Log in
Register|
mxf ☆ Italy, 2014-03-13 21:52 (4115 d 08:21 ago) Posting: # 12617 Views: 4,716 |
|
|
Dear All, after seeing other post in this forum, I would like to clear and point out the state of the art of incomplete data analysis. I haven't commercial softwares and usually work on R. Bear is a very good package, but it seems to be unable for performing analysis with incomplete dataset. If possible, I would appreciate any possible suggestion about incomplete data anlaysis with R or other not commercial software. Thank you at all and have patience for my terrible English! Max |
|
yjlee168 ★★★   Kaohsiung, Taiwan, 2014-03-18 11:17 (4110 d 18:56 ago) @ mxf Posting: # 12645 Views: 3,749 |
|
|
Dear Max, ❝ […] I haven't commercial softwares and usually work on R. Bear is a very good package, but it seems to be unable for performing analysis with incomplete dataset. If there is any appropriate statistic model for a incomplete crossover study design (I assume that it's your study design), it should be no problem to build this function into bear. BTW, I am thinking how to present the incomplete data in summary table. For instance, we require to list T/R values for pivotal parameters (such as Cmax and AUCs). If there is no T, then the T/R are all zero; If there is no R, oops! We will get all T/R as "inf"(inity) which is 'not a number (NaN)' in R. Please see this post by Helmut and others. In that post, Helmut got the same 90% CI for Cmax for "subject(sequence) fixed, n=24/23 (i.e., including the incomplete data)" and "subject(sequence) fixed, n=23/23 (without incomplete data)" using a commercial software. ❝ If possible, I would appreciate any possible suggestion about incomplete data analysis with R or other not commercial software. yes, only if possible... Hope this can help. — All the best, -- Yung-jin Lee bear v2.9.2:- created by Hsin-ya Lee & Yung-jin Lee Kaohsiung, Taiwan https://www.pkpd168.com/bear Download link (updated) -> here |
|
Dr_Dan ★★ Germany, 2014-03-18 12:45 (4110 d 17:28 ago) @ mxf Posting: # 12649 Views: 3,635 |
|
|
Dear Max Although not written for BE studies guideline CPMP/EWP/1776/99 could give you some information. One approach would be to determine the highest and lowest possible concentration which still would lead to a positive study result and discuss these concentration in a sensitivity analysis. Unfortunately I am not a statistician and can not provide advice with regard to statistical analysis software. Kind regards Dr_Dan — Kind regards and have a nice day Dr_Dan |
|
mxf ☆ Italy, 2014-03-26 18:58 (4102 d 11:14 ago) @ Dr_Dan Posting: # 12718 Views: 3,610 |
|
|
Sorry for my delay, but I had a problem in logged in!  Thanks for all your suggestions. I will think something... Max |
Ing. Helmut Schütz