Scribd
Upload
195 views

Survival Analysis Assignment 2

1. Describe the parametric inference procedures for estimating survival functions and parameters under different distributions (exponential, gamma, Weibull, log-normal) and censoring scenarios. 2. Construct likelihoods to estimate parameters of exponential and Weibull distributions based on relapse and death times, including with truncated sampling. 3. Derive the maximum likelihood estimator of the hazard rate parameter for an exponential survival distribution with independent right censoring. 4. Estimate the product limit survival function, median time, and conduct other analyses for remission duration data with censoring. 5. Compare immune therapies by estimating survival functions non-parametrically, testing differences in medians and distributions, and comparing patient characteristics between treatment groups

Uploaded by

girish
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
195 views

Survival Analysis Assignment 2

1. Describe the parametric inference procedures for estimating survival functions and parameters under different distributions (exponential, gamma, Weibull, log-normal) and censoring scenarios. 2. Construct likelihoods to estimate parameters of exponential and Weibull distributions based on relapse and death times, including with truncated sampling. 3. Derive the maximum likelihood estimator of the hazard rate parameter for an exponential survival distribution with independent right censoring. 4. Estimate the product limit survival function, median time, and conduct other analyses for remission duration data with censoring. 5. Compare immune therapies by estimating survival functions non-parametrically, testing differences in medians and distributions, and comparing patient characteristics between treatment groups

Uploaded by

girish
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 2

Survival Analysis – Assignment 2

1. Write down the complete parametric inference procedures of parameters and survival
function estimation under type-I, type-II and random censoring, when the survival time is
distributed according to the following distributions.

a. Exponential distribution
b. Gamma distribution,
c. Weibull distribution
d. Log normal distribution

2. The attached data (relapse_and_time) consists of the times to relapse and the times to
death following relapse of 10 bone marrow transplant patients. In the sample patients 4
and 6 were alive in relapse at the end of the study and patients 7-10 were alive, free of
relapse at the end of the study. Suppose the time to relapse had an exponential
distribution with hazard rate \lambda and the time to death in relapse had a Weibull
distribution with parameters \theta and \alpha.

a. Construct the likelihood for the relapse rate\lambda (exponential).


b. Construct the likelihood for the parameters\theta and \alpha (Weibull).
c. Suppose we were only allowed to observe a patients death time if the patient
relapsed. Construct the likelihood for \theta and \alpha based on this truncated
sample, and compare it to the results in (b).

3. Suppose that the time to death X has an exponential distribution with hazard rate \lambda
and that the right-censoring time C is exponential with hazard rate \theta. Let T = min(X,
C) and \delta = 1, if X \leq C ; 0, if X > C . Assume that X and C are independent.

a. Find P(\delta = 1)
b. Find the distribution of T .
c. Show that \delta and T are independent.
d. Let (T_1,\delta_1), . . . , (T_n, \delta_n) be a random sample from this model.
Find the maximum likelihood estimator of \lambda.
e. Compute mean and variance of \hat \lambda.

4. Suppose that the following remission durations are observed from 10 patients (n=10)
with solid tumors. Six patients relapse at 3.0, 6.5, 6.5, 10, 12, and 15 months; 1 patient is
lost to follow-up at 8.4 months; and 3 patients are still in remission at the end of the study
after 4.0, 5.7, and 10 months.

a. Calculate the product limit estimator \hat S(t).


b. Obtain (approximate) estimates of variance of PL estimator for t=6, t=10.
c. Plot the survival function.
d. Obtain the estimated median remission time.
5. Thirty melanoma patients (stages 2 to 4) were studied to compare the immune therapies
BCG (Bacillus Calmette-Guerin) and Corynebacterium parvum for their abilities to
prolong remission duration and survival time. The age, gender, disease stage, treatment
received, remission duration, and survival time are given in the attached document. All
the patients were resected before treatment began and thus had no evidence of melanoma
at the time of first treatment.

a. Compute and plot the PL estimates of the survivorship functions \hat S(t) of the
two treatment groups.
b. Compute the variance of \hat S(t) for every uncensored observation.
c. Estimate the median survival times of the two groups.
d. Examine the prognostic homogeneity of the two groups. That is, whether the
difference in survival between the two treatment groups is statistically significant.

Hint: Use non parametric test to test the difference of median of two distributions
(Mann-Whitney U-test or the Kolmogorov-Smirnov test (Marascuilo and
McSweeney, 1977) or generalized Wilcoxon test can be used)

e. Use a chi-square test to compare the two proportions; the number of male and
female patients in the two treatment groups.
f. Use nonparametric methods to compare the two survival distributions (Gehan’s
generalized Wilcoxon test) of the two treatment groups.

Submission date March 30, 2018

You might also like