Table 5 Mean AUCs of several variants of the proposed model
From: An interpretable neural network for outcome prediction in traumatic brain injury
Method | SHAP features | MRMR features | All 58 features |
|---|---|---|---|
TFNN | 0.794 (0.039) | 0.793 (0.039) | 0.786 (0.027) |
\(\ell _1\)-sparsity | 0.788 (0.033) | 0.794 (0.041) | 0.765 (0.022) |
\(\ell _0\)-sparsity | 0.799 (0.035) | 0.797 (0.041) | 0.784 (0.030) |
fewer rules | 0.784 (0.037) | 0.784 (0.042) | 0.775 (0.022) |