Cox s proportional hazards model and Cox s partial likelihood

Size: px
Start display at page:

Download "Cox s proportional hazards model and Cox s partial likelihood"

Transcription

1 Cox s proportional hazards model and Cox s partial likelihood Rasmus Waagepetersen October 12, / 27

2 Non-parametric vs. parametric Suppose we want to estimate unknown function, e.g. survival function. Approaches: Non-parametric using Kaplan-Meier. Advantage: no assumption regarding type of distribution. Disadvantage: requires iid observations. Parametric model. Advantage: we only need to estimate a few parameters that completely characterize distribution (e.g. exponential or Weibull) - gives low variance of estimates. Can be extended to non-iid observations using regression on covariates. Disadvantage: assumed model class may be (or always is) incorrect leading to model error or in other words, bias. Possible to combine the best of two approaches? 2 / 27

3 Semi-parametric approach - Cox s proportional hazards model Sir David Cox in a ground-breaking paper ( Regression models and life tables, 1972) suggested the following model for the hazard function given covariates z R p : h(t; z) = h 0 (t) exp(z T β), β R p. Here h 0 ( ) completely unspecified function except that it must be non-negative. Thus model combines great flexibility via non-parametric h 0 ( ) with the possibility of introducing covariate effects via exponential term exp(z T β) This model has become standard in medical statistics. 3 / 27

4 Some properties Cumulative hazard: H(t; z) = exp(z T β) Survival function t 0 h 0 (u)du = exp(z T β)h 0 (t) S(t; z) = S 0 (t) exp(zt β) S 0 (t) = exp( H 0 (t)) Proportional hazards: h(t; z) h(t; z ) = exp((z z ) T β) i.e. constant hazard ratio for two different subjects - curves can not cross! - this should be checked in any application. 4 / 27

5 Estimation - partial likelihood Model useless if we can not estimate parameter β. Problem: we can not use likelihood when h 0 ( ) unspecified. Second break-through contribution of Cox: invention of partial likelikehood for estimating β. Suppose we have observations (t i, δ i ) as well as (fixed) covariates z 1,..., z n, i = 1,..., n. We assume no ties (all t i distinct) and define D {1,..., n} as D = {l δ l = 1} - i.e. the index set of death times. For any t 0 we further define the risk set R(t) = {l t l t} i.e. the index set of subjects at risk at time t. 5 / 27

6 The partial likelihood The partial likelihood is L(β) = exp(zl T β) l D k R(t l ) exp(zt k β) Cox suggested to estimate β by maximizing L(β). does not depend on h 0 does not depend on actual death times - only their order censored observations only appear in risk set (as for Kaplan-Meier) Cox s idea has proven to work very well - but why? Lots of people have tried to make sense of this partial likelihood. 6 / 27

7 Cox s intuition Consider for simplicity the case of no censoring and let t (1),..., t (n) denote the set of ordered death times. We can equivalently represent data as the set of inter-arrival times v i = t (i) t (i 1) (taking t 0 = 0) together with the information r 1, r 2,..., r n about which subject died at each time of death - i.e. r i = l if subject l was the ith subject to die. Cox then factored likelihood of (v 1,..., v n, r 1,..., r n ) as (using generic notation for densities and probabilities) f (v 1 )p(r 1 v 1 )f (v 2 v 1, r 1 )p(r 2 v 1, v 2, r 1 ) f (v n v 1,..., v n 1, r 1,..., r n 1 )p(r n v 1,..., v n, r 1,..., r n 1 ) 7 / 27

8 Cox argued that terms f (v i...) could not contribute with information regarding β since the interarrival times can be fitted arbitrary well when h 0 is unrestricted - we can essentially just choose h 0 to consist of spikes at each death time. Thus estimation of β should be based on remaining factors L(β) = n p(r i H i ) i=1 where H i = {v 1,..., v i, r 1,..., r i 1 } history/previous observations. Here p(r i H i ) is the probability that subject r i is the ith person to die given the previous observations. 8 / 27

