Affiliations: [a] Department of Electrical Engineering, Islamic Azad University, South Tehran Branch, Iran | [b] Department of Biomedical Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran | [c] Department of Biomedical Engineering, Shahed University, Tehran, Iran | [d] Center for Informatics and Systems (CISUC), Department of Informatics Engineering, University of Coimbra, Portugal
Correspondence:
[*]
Corresponding author: Soroor Behbahani, Department of Electrical Engineering, Islamic Azad University, South Tehran Branch, Iran. E-mail:sor.behbahani@gmail.com
Abstract: BACKGROUND: Until now, different approaches have been
published to resolve the problem of predicting epileptic seizures. The
results are reminiscent of a substantial need for improvements in these
methods to reach the stage of the clinical application. Our aim is to
develop a reliable epileptic seizure prediction algorithm based on the Heart
Rate Variability (HRV) analysis. METHODS: We analyzed the HRV of sixteen epileptic patients with a
total of 170 seizures, to predict the occurrence of seizures based on the
dynamic changes of Electrocardiogram (ECG) during the pre-ictal period. Time
and frequency-domain features were computed forthe consecutive time windows
with a length of five minutes. An adaptive decision threshold method was
used for raising alarms. Predictions were made when selected features
exceeded the decision thresholds. RESULTS: For the seizure occurrence period (SOP) of 4:30 minutes,
and intervention time (IT) of 110 Sec, the presented method showed an
average sensitivity of 78.59%, and average false prediction rate of
0.21/Hr, which indicates that the system has superiority to the random
predictor. CONCLUSION: The proposed approach shows a potential in the
monitoring of epileptic patients and improving their life quality. The
overall performance of the algorithm is a step forward for clinical
implementation.
