Table 4 Median AUCPR scores for different AutoML models scaled according to median random forest performance. Models with the best performance for each disease are indicated in bold
From: Benchmarking AutoML frameworks for disease prediction using medical claims
Metric: Average Precision | Lung Cancer | Prostate Cancer | Rheumatoid Arthritis | Type 2 Diabetes | IBD | CKD |
|---|---|---|---|---|---|---|
Model | Â | Â | Â | Â | Â | Â |
AutoSklearn (Average Precision) | 1.957 | 1.787 | 1.471 | 1.675 | 1.212 | 1.395 |
AutoSklearn (Balanced Accuracy) | 2.043 | 1.260 | 1.647 | 1.608 | 1.259 | 1.300 |
AutoSklearn (ROC AUC) | 1.870 | 2.102 | 1.353 | 1.592 | 1.235 | 1.337 |
H2O (AUC) | 3.217 | 2.213 | 1.706 | 1.750 | 1.259 | 1.428 |
H2O (AUCPR) | 3.261 | 1.961 | 1.706 | 1.767 | 1.259 | 1.420 |
TPOT (Average Precision) | 2.565 | 1.346 | 1.147 | 1.650 | 1.200 | 1.160 |
TPOT (Balanced Accuracy) | 0.696 | 0.457 | 1.324 | 1.033 | 0.976 | 0.984 |
TPOT (ROC AUC) | 1.522 | 1.087 | 1.559 | 1.650 | 1.129 | 1.074 |
Random Forest | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 | 1.000 |