9 More precisely, let R i denote the random index of the ith subject that dies (R i = l means that T l is the ith smallest death time, i.e. T Ri = T (i) ). Assume that p(l H i ) only depends on H i through the knowledge that the ith death happens at time t (i) and that R(t (i) ) are the ones at risk at time t (i). Thus p(l H i ) = P(R i = l T Ri [t (i), t (i) + dt[, R(t (i) ) = A) This is the probability that l is the ith person to die given that the ith death happens at time t (i) and that the persons in A are at risk at time t (i) (thus probability is zero if l A) 9 / 27

10 We now express the conditional probability in terms of the hazard function: P(R i = l, T Ri [t (i), t (i) + dt[ R(t (i) ) = A) =P(T l [t (i), t (i) + dt[, T k > T l, k A R(t (i) ) = A) = h 0 (t (i) ) exp(zl T β)dt (1 h 0 (t (i) ) exp(zk T β)dt) k A\{l} Note = is because we actually replace T k > T l by T k > t (i) + dt. This does not really matter since dt infinitesimal. 10 / 27

11 Finally, P(R i = l T Ri [t (i), t (i) + dt[, R(t (i) ) = A) = P(R i = l, T Ri [t (i), t (i) + dt[ R(t (i) ) = A) P(T Ri [t (i), t (i) + dt[ R(t (i) ) = A) P(R i = l, T Ri [t (i), t (i) + dt[ R(t (i) = A)) = j R(t (i) ) P(R i = j, T Ri [t (i), t (i) + dt[ R(t (i) ) = A) h 0 (t (i) ) exp(zl T β)dt k R(t (i) )\{l} = (1 h 0(t (i) ) exp(zk T β)dt) j R(t (i) ) h 0(t (i) ) exp(zj T β)dt k R(t (i) )\{j} (1 h 0(t (i) ) exp(zk Tβ)dt) exp(zl T β) = k R(t (i) ) exp(zt k β) Note: last = follows after cancelling h 0 (t (i) )dt and noting that (1 h 0 (t (i) ) exp(zk T β)dt) tends to one when dt tends to zero. NB: denominator is hazard for minimum of T k, k R(t (i) ) (exercise 18) 11 / 27

12 Conditional likelihood for matched case-control study Cox s idea very closely related to conditional likelihood for matched case-control studies. Let X denote a binary random variable (e.g. sick/healthy) for an individual in a population. We want to study the impact of a covariate z on X. Assume that the population can be divided into homogeneous groups (strata) so that probability of being ill is given by a logistic regression P(X = 1) = p i (z) = exp(α i + βz) 1 + exp(α i + βz) for an individual in the ith strata and with the covariate z. 12 / 27

13 Suppose X 1 = 1 with covariate z 1 is observed for a sick person in the ith stratum. In a matched case-control study this observation is paired with an observation X 2 = 0 with covariate z 2 for a randomly selected healthy person in the same stratum. The conditional likelihood is now based on the conditional probabilities P(X 1 = 1 X 1 = 1, X 2 = 0 or X 1 = 0, X 2 = 1) = This reduces to p i (z 1 )(1 p i (z 2 )) p i (z 1 )(1 p i (z 2 )) + (1 p i (z 1 ))p i (z 2 ) exp(βz 1 ) exp(βz 1 ) + exp(βz 2 ) which is free of the strata specific intercept α i. Note α i is a nuisance parameter when we are just interested in β. 13 / 27

14 Invariance argument Again consider the case of no censoring. Kalbfleisch and Prentice noticed that if one applies a strictly monotone differentiable function g to the survival times T 1,..., T n then T i = g(t i ) again follow a proportional hazards model with a completely unspecified hazard function h 0 (exercise 17). Hence estimation problem for β the same regardless of whether we consider T i s or T i s. They thus concluded that only the ordering (ranks) of the survival times and not the magnitudes of the survival times could matter for inference on β. One can verify that for the ranks R i, P(R 1 = r 1,..., R n = r n ) = P(T r1 < T r2 < < T rn ) is precisely Cox s partial likelihood. 14 / 27

