Table 1:
Comparative results of proposed C-RNN and other architectures for LV status classification.
| Model | OR | Data | ACC | SEN | SPE | AUC | DR |
|---|---|---|---|---|---|---|---|
| MLP | 0 | Val | 63.99% | 11.00% | 92.56% | 0.5988 | 0.4856 |
| Test | 50.80% | 6.72% | 92.62% | 0.4934 | 0.2610 | ||
| 10% | Val | 63.17% | 20.40% | 86.22% | 0.6258 | 0.6986 | |
| Test | 51.81% | 10.91% | 90.63% | 0.5331 | 0.2958 | ||
| 20% | Val | 63.77% | 22.59% | 85.70% | 0.6272 | 0.7162 | |
| Test | 52.98% | 15.84% | 88.22% | 0.5448 | 0.3829 | ||
| 30% | Val | 63.76% | 24.90% | 84.62% | 0.6306 | 0.7574 | |
| Test | 52.03% | 16.59% | 85.67% | 0.5397 | 0.4229 | ||
| 40% | Val | 63.44% | 27.33% | 82.87% | 0.6361 | 0.7961 | |
| Test | 51.53% | 17.34% | 83.97% | 0.5267 | 0.4667 | ||
| 50% | Val | 62.96% | 30.37% | 80.50% | 0.6338 | 0.8734 | |
| Test | 53.28% | 20.63% | 84.26% | 0.5331 | 0.4747 | ||
| CNN+ MLP | 0 | Val | 64.98% | 8.92% | 94.92% | 0.6069 | 0.4764 |
| Test | 51.02% | 4.33% | 95.32% | 0.5361 | 0.2897 | ||
| 10% | Val | 64.62% | 9.30% | 94.52% | 0.6056 | 0.5341 | |
| Test | 51.38% | 5.14% | 95.32% | 0.5376 | 0.3253 | ||
| 20% | Val | 64.20% | 14.41% | 90.68% | 0.6245 | 0.6236 | |
| Test | 51.75% | 12.11% | 89.36% | 0.5649 | 0.4507 | ||
| 30% | Val | 64.70% | 19.06% | 89.22% | 0.6367 | 0.6504 | |
| Test | 53.57% | 37.82% | 68.51% | 0.5699 | 0.8127 | ||
| 40% | Val | 64.80% | 24.52% | 86.54% | 0.6487 | 0.7335 | |
| Test | 55.17% | 57.10% | 53.33% | 0.5803 | 1.1318 | ||
| 50% | Val | 64.82% | 25.68% | 85.98% | 0.6515 | 0.7723 | |
| Test | 55.39% | 67.41% | 43.97% | 0.5879 | 1.3324 | ||
| RNN+ MLP | 0 | Val | 72.93% | 59.35% | 80.84% | 0.7946 | 1.300 |
| Test | 65.50% | 47.68% | 82.41% | 0.7761 | 0.7572 | ||
| 10% | Val | 73.23% | 59.91% | 80.52% | 0.8024 | 1.2599 | |
| Test | 62.81% | 49.17% | 75.74% | 0.7360 | 1.0474 | ||
| 20% | Val | 72.98% | 57.73% | 81.50% | 0.8022 | 1.2755 | |
| Test | 65.42% | 58.59% | 71.91% | 0.7190 | 1.0073 | ||
| 30% | Val | 73.77% | 59.12% | 82.24% | 0.7962 | 1.2681 | |
| Test | 62.81% | 45.14% | 79.57% | 0.6839 | 0.9331 | ||
| 40% | Val | 72.74% | 56.87% | 81.24% | 0.7872 | 1.2601 | |
| Test | 62.74% | 55.75% | 69.36% | 0.7172 | 1.1059 | ||
| 50% | Val | 72.54% | 56.80% | 81.10% | 0.7869 | 1.2621 | |
| Test | 65.07% | 48.13% | 81.13% | 0.6947 | 1.1040 | ||
| Proposed C-RNN | 0 | Val | 78.39% | 72.02% | 81.82% | 0.8622 | 1.291 |
| Test | 75.32% | 68.01% | 82.27% | 0.8587 | 0.9636 | ||
| 10% | Val | 78.94% | 72.39% | 82.69% | 0.8680 | 1.2693 | |
| Test | 74.89% | 65.62% | 83.69% | 0.8613 | 0.9314 | ||
| 20% | Val | 78.62% | 72.46% | 81.98% | 0.8597 | 1.2672 | |
| Test | 75.96% | 71.00% | 78.72% | 0.8363 | 1.0201 | ||
| 30% | Val | 78.45% | 71.50% | 82.41% | 0.8642 | 1.2551 | |
| Test | 75.40% | 70.55% | 80.00% | 0.8540 | 1.0055 | ||
| 40% | Val | 78.56% | 70.99% | 82.67% | 0.8617 | 1.2470 | |
| Test | 76.35% | 70.70% | 81.70% | 0.8568 | 0.9766 | ||
| 50% | Val | 78.39% | 71.29% | 82.28% | 0.8597 | 1.2472 | |
| Test | 77.07% | 75.93% | 78.16% | 0.8678 | 1.0617 | ||
In the first row, the meanings of all the acronyms and abbreviations are explained as follows: OR: Overlapping Rate; Data: the model was either applied on validation sets or the testing set; ACC: Accuracy; SEN: sensitivity; SPE: specificity; AUC: Area under the ROC curve; DR: Duration ratio.