Choose Language Hide Translation Bar
Highlighted
Mi
Mi
Level III

Survival

Hello, everyone,
how should I build the file
to calculate a survival analysis.

In my file i have
27 days,
treatment 1and 2,
Opening_total,
dead_treatment,
alive_treatment and
alive_treatment censored 0=living
                                         1=dead

The way I built the file will not take the dead_treatment into account...

Total Number T=1 is 120 and T=2 114
              with T=1 live 41 and are 79 dead
                      T=2 live 6 and are 108 dead

 

Can somebody PLEASE help me.?
I really don't know what I'm doing wrong.

 

Best regards.

 

Mi

2 ACCEPTED SOLUTIONS

Accepted Solutions
Highlighted

Re: Survival

You do not define what you mean by "survival analysis."JMP expects the data in a particular layout for some analyses.

 

Your data includes 6 blocks of 27 days for each treatment. What is the blocking about?

 

Your data appear as though they came from a spreadsheet and you were setting it up for a manual calculation. That step is not necessary for JMP. JMP expects the following information for survival for each observation (subject):

 

  • The life time. This time can be exact (dead) or censored (alive).
  • An indicator of exact or censored life.

That information is all that is minimally required. You also have treatment and multiple deaths on some days. This layout can include a count of observations at a given time, which can be used in the Freq role.

 

I assume that the 120 subjects for Treatment = 1 and the 114 subjects for Treatment = 2 are the same throughout the 27 days. That is, you are not treating 120 new subjects each day. So a probit analysis is probably not what you want. You can compare the life distribution of the two populations with Analyze > Reliability and Survival > Life Distribution.

 

Set your data up like this:

 

table.JPG

 

I am only showing 1 block of 27 days for each treatment. Notice that only the exact deaths are entered as observations and all the alive subjects are counted at the end of the study. (Type I censoring).

 

Then set up the analysis like this:

 

dialog.JPG

 

The results appear like this:

 

platform.JPG

 

Please see JMP Help > JMP Documentation Library > Reliability and Survival guide for more information.

 

I attached my working data table so that you do not have to start from scratch.

Learn it once, use it forever!

View solution in original post

Highlighted

Re: Survival

I was asking about which analysis you wanted because some survival analyses expect a specific arrangement of the data table. The arrangement that I showed you should work with most analyses in JMP. That is, minimally you have the life time and the censor code, if any data are censored.

 

My question about subjects was to clarify if you started with 120 or 114 subjects and the risk pool was the same for the entire duration (27 days) or if it changed daily. I believe that it is the same subjects for the duration. I asked because it were the other case then a probit analysis might be what you wanted.

 

The Survival platform is a legacy object in JMP. Life Distribution is a newer replacement platform for Survival.

 

You can get the Kaplan-Meier estimate for Survival or from Life Distribution.

 

You can compare groups in Life Distribution. There is a special tab in the launch dialog for such an analysis.

 

You can get the hazard using the Hazard profiler in Life Distribution.

 

Cox regression (proportional hazards model) is available through Analyze > Fit Model.

 

See Help > JMP Documentation Library > Reliability and Survival guide for more information.

Learn it once, use it forever!

View solution in original post

3 REPLIES 3
Highlighted

Re: Survival

You do not define what you mean by "survival analysis."JMP expects the data in a particular layout for some analyses.

 

Your data includes 6 blocks of 27 days for each treatment. What is the blocking about?

 

Your data appear as though they came from a spreadsheet and you were setting it up for a manual calculation. That step is not necessary for JMP. JMP expects the following information for survival for each observation (subject):

 

  • The life time. This time can be exact (dead) or censored (alive).
  • An indicator of exact or censored life.

That information is all that is minimally required. You also have treatment and multiple deaths on some days. This layout can include a count of observations at a given time, which can be used in the Freq role.

 

I assume that the 120 subjects for Treatment = 1 and the 114 subjects for Treatment = 2 are the same throughout the 27 days. That is, you are not treating 120 new subjects each day. So a probit analysis is probably not what you want. You can compare the life distribution of the two populations with Analyze > Reliability and Survival > Life Distribution.

 

Set your data up like this:

 

table.JPG

 

I am only showing 1 block of 27 days for each treatment. Notice that only the exact deaths are entered as observations and all the alive subjects are counted at the end of the study. (Type I censoring).

 

Then set up the analysis like this:

 

dialog.JPG

 

The results appear like this:

 

platform.JPG

 

Please see JMP Help > JMP Documentation Library > Reliability and Survival guide for more information.

 

I attached my working data table so that you do not have to start from scratch.

Learn it once, use it forever!

View solution in original post

Highlighted
Mi
Mi
Level III

Re: Survival

Hallo Mark,

 

what exactly is the difference between Survival and Life Distribution?

"That is, you are not treating 120 new subjects each day. "

Is this the answer? Treating new subjects each day?

 

 

Here are the answers to your questions.

 

"You do not define what you mean by "survival analysis."JMP expects the data in a particular layout for some analyses."

I would like to make the following calculations with Survival Analysis:

Kaplan-Meier -> Survival time, Median survival time=50

 

Log-Rank Test -> Compare survival time of 2 groups,

 

Hazard -> the risk of suffering an event at the exact time of the event = current mortality  

                  rate (depends on the time course)

 

Hazard ratio -> Difference of survival times (comparison of two groups)

 

Cox regression-> multivariable models => how are the groups indiscriminate

 

"Your data includes 6 blocks of 27 days for each treatment. What is the blocking about?"

My file contains->3 trials, days, 6 cages(3 for each treatment), treatment, dead.

About suggestions where(how) I find to calculate Kaplan-Meier, Hazard, Hazart ratio correct I am very grateful.

 

Have a nice evening!

 

Best regards

Mi

 

Highlighted

Re: Survival

I was asking about which analysis you wanted because some survival analyses expect a specific arrangement of the data table. The arrangement that I showed you should work with most analyses in JMP. That is, minimally you have the life time and the censor code, if any data are censored.

 

My question about subjects was to clarify if you started with 120 or 114 subjects and the risk pool was the same for the entire duration (27 days) or if it changed daily. I believe that it is the same subjects for the duration. I asked because it were the other case then a probit analysis might be what you wanted.

 

The Survival platform is a legacy object in JMP. Life Distribution is a newer replacement platform for Survival.

 

You can get the Kaplan-Meier estimate for Survival or from Life Distribution.

 

You can compare groups in Life Distribution. There is a special tab in the launch dialog for such an analysis.

 

You can get the hazard using the Hazard profiler in Life Distribution.

 

Cox regression (proportional hazards model) is available through Analyze > Fit Model.

 

See Help > JMP Documentation Library > Reliability and Survival guide for more information.

Learn it once, use it forever!

View solution in original post