This paper is concerned with Electroencephalography (EEG) seizure prediction, which means the detection of the pre-ictal state prior to ictal activity occurrence. The basic idea of the proposed approach for EEG seizure prediction is to work on the signals in the Hilbert domain. The operation in the Hilbert domain guarantees working on the low-pass spectra of EEG signal segments to avoid artifacts. Signal attributes in the Hilbert domain including amplitude, derivative, local mean, local variance, and median are analyzed statistically to perform the channel selection and seizure prediction tasks. Pre-defined prediction and false-alarm probabilities are set to select the channels, the attributes, and bins of probability density functions (PDFs) that can be useful for seizure prediction. Due to the multi-channel nature of this process, there is a need for a majority voting strategy to take a decision for each signal segment. Simulation results reveal an average prediction rate of 96.46%, an average false-alarm rate of 0.028077/h and an average prediction time of 60.1595 min for a 90-min prediction horizon.

本文涉及脑电图（EEG）癫痫发作的预测，这是指在发生发作前先检测发作前状态。提出的脑电图癫痫发作预测方法的基本思想是处理希尔伯特域中的信号。希尔伯特域中的操作可确保在EEG信号段的低通频谱上工作，从而避免伪像。对希尔伯特域中的信号属性（包括幅度，导数，局部均值，局部方差和中位数）进行统计分析，以执行信道选择和癫痫发作预测任务。设置预定义的预测和错误警报概率，以选择可能对癫痫发作预测有用的通道，属性和概率密度函数（PDF）箱。由于此过程具有多渠道性质，需要多数表决策略为每个信号段做出决定。仿真结果表明，对于90分钟的预测范围，平均预测率为96.46％，平均错误警报率为0.028077 / h，平均预测时间为60.1595分钟。