Survival Data Analysis
BIOS 7210 / STAT 7570
Fall 2015
Note: There is a newer version of this course (Fall 2019) available here.
Announcements:
9-10: Assignment 3 is now due September 22, not September 17.
8-18: Welcome to the website for Survival Data Analysis. On the menu above are links to the syllabus, lecture notes, and assignments. There are also some data sets available in a tab-delimited format (we will be adding data sets to this page over the course of the semester). We will discuss these data in class and on homework assignments as the semester goes on.
Corrections:
11-10 – Lecture 10-15, slides 8-10: I didn’t notice that the 10-15 sign error correction below also propagated to the next few slides. Slides 8-10 have now been fixed as well, with several η terms replaced by -η terms. All of these plots and figures should be correct now. Again, very sorrt for the sign errors on these AFT notes. Thank you very much to Can for pointing these errors out.
10-28 – Lecture 10-27, slides 9 and 11: The notation concerning the “expected value” of covariate vector has been changed/clarified. The original notation was both problematic/unclear and didn’t give correct expressions for the score and Hessian.
10-27 – Lecture 10-15, slide 15: In the second equation, d - wTa should be d + wTa. This has been corrected now. Thank you very much to Brandon for pointing this out.
10-15 – Lecture 10-15, slide 7: In the equation at the top, η should be -η. Correspondingly, the red/blue lines should be switched. Both mistakes have been corrected now. These are pretty important corrections, as they completely reverse the interpretation – my apologies for the mistake.
10-06 – Lecture 9-29, slide 6: In example two, the equation should have a negative sign. This has been corrected now. Thank you to Daniel for pointing this out.
10-06 – Lecture 8-27, slide 13: Two issues with this equation! (1) I’ve overloaded j in this formula, and (2) the inequality in the product should be strict. The formula is corrected now. Thank you very much to Bo for pointing this out.
10-01 – Lecture 9-10, slides 28-29: < should be ≤ (the cumulative hazard in the discrete case should go up to and include time t). Thank you to Ryan for pointing this out.
9-27 – Lecture 9-22, slide 21: e-λt* should be eλt*. This also applies to Assignment 4, Problem 2. Both have been fixed online now (thank you to Bo for pointing this out).
9-10 – Lecture 9-8, Slide 19: The groups are labeled “CSP+MTX” and “CSP”; that should be “MTX+CSP” and “MTX” (thank you to Anthony for pointing this out!). This is fixed now.
9-8 – Lecture 9-3, Slide 18: The hazard terms in the product should be raised to the power di (i.e., only subjects observed to fail should contribute a hazard term to the likelihood). The same issue applies to the Assignment 2, Problem 3, second equation. Both typos have now been fixed online.