Table 4 Ablation study of the key modules
From: MOCAT: multi-omics integration with auxiliary classifiers enhanced autoencoder
Dataset | Method | ACC (95% CI) | F1 (95% CI) | AUC (95% CI) |
ROSMAP | \(\overline{\text {AC}}\): AE\(_{\text {os}}\)+AE\(_{\text {f}}\)+Att+ConfNet | 79.4\(^*\) (76.9-81.9) | 79.6\(^*\) (76.6-82.6) | 88.2\(^*\) (87.3-89.1) |
\(\overline{\text {Att}}\): AE\(_{\text {os}}\)+AC+AE\(_{\text {f}}\)+ConfNet | 86.7 (85.8-87.6) | 86.5 (85.8-87.2) | 92.3 (91.8-92.8) | |
\(\overline{\text {ConfNet}}\): AE\(_{\text {os}}\)+AC+AE\(_{\text {f}}\)+Att | 85.5\(^*\) (84.5-86.5) | 85.6\(^*\) (84.6-86.6) | 92.2 (91.3-93.1) | |
Ours | 87.6 (86.7-88.5) | 87.5 (86.8-88.2) | 92.3 (91.2-93.4) | |
LGG | \(\overline{\text {AC}}\): AE\(_{\text {os}}\)+AE\(_{\text {f}}\)+Att+ConfNet | 80.0\(^*\) (78.5-81.5) | 78.2\(^*\) (76.5-79.9) | 88.8 (88.1-89.5) |
\(\overline{\text {Att}}\): AE\(_{\text {os}}\)+AC+AE\(_{\text {f}}\)+ConfNet | 83.6\(^*\) (83.0-84.2) | 83.4\(^*\) (82.5-84.3) | 88.9 (88.3-89.5) | |
\(\overline{\text {ConfNet}}\): AE\(_{\text {os}}\)+AC+AE\(_{\text {f}}\)+Att | 84.1\(^*\) (83.5-84.7) | 83.8\(^*\) (82.8-84.8) | 89.0 (88.5-89.5) | |
Ours | 85.1 (84.4-85.8) | 85.1 (84.1-86.1) | 88.5 (88.0-89.0) | |
Dataset | Method | ACC (95% CI) | F1_w (95% CI) | F1_m (95% CI) |
BRCA | \(\overline{\text {AC}}\): AE\(_{\text {os}}\)+AE\(_{\text {f}}\)+Att+ConfNet | 87.8 (87.2-88.4) | 88.0 (87.1-88.9) | 84.5 (82.4-86.6) |
\(\overline{\text {Att}}\): AE\(_{\text {os}}\)+AC+AE\(_{\text {f}}\)+ConfNet | 87.4\(^*\) (86.8-88.0) | 87.7\(^*\) (87.1-88.3) | 84.8\(^*\) (84.7-84.9) | |
\(\overline{\text {ConfNet}}\): AE\(_{\text {os}}\)+AC+AE\(_{\text {f}}\)+Att | 87.9\(^*\) (87.5-88.3) | 88.1\(^*\) (87.7-88.5) | 85.0\(^*\) (84.4-85.6) | |
Ours | 88.5 (88.0-89.0) | 88.9 (88.4-89.4) | 86.2 (85.2-87.2) |