I really appreciate all the work that has gone into making this and other videos. They are simply invaluable for me getting to grips with R commander. You don't mention adjusting or controlling for a variable in the Cox Model. I would find this very useful. Many thanks
Thank you, that was very helpful. I found it while trying to figure out how to write a Cox PH model in R (I'm familiar with R but not with survival models) and this was exactly what I needed. A suggested update for future videos: most people now use RStudio rather than R commander, so it would be interesting to add an example using that, one day, although as it stands this is already super useful.
I think that there is a small typo on p 13 of 22 of the handout (12:32 minutes). It should be 'e' to the power of'.451+34.74' for a total of 'e' to the power of 35.191, not 44.191
Cox regression is for modelling durations in general, not necessarily duration of life! So, a HR > 1 is not necessary something bad. Imagine for example that you are wishing to model the duration of a disease. The higher the HR (above 1), the smaller will the duration of the disease be. I am myself working for years on and with Cox models (and I published several results on this). For most of the problems I am working on, I will put the smiling face (smiley) of the HR >1 side..
Good point Eric? I was just trying to get the students to understand the examples I provided which is as stated - my pdf handouts go into more depth. I probably need to post one where length of time is the undesirable aspect. Thanks for the comment.
Very nice presentation. Thank you, Robin. I never used the R commander. Imported the survival and tcltk packages, but my plug-ins under Tools option is grayed out. I am not sure if I should upload the plug-in and how?
I have a question. Suppose I had data with two categorical covariates, A and B, which creates four groups (A=0 B=0, A=1 B=0, A=0 B=1, and A=1 B=1). The P values for pairwise comparisons are thus: A=0 B=0 vs A=0 B=1:
@theoldorganplayer This was incredibly helpful, thank you so much for doing it. I would love to see a bit more detail on the various options/choices available in running the SPSS analysis and what they all mean. Also would like to see it with more than two variables, since this seems to introduce additional choices in the analysis. Thanks again!
Hello there, Can anyone please help me figure which survival analysis test to do for organ transplant survival comparing two groups using spss. and how should I find out covariates?
Great video thanks ! Just had a quick question. When you check for treat AND age in SPSS in the cox P-H model how is the interpretation of your results ? treatment has no effect on survival but age has an influence on survival with treatment ?
Very good, thank you - you've helped me understanding my research better (im researching the influence of patent-grants onthe time between venture capital funding phases. I'm working on creating a duration / hazard model to get this done..
This could be a silly question is cox regression also known as cox multivariate analysis? If yes what is the different between univariate and multivariate Thanks
Hi, I've been trying to move from spss to R. This video was somewhat useful. However when I add categorical variables with more than 2 categories, R starts having problem. For example I'm working with cancer datasets where the stage of cancer is important in the cox model. So stage 1,2,3,4 should be different. Spss shows this and calculates pval/hazard ratios for all stages. If i use coxph in R, it only shows p value for stage 2,3,4 and furthermore answer is different from spss. Any advice?
Hi billselig the link to my material in the info box below the video provides a link to a excellent tutorial by Chan using arounf 4-6 covariates and also a model development process. All the best robin
I have been thinking a kinda confusing question to ask for someone who could give me a answer: when we deal with the multiple variates in survival analysis, why we have to do univariate analysis first and then multivariate analysis? Cann´t we do pool all the variables into one multivariate analysis model? And further, we only pick up the variable with p
I think you need to revise you basic knowledge of multiple linear regression - that will help you understand the situation. have a look at my book chapter on that title (available from amazon.uk. note that 'multiple' regression (many inputs one output) of any variety is very different from multivariate analysis (input any number outputs any number but classed as multidimensionally normally distributed
Seems like a lot of people are looking at this video - Be interesting in some feedback, specifically would people like more about this topic - any suggestions? Robin
Yep R treats categorical variables differently be default compared to SPSS (its due to how it sets the contrasts up) have a look at the survival plugin article by Fox search "The RcmdrPlugin.survival Package: Extending the R Commander Interface to Survival Analysis by john Fox"
I really appreciate all the work that has gone into making this and other videos. They are simply invaluable for me getting to grips with R commander. You don't mention adjusting or controlling for a variable in the Cox Model. I would find this very useful. Many thanks
Thank you, that was very helpful. I found it while trying to figure out how to write a Cox PH model in R (I'm familiar with R but not with survival models) and this was exactly what I needed.
A suggested update for future videos: most people now use RStudio rather than R commander, so it would be interesting to add an example using that, one day, although as it stands this is already super useful.
