Table 1 Performance (AUC) of machine learning classifiers differentiating between real healthy control subjects and real as well as synthetic patients, respectively.
From: Generation of realistic synthetic data using Multimodal Neural Ordinary Differential Equations
PPMI | Trained on synthetic PPMI tested on real | NACC | Trained on synthetic NACC tested on real | |
|---|---|---|---|---|
Real patients | 0.97 ± 0.02 | 0.90 ± 0.01 | ||
Synthetic (prior sampling) | 0.97 ± 0.02 | 0.97 ± 0.002 | 0.96 ± 0.01 | 0.85 ± 0.002 |
Synthetic (posterior sampling) | 0.97 ± 0.01 | 0.98 ± 0.002 | 0.93 ± 0.01 | 0.87 ± 0.002 |
Synthetic (VAMBN) | 0.96 ± 0.01 | 0.98 ± 0.004 | 0.88 ± 0.01 | 0.89 ± 0.001 |
