A scoping methodological review of simulation studies comparing statistical and machine learning approaches to risk prediction for time-to-event data

Table 4 Number of covariates, distribution type and relationships between covariates in each article’s simulations

	Covariates
	Number of covariates	Distribution				Relationships
	Number of covariates	Binomial	Normal	Uniform	Real Data	Independent	Correlation	Interaction, e.g. X_{3 =} X₁X₂	Correlation and interaction
Geng et al. (2014) [26]	2		✓				✓		✓
Golmakani et al. (2020) [31]	50, 1000		✓			✓		✓
Gong et al. (2018)* [27]	2, 3, 250	✓	✓			✓		✓	✓
Hu and Steingrimsson (2018) [28]	50		✓				✓		✓
Katzman et al. (2018) [29]	10			✓		✓
Lowsky et al. (2012)** [25]	13				✓
Omurlu et al. (2009) [24]	5	✓		✓		✓
Steingrimsson and Morrison (2020) [32]	30, 100		✓				✓
Wang and Li (2019) [30]	500, 1000, 2000, 5000		✓				✓
Xiang et al. (2000) [23]	2, 4	✓	✓			✓		✓

*Gong et al. (2018) [27] used distributions and parameter values to model clinical data in their clinically relevant datasets and included three data-generating mechanisms where the covariate relationships were modelled to be clinically relevant
**Lowsky et al. (2012) [25] used real clinical data for their covariates and so exact relationships are unknown

ISSN: 2397-7523