Skip to main content
Fig. 1 | BMC Medical Research Methodology

Fig. 1

From: Robust estimation of the expected survival probabilities from high-dimensional Cox models with biomarker-by-treatment interactions in randomized clinical trials

Fig. 1

Schematic representation of the proposed framework. 1CV and 2CV: single and double cross-validation, ĥ 0(t) baseline hazard at time t, \( \widehat{\alpha},\widehat{\boldsymbol{\beta}},\widehat{\boldsymbol{\gamma}} \): estimated regression parameters, \( {\widehat{\varphi}}_i = {\displaystyle \sum_{j=1}^p}{\widehat{\beta}}_j{X}_{i j} \): prognostic score, \( {\widehat{\eta}}_i = {\displaystyle \sum_{j=1}^p}{\widehat{\gamma}}_j{X}_{i j} \): treatment-effect modifying score, \( \widehat{\boldsymbol{\pi}} \): entire linear predictor of the selected model, iBrier: integrated Brier score, C: Uno’s C-statistic, ΔC: different in arm-specific Uno’s C-statistics, Ŝ(t): estimated survival probability at time t

Back to article page