Discussion of "A risk-based measure of time-varying prognostic discrimination for survival models," by C. Jason Liang and Patrick J. Heagerty.
Discussion of "A risk-based measure of time-varying prognostic discrimination for survival models," by C. Jason Liang and Patrick J. Heagerty.
复制标题
作者：
Michael H;Tian L
我们祝贺梁和Heagerty巧妙地推广了区分斜率(DS)，以便在生存分析中评估生物标记物的时变区分性能。最近，人们越来越有兴趣开发良好的生物标记物(或一组生物标记物)的表现来预测生存结果。例如，在评估一种新的生物标记物，如C反应蛋白或冠状动脉钙化评分相对于预测个体患者未来心血管风险的传统危险因素的增量价值时，对这些措施的适当选择和解释是至关重要的(Lloyd-Jones等人，2006年)。为此，人们提出了许多总结统计量的推广，如c-−指数和r2。梁和Heagerty提出的基于风险函数的度量是独一无二的，这使得它特别适合于基于风险函数的回归模型，如流行的Cox模型。具体地说，梁和希格蒂提出了这个比例
We congratulate Liang and Heagerty for their clever generalization of the discrimination slope (DS) to allow for evaluation of the time-varying discrimination performance of a biomarker in survival analysis. Lately, there has been growing interest in developing good measures of the performance of a biomarker (or a set of biomarkers) for predicting survival outcomes. The appropriate choice and interpretation of such measures are crucial, for example, in evaluating the incremental value of a novel biomarker such as CRP or coronary artery calcium score relative to conventional risk factors in predicting future cardiovascular risk for an individual patient (Lloyd-Jones et al., 2006). To this end, many generalizations of summary statistics such as c− index and R2 have been proposed. Liang and Heagerty’s proposed metric is unique in being based on the risk function, making it particularly suitable for hazard functionbased regression models such as the popular Cox model. Specifically, Liang and Heagerty propose the ratio