is the Spizella Optimization-based Bi-LSTM model (SBTM) for seizure prediction with enhanced accuracy and integration into a blockchain-enabled IoT framework.

## Key findings

- The SBTM model achieves 97.52% accuracy, 97.51% sensitivity, and 98.51% specificity in predicting epileptic seizures.
- The proposed framework improves data security and remote monitoring in tech-aided healthcare systems.

## Abstract

Epileptic Seizure prediction is highly significant for the identification and reduction of high risks related to serious brain injuries, strokes, and brain tumors. Early and accurate diagnosis is vital for providing intervention measures for enhancing the quality of life of the affected individuals. Numerous techniques have been developed based on Machine vision techniques to predict epileptic seizures. Nonetheless, the acquisition of precise epileptic seizure detection with low false positive rates is challenging. Moreover, the emergence of the Internet of Things (IoT) revolutionized healthcare monitoring with technological improvements, aiming to handle the concerns related to data interoperability, scalability, as well as privacy issues. Hence, this research proposes the Smart Healthcare Monitoring Framework, namely Spizella Optimization-based Bidirectional Short Term Memory Network (SBTM), for determining the seizure states, thereby allowing the provision of remote care. Specifically, the proposed model exploits the Bi-LSTM architecture that captures the temporal dependencies and nonlinear dynamics of EEG signals, making the model highly efficient for predicting the seizure patterns. Besides, the Spizella Optimization is applied for fine-tuning the hyperparameters of the classifier, thereby leading to accurate prediction. Experimental results demonstrate that the proposed SBTM model accomplishes superior results by achieving high accuracy, sensitivity, and specificity equivalent to 97.52%, 97.51% and 98.51% with 90% training, outperforming the state-of-the-art techniques. Moreover, the presented approach significantly improves the remote monitoring, guaranteeing on-time medical care, ensuring data security, and enhancing the overall performance of applications in tech-aided healthcare systems.