Table 4 Comparison of different algorithms for fetal acidosis diagnosis
From: Hybrid-FHR: a multi-modal AI approach for automated fetal acidosis diagnosis
Reference | Method | Database | Division criterion | Performance% |
|---|---|---|---|---|
[1] | Expert +1D-CNN | CTU-UHB | pH < 7.15 | Sen: 75.23; Spe: 70.82 |
[5] | Expert + PCA + ANN | Private | pH < 7.1 | Sen:60.3; Spe:67.5 |
[6] | Expert + RELIEF + AdaBoost | CTU-UHB | pH < 7.05 | Sen:64.1; Spe:65.2 |
[7] | Expert + Random Forest | CTU-UHB | Expert annotation | Sen:72.41; Spe:78.4 |
[16] | 1D-CNN-GRU | CTU-UHB | pH < 7.15 | Acc: 95.15; Sen: 96.20 Spe: 94.09; Pre: 94.21 |
[26] | Expert + SVM / Naive Bayes | CTU-UHB | pH < 7.15 | Sen: 73.4 / 72.3; Spe: 76.3 / 75.6 |
Current work | Hybrid-FHR (Expert + SE-TCN + CMFF) | CTU-UHB | pH < 7.15 | Acc: 96.8; Sen: 96.0 Spe: 97.5; Pre: 97.5 F1: 96.7 |