【摘要】 为了更准确地进行癫痫预测，提出基于特征融合的脑电通道选择算法。将样本熵、排列熵和方差特征根据脑电电极位置进行特征融合，融合特征作为脑电通道选择算法的输入进行癫痫预测。结果表明，融合特征在使用9个脑电通道的情况下得到96.74%的准确率和0.93的适应度函数值，远高于其他特征组合。更多还原

【Abstract】 In order to predict epilepsy more accurately, an algorithm of EEG channel selection based on feature fusion is proposed. The sample entropy, permutation entropy, and variance features are fused based on the position of electroencephalogram electrodes, and the fused features are used as inputs for the electroencephalogram channel selection algorithm for epilepsy prediction. The results show that the fused features achieved an accuracy of 96.74% and a fitness function value of 0.93 when using 9 electroencephalogram channels, which is much higher than other feature combinations.更多还原