15 Profile likelihood Cox s partial likelihood can also be derived as a profile likelihood. Consider likelihood (assuming no ties) n [h 0 (t i )dt exp(zl T β)] δ i exp[ exp(zi T β) i=1 ti 0 h 0 (u)du]. Let s try to maximize wrt h 0. First, we need h 0 (t l ) > 0 for l D. At the same time we should take h 0 (u) = 0 between death times. So we let h 0 (t)dt = α l in very small intervals around death times, [t l, t l + dt[, l D, and zero elsewhere. Note likelihood does not inform about h 0 (t) for t larger than max i t i. 15 / 27

16 Then likelihood becomes ( ) L(α, β) = α l exp[zl T β] exp( l D ( ) = α l exp[zl T β] exp( l D l D α l n exp(zi T β) α l ) l D:t l t i i=1 i R(t l ) Taking log and differentiating wrt α l we obtain log L(α, β) = 1 α l α l j R(t l ) Setting equal to zero and solving wrt α l gives exp(z T i β)) exp(z T j β) ˆα l (β) = 1 j R(t l ) exp(zt j β) 16 / 27

17 Plugging in ˆα l (β) for α l we finally obtain profile likelihood: ( ) exp(zl T β) L p (β) = L(ˆα, β) = exp( D ) j R(t l ) exp(zt j β) l D which is Cox s partial likelihood. As a byproduct we obtain the Breslow estimate of H 0 : Ĥ 0 (t) = 1 l D: j R(t l ) exp(zt j β) t l t where we replace β by partial likelihood estimate ˆβ. This reduces to Nelson-Aalen estimator if β = 0. Note Ĥ 0 (t) is discontinuous in contrast to H 0 (t) = t 0 h 0(u)du. Ĥ 0 (t) limiting case of H 0 with mass increasingly concentrated around death times. 17 / 27

18 Estimating function point of view All previous derivations more or less heuristic. However, not crucial to understand Cox s partial likelihood as a likelihood or as derived from a likelihood. Just consider properties of associated estimating function. Score of partial likelihood is an estimating function which (see next slide) is unbiased (each term mean zero) sum of uncorrelated terms (gives CLT) - general theory for estimating functions suggests that partial likelihood estimates asymptotically consistent and normal. 18 / 27

19 Score function is sum of terms u(β) = d log L(β) dβ u i (β) = z Ri E[z Ri T Ri [t (i), t (i) + dt[, R(t (i) )]. Each term has mean zero: E[u i (β)] = E[E[u i (β) H i ]] = 0 Moreover, terms are uncorrelated. For i < j: E[u i (β)u j (β)] = E[u i (β)e[u j (β) H j ]] = 0 Thus good reason to believe that CLT works for score function. 19 / 27

20 Asymptotic properties of estimates and tests The observed information for the partial likelihood is j(β) = d dβ T u(β) = Var[z Ri T Ri [t (i), t (i) + dt[, R(t (i) )] = i D Var[u i (β) H i ] Information i(β) = Ej(β) = Var(u(β)) i D In analogy with usual asymptotic results we obtain for large n, ( ˆβ β) N(0, i(β) 1 ) In practice we estimate i(β) by j( ˆβ). This can be used for constructing confidence intervals in the usual way. Moreover, we can construct Wald tests, score-tests and likelihood-ratio tests in the usual way. 20 / 27

21 Asymptotic distribution - sketch Let β denote true value of regression parameter. First order (multivariate) Taylor: u(β ) = u( ˆβ) + where β β ˆβ β. Thus d dβ T u(β) β= β (β ˆβ) u(β ) = j( β)( ˆβ β ) i(β ) 1/2 u(β ) = i(β ) 1/2 j( β)( ˆβ β ) 21 / 27

22 Assume now as n tends to infinity, i(β ) 1/2 u(β ) N(0, I ) (CLT) (convergence in distribution) and i(β ) 1/2 j( β)i(β ) 1/2 I (convergence in probability). Since i(β ) 1/2 ( ˆβ β ) = (i(β ) 1/2 j( β)i(β ) 1/2 ) 1 i(β ) 1/2 u(β ) it follows by the assumptions above that i(β ) 1/2 ( ˆβ β ) N(0, I ) in distribution. 22 / 27

