Reference no: EM131883019

Identify the (time, censor) pair for each of the following analyses:

Do you have the answers to your Biology questions?
These problems use the Copelan bone marrow transplant dataset, available in the KMsurv package.

Problem 1

You need to replace "time" and "censor" in the example with the actual names of the variables you are using as the time and censoring variable (these are different for each "time to" and "indicator" pair)
Use the cleaned ‘bmt' dataset from Homework 3.

Identify the (time, censor) pair for each of the following analyses:
a. Patient survival -
1= indicates that the patient died from event
2. = still alive = Time
3 = died from causes unrelated to - time
Cens alive- 1 if status=2, 0 otherwise Tmax
Cens all 1= 1 if status =2, 0 otherwise Tmax
b. Disease-free survival = Event
c. Time to chronic graft versus host disease- duration would be at least T=>t Tmax?
d. Time to platelet recovery= Tmax

Identify the (time, censor) pair for each of the following analyses:
a. Patient survival -
1= indicates that the patient died from event
2. = still alive = Time
3 = died from causes unrelated to - time
Cens alive- 1 if status=2, 0 otherwise Tmax
Cens all 1= 1 if status =2, 0 otherwise Tmax
b. Disease-free survival = Event
c. Time to chronic graft versus host disease- duration would be at least T=>t Tmax?
d. Time to platelet recovery= Tmax

Identify the (time, censor) pair for each of the following analyses:
a. Patient survival -
1= indicates that the patient died from event
2. = still alive = Time
3 = died from causes unrelated to - time
Cens alive- 1 if status=2, 0 otherwise Tmax
Cens all 1= 1 if status =2, 0 otherwise Tmax
b. Disease-free survival = Event
c. Time to chronic graft versus host disease- duration would be at least T=>t Tmax?
d. Time to platelet recovery= Tmax

Problem 2
plotted survival probabilty for all three at once looking at t1,d1 compared (~) to group according to the code in the lectures. For part c, what I did was used the KM graph and looked at the probability of survival at 365 days (by putting and abline at 365) and guesstimating the probability with my very good eyesight

> library(survival)
> library(KMsurv)
> Km = survfit(Surv(time,censor)~1)
> Km = survfit(Surv(time,censor)~ group)

For patient post-transplant survival time,
a. Calculate Kaplan-Meier curves for each disease group. Plot the curves and comment on what you see.
b. Check whether these curves are different using the log-rank test.
c. What is the estimated one year survival of the three disease types? How does this compare with the estimates generated from a logistic regression model? You can use your results from Homework 3 if you generated the simple model (OneYearSurvival ~ group), but if you do, make sure that you are using the same data exclusions/modifications in each model.

Problem 3
a. Build a multivariate Cox model for post-transplant survival. As before, manually build your best model, then run an automated variable selection algorithm and compare the results.
b. Compare your results to the logistic regression model from Homework 3. Which do you think is the better approach, and why?
c. Build another Cox model predicting disease-free survival. Are the predictors of disease-free survival different from those for patient survival?