This study presents a new attempt to quantify and predict changes in the ECG signal in the pre-ictal period. In the proposed approach, threshold techniques were applied to the standard deviation of two heart rate variability features (The number of heartbeats per two minutes and approximate entropy) computed to ensure prediction and quantification of the pre-ictal state. We analyzed clinical data taken from two epileptic public databases, Siena scalp EEG and post-ictal heart rate oscillations in partial epilepsy and a local database. By testing the proposed approach on the Siena scalp EEG database, we achieved a sensitivity of 100%, specificity of 95%, and an accuracy of 96.4% whereas using acquisitions from the post-ictal database, we achieved a sensitivity of 100%, specificity of 91% and an accuracy of 94% and using the local database we achieved a sensitivity of 100%, a specificity of 97% and an accuracy of 97.5%. Furthermore, the proposed approach predicted 58.7%, 57.2, and 40% of the seizures before the onset by more than 10 min for the data taken from post-ictal, local and Siena database, respectively. Using the automatic threshold technique, we were able to achieve a sensitivity, specificity, and accuracy of 85%, 81%, 82% using our local database, respectively, whereas using acquisitions take from the Siena scalp EEG database, we achieved a sensitivity of 75%, specificity of 85% and an accuracy of 82%. Besides, using the post-ictal database, we achieved a sensitivity of 90%, a specificity of 83% and an accuracy of 85%.

这项研究提出了量化和预测发作前心电图信号变化的新尝试。在所提出的方法中，将阈值技术应用于计算的两个心率变异性特征（每两分钟的心跳次数和近似熵）的标准差，以确保对发作前状态的预测和量化。我们分析了来自两个癫痫公共数据库、锡耶纳头皮脑电图和部分性癫痫发作后心率振荡以及本地数据库的临床数据。通过在锡耶纳头皮脑电图数据库上测试所提出的方法，我们实现了 100% 的敏感性、95% 的特异性和 96.4% 的准确性，而使用从发作后数据库采集的数据，我们实现了 100% 的敏感性、95% 的特异性和 96.4% 的准确性。 91% 的准确度为 94%，使用本地数据库我们实现了 100% 的敏感性、97% 的特异性和 97.5% 的准确度。此外，对于从发作后、本地和锡耶纳数据库获取的数据，所提出的方法分别预测了发作前 10 分钟以上的 58.7%、57.2 和 40% 的癫痫发作。使用自动阈值技术，我们能够使用本地数据库分别实现 85%、81%、82% 的灵敏度、特异性和准确性，而使用锡耶纳头皮脑电图数据库采集的数据，我们实现的灵敏度为75%，特异性为 85%，准确度为 82%。此外，使用发作后数据库，我们实现了 90% 的敏感性、83% 的特异性和 85% 的准确性。