23 Consider H 0 : β = β 0. Several possibilities under H 0 : (Wald) j(β 0 ) 1/2 ( ˆβ β 0 ) N(0, I ) (Score test) j(β 0 ) 1/2 u(β 0 ) N(0, I ) ( likelihood ratio) 2 log(l(β 0 )/L( ˆβ)) χ 2 (p 1) See KM 8.3 and 8.5 for further details. NB: in the case of z i {0, 1} (two-group scenario), score-test for H 0 : β = 0 is equivalent with log-rank test (exercise 19). 23 / 27

24 Data with ties Suppose we have tied death times t 11 = t 12 = = t 1d 1 < t 21 = = t 2d 2 < < t r1 = = t rd r I.e. r distinct death times with d l deaths at the l distinct time. Let z lj be the covariate for the individual with death time t lj and let z l = d l j=1 z lj. Suppose we knew t l1 < t l2 < < t ld l, l = 1,..., r and let B l(j 1) consist of individuals who die at times t l1,..., t l(j 1). Then Cox s partial likelihood is r d l exp(β T zlj ) l=1 j=1 k R(tl1 )\B exp(zt l(j 1) k β) r exp(β T z = l ) dl l=1 j=1 [ k R(t l1 ) exp(zt k β) k B l(j 1) exp(zk Tβ)] 24 / 27

25 When we do not know the ordering of tl1,..., t ld l we can not compute term k B l(j 1) exp(zk Tβ). Breslow: simply ignore this sum. Resulting partial likelihood becomes r exp(β T zl ) l=1 ( k R(t l1 ) exp(zt k β))d l Efron: replace sum by j 1 times average, that is k B l(j 1) exp(z T k β) (j 1) 1 d l d l exp(β T zlk ) k=1 25 / 27

26 Cox s discrete time proportional odds model Reuse notation from actuarial estimate but introduce covariates: p k (z) = P(indiv. with covariates z dies in [u k 1, u k [ alive at time u k 1 ). Cox proposed proportional odds model: O k (z) = p k(z) 1 p k (z) = p k(0) 1 p k (0) exp(zt β) = O k (0) exp(z T β) Let D k be index set of d k individuals who die in [u k 1, u k [. Probability that precisely individuals in D k die given risk set R(u k 1 ) is p k (z l ) (1 p k (z l )) = O k (z l ) (1 p k (z l )) l D k l R(u k 1 )\D k l D k Probability that d k individuals die: O k (z l ) A R(u k 1 ): #A=d k l A l R(u k 1 ) l R(u k 1 ) (1 p k (z l )) 26 / 27

27 Discrete time partial likelihood Partial likelihood based on probabilities that individuals in D k die given d k individuals die and given R(u k 1 ). Only consider intervals with d k > 0 L(β) = k:d k >0 exp( l D k zl T β) A R(u k 1 ): exp( l A zt l β) #A=d k Note: O k (0) plays the same role as exp(α i ) in matched case control model. Different approaches to handling ties vary regarding computational complexity. On modern computers all options usually feasible. 27 / 27

STAT331. Cox s Proportional Hazards Model

STAT331. Cox s Proportional Hazards Model STAT331 Cox s Proportional Hazards Model In this unit we introduce Cox s proportional hazards (Cox s PH) model, give a heuristic development of the partial likelihood function, and discuss adaptations

More information

Cox s proportional hazards/regression model - model assessment

Cox s proportional hazards/regression model - model assessment Cox s proportional hazards/regression model - model assessment Rasmus Waagepetersen September 27, 2017 Topics: Plots based on estimated cumulative hazards Cox-Snell residuals: overall check of fit Martingale

More information

Survival Analysis Math 434 Fall 2011

Survival Analysis Math 434 Fall 2011 Survival Analysis Math 434 Fall 2011 Part IV: Chap. 8,9.2,9.3,11: Semiparametric Proportional Hazards Regression Jimin Ding Math Dept. www.math.wustl.edu/ jmding/math434/fall09/index.html Basic Model Setup

