标题: 求助：生存曲线的标记问题 [打印本页] 作者: shiyiming 时间: 2010-4-12 22:13 标题: 求助：生存曲线的标记问题 [size=150:6f8pn2cb]我作出的生存曲线上只有删失值的标记没有正常值的标记，有时候连删失值的标记都没有。不会贴图，先请大伙帮忙看看程序吧。[/size:6f8pn2cb]
data yt7_2;
input t @@;
if t<0 then censor=1;
else censor=0;
if _N_<16 then group='A';
else group='B';
t=abs(t);
cards;
28 29 175 195 309 -377 -393 -421 -447 452 -709 -744 -770 -1106 1206
34 88 137 199 280 291 -299 -300 309 351 358 369 370 375 382 392 -429 451 -1119
;
proc lifetest method=pl plots=(s,ls);
time t * censor(1);
strata group;
proc lifetest method=lt plots=(s,h);
time t * censor(1);
strata group;
run;作者: shiyiming 时间: 2010-4-13 05:54 标题: Re: 求助：生存曲线的标记问题 When a METHOD = PL, also the default, was implemented in PROC LIFETEST, step-wise KM curve was given with censor symbol as circle by default except you specified CS = NONE or other symbols. I do not know why you did not see it in your first model. When METHOD = LT, lifetime table, a different style of survival was given, point-to-point connection. It is very dubious you will need the censored marked in this method, because in this type of plot, two censors between the consecutive events could be representing two different survivals (the slope is not horizontal at all) and that is against common sense. Anyhow, in your first method, you are supposed to see the censorship mark and you can change it by an option of CS = none/star/…and so on in the statement of LIFETEST.作者: shiyiming 时间: 2010-4-13 13:50 标题: Re: 求助：生存曲线的标记问题 我是看到一篇文献上的图是有标记的，自己画不出来。刚才确认了一下，是一款stat soft 公司的statistica软件画的。作者: shiyiming 时间: 2010-4-13 20:49 标题: Re: 求助：生存曲线的标记问题 我说的是SAS。