Table 3 Results of the comparative experiments evaluating overfitting. Per dataset, the least overfitting model (bold) and the models within the 95% CI of the least overfitting model (italics) are highlighted
From: Bayesian parametric models for survival prediction in medical applications
Characteristic | CoxPH N = 47a | DeepSurv N = 47a | BPS Exp N = 47a | BPS Wb N = 47a | BPS WbNN N = 47a | RSF N = 47a |
|---|---|---|---|---|---|---|
ACTG | ||||||
C-Index | 0.031 [0.005—0.056] | 0.096 [0.059—0.131] | 0.038 [0.011—0.063] | 0.037 [0.007—0.065] | 0.043 [0.015—0.074] | 0.090 [0.058—0.122] |
GBCS | ||||||
C-Index | 0.007 [-0.014—0.029] | 0.032 [0.008—0.055] | 0.014 [-0.006—0.032] | 0.016 [-0.005—0.038] | 0.030 [0.010—0.052] | 0.053 [0.034—0.074] |
PBC | ||||||
C-Index | 0.018 [-0.004—0.040] | 0.064 [0.040—0.089] | 0.015 [-0.005—0.036] | 0.018 [-0.008—0.041] | 0.017 [-0.003—0.037] | 0.074 [0.053—0.095] |
Veteran | ||||||
C-Index | 0.022 [0.002—0.044] | 0.063 [0.035—0.089] | 0.029 [-0.005—0.062] | 0.027 [-0.005—0.056] | 0.037 [0.014—0.060] | 0.066 [0.034—0.097] |
WHAS | ||||||
C-Index | 0.028 [0.004—0.051] | 0.039 [0.014—0.066] | 0.020 [-0.005—0.046] | 0.016 [-0.009—0.040] | 0.034 [0.012—0.057] | 0.098 [0.071—0.125] |