More information

STAT 6350 Analysis of Lifetime Data. Failure-time Regression Analysis

STAT 6350 Analysis of Lifetime Data. Failure-time Regression Analysis STAT 6350 Analysis of Lifetime Data Failure-time Regression Analysis Explanatory Variables for Failure Times Usually explanatory variables explain/predict why some units fail quickly and some units survive

More information

Survival Regression Models

Survival Regression Models Survival Regression Models David M. Rocke May 18, 2017 David M. Rocke Survival Regression Models May 18, 2017 1 / 32 Background on the Proportional Hazards Model The exponential distribution has constant

More information

Semiparametric Regression

Semiparametric Regression Semiparametric Regression Patrick Breheny October 22 Patrick Breheny Survival Data Analysis (BIOS 7210) 1/23 Introduction Over the past few weeks, we ve introduced a variety of regression models under

More information

In contrast, parametric techniques (fitting exponential or Weibull, for example) are more focussed, can handle general covariates, but require

In contrast, parametric techniques (fitting exponential or Weibull, for example) are more focussed, can handle general covariates, but require Chapter 5 modelling Semi parametric We have considered parametric and nonparametric techniques for comparing survival distributions between different treatment groups. Nonparametric techniques, such as

More information

Other Survival Models. (1) Non-PH models. We briefly discussed the non-proportional hazards (non-ph) model

Other Survival Models. (1) Non-PH models. We briefly discussed the non-proportional hazards (non-ph) model Other Survival Models (1) Non-PH models We briefly discussed the non-proportional hazards (non-ph) model λ(t Z) = λ 0 (t) exp{β(t) Z}, where β(t) can be estimated by: piecewise constants (recall how);

More information

UNIVERSITY OF CALIFORNIA, SAN DIEGO

UNIVERSITY OF CALIFORNIA, SAN DIEGO UNIVERSITY OF CALIFORNIA, SAN DIEGO Estimation of the primary hazard ratio in the presence of a secondary covariate with non-proportional hazards An undergraduate honors thesis submitted to the Department

More information

Part [1.0] Measures of Classification Accuracy for the Prediction of Survival Times

Part [1.0] Measures of Classification Accuracy for the Prediction of Survival Times Part [1.0] Measures of Classification Accuracy for the Prediction of Survival Times Patrick J. Heagerty PhD Department of Biostatistics University of Washington 1 Biomarkers Review: Cox Regression Model

More information

Tied survival times; estimation of survival probabilities

Tied survival times; estimation of survival probabilities Tied survival times; estimation of survival probabilities Patrick Breheny November 5 Patrick Breheny Survival Data Analysis (BIOS 7210) 1/22 Introduction Tied survival times Introduction Breslow approximation

More information

Exercises. (a) Prove that m(t) =

Exercises. (a) Prove that m(t) = Exercises 1. Lack of memory. Verify that the exponential distribution has the lack of memory property, that is, if T is exponentially distributed with parameter λ > then so is T t given that T > t for

More information

Lecture 22 Survival Analysis: An Introduction

Lecture 22 Survival Analysis: An Introduction University of Illinois Department of Economics Spring 2017 Econ 574 Roger Koenker Lecture 22 Survival Analysis: An Introduction There is considerable interest among economists in models of durations, which

More information

Statistics 262: Intermediate Biostatistics Non-parametric Survival Analysis

Statistics 262: Intermediate Biostatistics Non-parametric Survival Analysis Statistics 262: Intermediate Biostatistics Non-parametric Survival Analysis Jonathan Taylor & Kristin Cobb Statistics 262: Intermediate Biostatistics p.1/?? Overview of today s class Kaplan-Meier Curve

More information

MAS3301 / MAS8311 Biostatistics Part II: Survival

MAS3301 / MAS8311 Biostatistics Part II: Survival MAS3301 / MAS8311 Biostatistics Part II: Survival M. Farrow School of Mathematics and Statistics Newcastle University Semester 2, 2009-10 1 13 The Cox proportional hazards model 13.1 Introduction In the

