In this text, the methodology developed by Tian et al. is verified by the author via a number of numerical simulations: an arbitrary collection of random variables are generated to represent baseline covariates, and “true” treatment effect for each subject is calculated using a preset formula. By coding the treatment variable as ±1 and fitting the products of the treatment variable and baseline covariates (which essentially are treatment/covariate interaction terms) in a LASSO regression model, a score could be constructed to select a subgroup of patients who may benefit from a specific treatment. Subsequently, the method is applied to the collection of random variables to determine treatment scores for all subjects. Finally, Spearman’s correlation coefficients between treatment scores and treatment effect are calculated to evaluate the performance of the score. A high Spearman’s correlation coefficient suggests that the calculated score is a good predictor of “true” treatment effect, and vice versa.