I think that there is a small typo on p 13 of 22 of the handout (12:32 minutes). It should be 'e' to the power of'.451+34.74' for a total of 'e' to the power of 35.191, not 44.191
Thanks Robin.. very helpful, would appreciate a simplified video on checking model assumptions (proportional hazards)
Cox regression is for modelling durations in general, not necessarily duration of life! So, a HR > 1 is not necessary something bad. Imagine for example that you are wishing to model the duration of a disease. The higher the HR (above 1), the smaller will the duration of the disease be. I am myself working for years on and with Cox models (and I published several results on this). For most of the problems I am working on, I will put the smiling face (smiley) of the HR >1 side..
i'd like to see more detail about cox model building and selection. Does R have a function for model selection?
Good point Eric?
I was just trying to get the students to understand the examples I provided which is as stated - my pdf handouts go into more depth. I probably need to post one where length of time is the undesirable aspect. Thanks for the comment.
Very nice presentation. Thank you, Robin. I never used the R commander. Imported the survival and tcltk packages, but my plug-ins under Tools option is grayed out. I am not sure if I should upload the plug-in and how?
+Peter Paprzycki Ok. Figured it out. It requires the ‘RcmdrPlugin.survival’ plug-in package in case someone was wondering.
+Peter Paprzycki I met the same question. How did you solve it? I didn't get what you said here
i want to know how these datas were inputted in the spss because im actually using accident cases that happened in the past.... please please
I have a question. Suppose I had data with two categorical covariates, A and B, which creates four groups (A=0 B=0, A=1 B=0, A=0 B=1, and A=1 B=1). The P values for pairwise comparisons are thus:
A=0 B=0 vs A=0 B=1:
Very useful video, thank you. Would appreciate something on diagnostics/checking assumptions for Cox proportional hazards. Best wishes, Ewen
@theoldorganplayer This was incredibly helpful, thank you so much for doing it. I would love to see a bit more detail on the various options/choices available in running the SPSS analysis and what they all mean. Also would like to see it with more than two variables, since this seems to introduce additional choices in the analysis. Thanks again!
Hello there, Can anyone please help me figure which survival analysis test to do for organ transplant survival comparing two groups using spss. and how should I find out covariates?
Thank you, your videos and notes are very useful.
Great video thanks ! Just had a quick question. When you check for treat AND age in SPSS in the cox P-H model how is the interpretation of your results ? treatment has no effect on survival but age has an influence on survival with treatment ?
Very good, thank you - you've helped me understanding my research better (im researching the influence of patent-grants onthe time between venture capital funding phases. I'm working on creating a duration / hazard model to get this done..
This could be a silly question is cox regression also known as cox multivariate analysis?
If yes what is the different between univariate and multivariate
Thanks
Hi, I've been trying to move from spss to R. This video was somewhat useful. However when I add categorical variables with more than 2 categories, R starts having problem. For example I'm working with cancer datasets where the stage of cancer is important in the cox model. So stage 1,2,3,4 should be different. Spss shows this and calculates pval/hazard ratios for all stages. If i use coxph in R, it only shows p value for stage 2,3,4 and furthermore answer is different from spss. Any advice?
Hi billselig the link to my material in the info box below the video provides a link to a excellent tutorial by Chan using arounf 4-6 covariates and also a model development process. All the best robin
Mr Beaumont THANK YOU - feedback is it's all good!!!!!
I have been thinking a kinda confusing question to ask for someone who could give me a answer: when we deal with the multiple variates in survival analysis, why we have to do univariate analysis first and then multivariate analysis? Cann´t we do pool all the variables into one multivariate analysis model? And further, we only pick up the variable with p
I think you need to revise you basic knowledge of multiple linear regression - that will help you understand the situation. have a look at my book chapter on that title (available from amazon.uk. note that 'multiple' regression (many inputs one output) of any variety is very different from multivariate analysis (input any number outputs any number but classed as multidimensionally normally distributed
Very useful video even for beginers. Thanks very much
Seems like a lot of people are looking at this video - Be interesting in some feedback, specifically would people like more about this topic - any suggestions?
Robin
hi robin do you have any examples with time varying covariates?
good suggestion will do one when i get some time:)
yep same thing. Univariate = one predictor, multivariate = more than one predictor.
hazard ratio: e^44.191 is coming horrible value in calculator. then how did you write hazard ratio: e^44.191 = 3.788 ??? plz explan
I mean in Cox regression model
Yep R treats categorical variables differently be default compared to SPSS (its due to how it sets the contrasts up) have a look at the survival plugin article by Fox search "The RcmdrPlugin.survival Package: Extending the R Commander Interface to Survival Analysis by john Fox"
thank you for sharing
thank you!!!
Destroy the mouse click sound that you have prior to recording. Just a presentation tip. The rest was great. Thanks
IT IS AWESOME