基于多域特征和GATv2网络的癫痫发作预测方法

韩哲,孟庆芳,张强,张相龙,赵亚欧

Seizure prediction method based on multi-domain feature and GATv2 network

Zhe HAN,Qingfang MENG,Qiang ZHANG,Xianglong ZHANG,Yaou ZHAO

表 2 所提癫痫发作预测方法在CHB-MIT数据集上的评估结果

Tab.2 Evaluation results of proposed seizure prediction method on CHB-MIT dataset %

病例 Acc Sen Spe F1 AUC chb1 98.00 99.56 96.75 98.92 98.48 chb2 97.74 98.05 98.18 97.80 99.14 chb3 98.03 97.25 98.96 97.23 98.39 chb4 97.45 96.94 96.92 96.97 98.70 chb5 92.95 96.39 95.74 93.34 96.82 chb6 95.50 96.28 95.72 95.46 96.80 chb7 98.67 98.86 97.24 99.32 98.75 chb9 98.59 98.55 97.77 96.58 98.21 chb10 99.63 99.62 99.93 99.43 99.86 chb11 97.00 98.34 97.01 97.52 98.81 chb13 98.59 97.28 98.83 97.83 98.84 chb14 95.10 94.39 96.50 95.21 97.98 chb17 98.00 99.03 97.56 98.86 99.85 chb18 98.52 98.48 95.43 97.42 99.21 chb19 97.50 98.07 96.98 97.38 98.86 chb20 98.51 99.51 96.97 99.26 98.95 chb21 98.06 98.67 97.45 98.42 99.88 chb22 99.53 99.62 99.38 99.54 96.85 chb23 98.57 99.11 97.26 98.07 99.58 平均值 97.58 98.11 97.88 97.61 98.63