More information

Lecture 6 PREDICTING SURVIVAL UNDER THE PH MODEL

Lecture 6 PREDICTING SURVIVAL UNDER THE PH MODEL Lecture 6 PREDICTING SURVIVAL UNDER THE PH MODEL The Cox PH model: λ(t Z) = λ 0 (t) exp(β Z). How do we estimate the survival probability, S z (t) = S(t Z) = P (T > t Z), for an individual with covariates

More information

FULL LIKELIHOOD INFERENCES IN THE COX MODEL

FULL LIKELIHOOD INFERENCES IN THE COX MODEL October 20, 2007 FULL LIKELIHOOD INFERENCES IN THE COX MODEL BY JIAN-JIAN REN 1 AND MAI ZHOU 2 University of Central Florida and University of Kentucky Abstract We use the empirical likelihood approach

More information

β j = coefficient of x j in the model; β = ( β1, β2,

β j = coefficient of x j in the model; β = ( β1, β2, Regression Modeling of Survival Time Data Why regression models? Groups similar except for the treatment under study use the nonparametric methods discussed earlier. Groups differ in variables (covariates)

More information

Survival Analysis. Stat 526. April 13, 2018

Survival Analysis. Stat 526. April 13, 2018 Survival Analysis Stat 526 April 13, 2018 1 Functions of Survival Time Let T be the survival time for a subject Then P [T < 0] = 0 and T is a continuous random variable The Survival function is defined

More information

Practice Exam 1. (A) (B) (C) (D) (E) You are given the following data on loss sizes:

Practice Exam 1. (A) (B) (C) (D) (E) You are given the following data on loss sizes: Practice Exam 1 1. Losses for an insurance coverage have the following cumulative distribution function: F(0) = 0 F(1,000) = 0.2 F(5,000) = 0.4 F(10,000) = 0.9 F(100,000) = 1 with linear interpolation

More information

PhD course in Advanced survival analysis. One-sample tests. Properties. Idea: (ABGK, sect. V.1.1) Counting process N(t)

PhD course in Advanced survival analysis. One-sample tests. Properties. Idea: (ABGK, sect. V.1.1) Counting process N(t) PhD course in Advanced survival analysis. (ABGK, sect. V.1.1) One-sample tests. Counting process N(t) Non-parametric hypothesis tests. Parametric models. Intensity process λ(t) = α(t)y (t) satisfying Aalen

More information

Multistate Modeling and Applications

Multistate Modeling and Applications Multistate Modeling and Applications Yang Yang Department of Statistics University of Michigan, Ann Arbor IBM Research Graduate Student Workshop: Statistics for a Smarter Planet Yang Yang (UM, Ann Arbor)

More information

log T = β T Z + ɛ Zi Z(u; β) } dn i (ue βzi ) = 0,

log T = β T Z + ɛ Zi Z(u; β) } dn i (ue βzi ) = 0, Accelerated failure time model: log T = β T Z + ɛ β estimation: solve where S n ( β) = n i=1 { Zi Z(u; β) } dn i (ue βzi ) = 0, Z(u; β) = j Z j Y j (ue βz j) j Y j (ue βz j) How do we show the asymptotics

More information

THESIS for the degree of MASTER OF SCIENCE. Modelling and Data Analysis

THESIS for the degree of MASTER OF SCIENCE. Modelling and Data Analysis PROPERTIES OF ESTIMATORS FOR RELATIVE RISKS FROM NESTED CASE-CONTROL STUDIES WITH MULTIPLE OUTCOMES (COMPETING RISKS) by NATHALIE C. STØER THESIS for the degree of MASTER OF SCIENCE Modelling and Data

More information

11 Survival Analysis and Empirical Likelihood

11 Survival Analysis and Empirical Likelihood 11 Survival Analysis and Empirical Likelihood The first paper of empirical likelihood is actually about confidence intervals with the Kaplan-Meier estimator (Thomas and Grunkmeier 1979), i.e. deals with

More information

Time-dependent covariates

Time-dependent covariates Time-dependent covariates Rasmus Waagepetersen November 5, 2018 1 / 10 Time-dependent covariates Our excursion into the realm of counting process and martingales showed that it poses no problems to introduce

