Table 3 Experimental results for LBW prediction using anomaly detection algorithms.
From: Node embedding-based graph autoencoder outlier detection for adverse pregnancy outcomes
LBW prediction | Parous | Nulliparous | ||
|---|---|---|---|---|
Method | AUC | AUC-PR | AUC | AUC-PR |
Original GAE | 0.4982 | 0.0966 | 0.4891 | 0.1584 |
FastRP 16 | 0.5780 | 0.1280 | 0.5116 | 0.2434 |
FastRP 32 | 0.5773 | 0.1349 | 0.5796 | 0.2305 |
FastRP 64 | 0.4876 | 0.1002 | 0.5248 | 0.1970 |
Node2Vec 16 | 0.5311 | 0.1522 | 0.5116 | 0.2389 |
Node2Vec 32 | 0.5506 | 0.1541 | 0.5017 | 0.1722 |
Node2Vec 64 | 0.5268 | 0.1128 | 0.5244 | 0.1945 |
combine 16 | 0.5654 | 0.1439 | 0.5092 | 0.2058 |
combine 32 | 0.6077 | 0.1545 | 0.5428 | 0.1877 |
combine 64 | 0.5058 | 0.1086 | 0.5384 | 0.2022 |
