Favoring the hierarchical constraint in penalized survival models for randomized trials in precision medicine

BMC Bioinformatics

Table 1 Scenarios of the simulation study

Scenarios		Biomarker effect: \(ln(HR)=\beta \pm 0.5\gamma\)		Average censoring probability
Scenarios		Ctrl: \(\beta -0.5\gamma\)	Exp: \(\beta +0.5\gamma\)	Average censoring probability
1	No effect	0	0	0.10
2	\(q_{Pe}= 1\) predictive biomarker	-\(0.5 \, ln(0.5)\)	\(0.5 \, ln(0.5)\)	0.13
3	\(q_{Pe}= 10\) predictive biomarkers	-\(0.5 \, ln(0.5)\)	\(0.5 \, ln(0.5)\)	0.20
4	\(q_{Pe}= 10\) predictive biomarkers	-\(0.5 \, ln(0.5)\)	\(0.5 \, ln(0.5)\)	0.31
4	\(q_{Po} = 10\) prognostic biomarkers	ln(0.5)	ln(0.5)	0.31

HR hazard ratio, Ctrl control arm, Exp experimental arm, \(q_{Po}\) number of prognostic biomarkers, \(q_{Pe}\) number of predictive biomarkers

ISSN: 1471-2105