研究文章
自动化的心房纤维性颤动检测基于特征融合使用判别典型相关分析
表3
比较以前的研究基于生理网/ CinC挑战2017的心电图公共数据集。
|
| 方法 |
|
|
|
|
Acc |
Spe |
森 |
|
| 卷积递归神经网络(23] |
92.4% |
81.4% |
80.9% |
84.9% |
87.5% |
94.6% |
82.9% |
| 决策树合奏(24] |
88.9% |
79.1% |
70.2% |
79.4% |
- - - |
- - - |
- - - |
| 16层1 d残留卷积网络(25] |
90.0% |
82.0% |
75.0% |
82.0% |
80.2% |
- - - |
- - - |
| 二维卷积网络LSTM层(26] |
88.8% |
76.4% |
72.6% |
79.2% |
82.3% |
- - - |
- - - |
| 1 dcnn包含剩余块和复发性层(27] |
91.9% |
85.8% |
81.6% |
86.4% |
- - - |
- - - |
- - - |
| 摘要提出了 |
93.1% |
88.3% |
84.0% |
88.3% |
91.7% |
93.2% |
90.4% % |
|
|
