Table 1 Predictive performance

From: Deep hierarchical subtyping of multi-organ systemic sclerosis trajectories - a EUSTAR study

 

Continuous (MAE)

Categorical (weighted F1)

Ours

0.436 ± 0.006

0.879 ± 0.002

MLP

0.464 ± 0.003

0.872 ± 0.003

Ours – without feature masking

0.439 ± 0.006

0.883 ± 0.004

Cohort mean

0.572 ± 0.009

0.756 ± 0.000

Patient-specific

0.522 ± 0.008

0.677 ± 0.000

  1. We compare our final model ("Ours”) to multiple ML and non-ML-driven baselines in terms of predictive performance for variables in G. We computed the Mean Absolute Error (MAE) for continuous variables and the weighted F1 score for binary/categorical variables.
Back to article page