研究文章
设计和评估一个健壮且可概括的ANN-Based分类器的预测早产的多通道Electrohysterographic记录
表3
平均值和标准偏差(括号)之间不同的分类指标获得的培训和验证数据集10平衡使用击打技术获得的数据。对于每一个平衡的数据集,进行了30倍交叉验证方法。
|
|
精度(%) |
AUC (%) |
感性(%) |
特异性(%) |
PPV (%) |
NPV (%) |
| 火车 |
验证 |
火车 |
验证 |
火车 |
验证 |
火车 |
验证 |
火车 |
验证 |
火车 |
验证 |
|
| C1 |
91.4 (1.1) |
90.2 (1.8) |
91.3 (1.3) |
90.4 (2.1) |
91.4 (1.1) |
89.7 (1.6) |
91.4 (1.1) |
90.7 (1.9) |
91.4 (1.1) |
90.7 (1.9) |
91.4 (1.1) |
89.7 (1.6) |
| C2 |
95.5 (0.9) |
94.2 (1.4) |
95.4 (1.1) |
94.5 (1.9) |
95.6 (0.7) |
93.7 (1.2) |
95.5 (1.1) |
94.7 (1.8) |
95.5 (1.0) |
94.7 (1.8) |
95.6 (0.8) |
93.8 (1.2) |
| C3 |
98.7 (1.0) |
98.1 (1.6) |
99.8 (1.1) |
99.6 (1.9) |
99.5 (0.8) |
98.8 (1.3) |
99.4 (1.3) |
99.5 (0.021) |
99.6 (1.2) |
99.5 (2.0) |
99.2 (0.8) |
98.8 (1.3) |
| C4 |
98.6 (0.6) |
98.3 (1.2) |
99.6 (0.7) |
98.9 (1.5) |
99.8 (0.5) |
97.9 (1.0) |
99.5 (0.8) |
98.8 (1.6) |
99.5 (0.8) |
98.8 (1.6) |
99.8 (0.5) |
97.9 (1.0) |
| C5 |
98.8 (1.3) |
98.4 (2.0) |
99.5 (1.5) |
98.8 (2.2) |
99.7 (1.3) |
98.8 (1.9) |
99.4 (1.3) |
98.9 (2.2) |
99.6 (1.3) |
99.1 (2.1) |
99.8 (1.3) |
98.9 (1.9) |
| C6 |
98.8 (1.0) |
98.7 (1.6) |
99.6 (1.3) |
99.2 (2.1) |
98.9 (0.8) |
98.5 (1.3) |
99.6 (1.2) |
99.1 (2.0) |
99.5 (1.2) |
98.9 (2.0) |
99.6 (0.8) |
98.8 (1.3) |
| C7 |
98.7 (1.0) |
98.5 (1.7) |
99.8 (1.2) |
99.5 (1.9) |
99.3 (0.8) |
99.0 (1.3) |
99.6 (1.3) |
99.4 (2.1) |
99.7 (1.3) |
99.1 (2.1) |
99.7 (0.8) |
99.2 (1.3) |
| C8 |
99.1 (0.7) |
98.8 (1.3) |
99.7 (0.7) |
99.0 (1.5) |
99.7 (0.5) |
99.1 (1.5) |
99.7 (0.8) |
99.2 (1.7) |
99.5 (0.8) |
98.8 (1.7) |
99.8 (0.5) |
99.5 (1.0) |
| 制备过程 |
98.8 (1.3) |
98.6 (2.0) |
99.7 (1.5) |
99.3 (2.3) |
99.5 (1.3) |
99.0 (1.8) |
99.5 (1.3) |
99.5 (2.1) |
99.8 (1.3) |
96.5 (2.1) |
99.8 (1.3) |
99.3 (1.8) |
| 10大 |
98.7 (0.9) |
98.2 (1.5) |
99.8 (1.2) |
99.4 (2.0) |
99.7 (0.8) |
99.8 (1.2) |
99.7 (1.1) |
99.5 (1.9) |
99.7 (1.1) |
99.4 (1.9) |
99.8 (0.8) |
99.8 (1.3) |
|
|
