Table 3.
Evaluating COVID-19 omics data on the testing set: balanced accuracy, weighted F1, precision, and recall.
| Method | Balanced accuracy | Weighted F1 | Weighted precision | Weighted recall |
|---|---|---|---|---|
| Deep IDA with top 100 selected features | 0.76(0.08) | 0.79(0.06) | 0.81(0.06) | 0.80(0.06) |
| Deep IDA with top 10% selected features | 0.76(0.08) | 0.79(0.07) | 0.80(0.07) | 0.79(0.07) |
| Deep IDA | 0.70(0.07) | 0.76(0.05) | 0.77(0.06) | 0.76(0.05) |
| Deep GCCA + SVM with top 10% selected features | 0.55(0.16) | 0.61(0.13) | 0.66(0.12) | 0.64(0.11) |
| Deep GCCA +SVM | 0.61(0.14) | 0.64(0.12) | 0.68(0.10) | 0.65(0.11) |
| Deep GCCA + NCC with top 10% selected features | 0.60(0.12) | 0.64(0.12) | 0.68(0.11) | 0.64(0.12) |
| Deep GCCA +NCC | 0.67(0.08) | 0.67(0.08) | 0.70(0.08) | 0.67(0.08) |
| SIDA | 0.60(0.11) | 0.67(0.10) | 0.69(0.10) | 0.67(0.11) |
| PMA + SVM | 0.40(0.03) | 0.56(0.04) | 0.52(0.05) | 0.62(0.05) |
| SVM with top 10% selected features | 0.78(0.07) | 0.82(0.05) | 0.83(0.05) | 0.82(0.05) |
| SVM | 0.73(0.09) | 0.78(0.06) | 0.79(0.06) | 0.78(0.06) |
| NCC with top 10% selected features | 0.72(0.07) | 0.72(0.06) | 0.75(0.06) | 0.72(0.06) |
| NCC | 0.65(0.07) | 0.67(0.05) | 0.70(0.05) | 0.67(0.05) |
The top selected features are obtained by our proposed Deep IDA + Bi-Bootstrap. Each value is based on 20 random train-test splits of data. Mean value is followed by standard deviation in the parentheses.