如何模拟产生指定HR的生存数据

指定生存分析风险比HR,  模拟产生一个相应的生存数据集。

如： 指定HR=4，模拟产生数据集SURV， SURVE中有变量GRP(分组：1，0)， TIME(生存时间)，STATUS(状态，0表示右删失)。

when simulating survival data, you need to know or assume the distribution of the survival time. For example, in the following code, i assume t~exp(1), and the right censored at end of follow-up by 2 years.
I may develop more dists in my blog later on.

1. data x;
   
2. call streaminit(1234);
   
3. lambda=1;/*Survival time ~ exponential(1)*/
   
4. beta =log(4);
   
5. tc=2.0;/*End of followup data T=2*/
   
6. do i=1 to 800;
   
7. censor=0;
   
8. x =rand('table',0.5)-1;/*0 vs 1 = 50% vs 50%*/
   
9. u=rand('uniform');
   
10. T =-log(u)/(lambda *exp(beta*x));
    
11. if T >tc then do;
    
12. censor=1;
    
13. T=tc;
    
14. end;
    
15. output;
    
16. end;
    
17. run;

复制代码

JingJu

jingju11 发表于 2015-2-28 07:17
when simulating survival data, you need to know or assume the distribution of the survival time. For ...

非常感谢， 期待！

jingju11 发表于 2015-2-28 07:17
when simulating survival data, you need to know or assume the distribution of the survival time. For ...

我找到了Ralf Bender的PAPER：Generating Survival Times to Simulate Cox Proportional Hazards Models  , 非常好的参考，也非常感谢京剧大哥的帮助！

jingju11 发表于 2015-2-27 20:32
when simulating survival data, you need to know or assume the distribution of the survival time. For ...

再次请教京剧大哥，如果考虑再加一个和分组相关的协变量， 可行吗？