Interpretable CNN for ischemic stroke subtype classification with active model adaptation

Table 7 Comparison of different models for baseline

Method	Accuracy	AUC	Recall	Precision	F1-score
NB [22]	0.5023	0.6054	0.5023	0.4493	0.4231
Multinomial NB [23]	0.1728	0.5402	0.1728	0.4471	0.2070
DT [22]	0.5421	0.6138	0.5421	0.4538	0.4594
RF [21, 22, 24]	0.5671	0.6532	0.5671	0.4865	0.4755
ET [21, 23]	0.5786	0.6504	0.5786	0.5022	0.5016
CART [24]	0.4431	0.5476	0.4431	0.4527	0.4557
GDBT [21]	0.5639	0.5956	0.5639	0.4321	0.4544
XGBoost [21]	0.5605	0.6453	0.5605	0.4734	0.4702
AdaBoost [23]	0.5409	0.5812	0.5409	0.4639	0.4716
LDA	0.5647	0.6302	0.5647	0.4577	0.4653
QDA	0.2616	0.5667	0.2616	0.4144	0.2039
LR [22, 24]	0.5565	0.6309	0.5565	0.4452	0.4290
KNN [21, 22, 24]	0.5366	0.6031	0.5366	0.4513	0.4564
SVM [21, 22, 24]	0.5646	0.6228	0.5646	0.4461	0.4570
NN [22, 26]	0.5539	0.5192	0.5539	0.3649	0.4083
MLP [23]	0.5353	0.5015	0.5353	0.3140	0.3956
LSTM	0.1295	0.5544	0.1295	0.4978	0.1252
LSTM+Att	0.0879	0.5781	0.0879	0.2701	0.0634
Bi-LSTM [25]	0.1923	0.6032	0.1923	0.7009	0.1924
Bi-LSTM+Att	0.1515	0.6020	0.1515	0.6986	0.1446
Ours	0.6020	0.6757	0.6020	0.6213	0.5141

ISSN: 1472-6947