More information

Lecture 12. Multivariate Survival Data Statistics Survival Analysis. Presented March 8, 2016

Lecture 12. Multivariate Survival Data Statistics Survival Analysis. Presented March 8, 2016 Statistics 255 - Survival Analysis Presented March 8, 2016 Dan Gillen Department of Statistics University of California, Irvine 12.1 Examples Clustered or correlated survival times Disease onset in family

More information

9 Estimating the Underlying Survival Distribution for a

9 Estimating the Underlying Survival Distribution for a 9 Estimating the Underlying Survival Distribution for a Proportional Hazards Model So far the focus has been on the regression parameters in the proportional hazards model. These parameters describe the

More information

Introduction to Statistical Analysis

Introduction to Statistical Analysis Introduction to Statistical Analysis Changyu Shen Richard A. and Susan F. Smith Center for Outcomes Research in Cardiology Beth Israel Deaconess Medical Center Harvard Medical School Objectives Descriptive

More information

Analysis of Time-to-Event Data: Chapter 6 - Regression diagnostics

Analysis of Time-to-Event Data: Chapter 6 - Regression diagnostics Analysis of Time-to-Event Data: Chapter 6 - Regression diagnostics Steffen Unkel Department of Medical Statistics University Medical Center Göttingen, Germany Winter term 2018/19 1/25 Residuals for the

More information

Follow this and additional works at: Part of the Applied Mathematics Commons

Follow this and additional works at:   Part of the Applied Mathematics Commons Louisiana State University LSU Digital Commons LSU Master's Theses Graduate School 2004 Cox Regression Model Lindsay Sarah Smith Louisiana State University and Agricultural and Mechanical College, lsmit51@lsu.edu

More information

Improving Efficiency of Inferences in Randomized Clinical Trials Using Auxiliary Covariates

Improving Efficiency of Inferences in Randomized Clinical Trials Using Auxiliary Covariates Improving Efficiency of Inferences in Randomized Clinical Trials Using Auxiliary Covariates Anastasios (Butch) Tsiatis Department of Statistics North Carolina State University http://www.stat.ncsu.edu/

More information

Beyond GLM and likelihood

Beyond GLM and likelihood Stat 6620: Applied Linear Models Department of Statistics Western Michigan University Statistics curriculum Core knowledge (modeling and estimation) Math stat 1 (probability, distributions, convergence

More information

MS&E 226: Small Data. Lecture 11: Maximum likelihood (v2) Ramesh Johari

MS&E 226: Small Data. Lecture 11: Maximum likelihood (v2) Ramesh Johari MS&E 226: Small Data Lecture 11: Maximum likelihood (v2) Ramesh Johari ramesh.johari@stanford.edu 1 / 18 The likelihood function 2 / 18 Estimating the parameter This lecture develops the methodology behind

More information

Power and Sample Size Calculations with the Additive Hazards Model

Power and Sample Size Calculations with the Additive Hazards Model Journal of Data Science 10(2012), 143-155 Power and Sample Size Calculations with the Additive Hazards Model Ling Chen, Chengjie Xiong, J. Philip Miller and Feng Gao Washington University School of Medicine

More information

Statistical Inference and Methods

Statistical Inference and Methods Department of Mathematics Imperial College London d.stephens@imperial.ac.uk http://stats.ma.ic.ac.uk/ das01/ 31st January 2006 Part VI Session 6: Filtering and Time to Event Data Session 6: Filtering and

More information

Definitions and examples Simple estimation and testing Regression models Goodness of fit for the Cox model. Recap of Part 1. Per Kragh Andersen

Definitions and examples Simple estimation and testing Regression models Goodness of fit for the Cox model. Recap of Part 1. Per Kragh Andersen Recap of Part 1 Per Kragh Andersen Section of Biostatistics, University of Copenhagen DSBS Course Survival Analysis in Clinical Trials January 2018 1 / 65 Overview Definitions and examples Simple estimation

More information

University of California, Berkeley

University of California, Berkeley University of California, Berkeley U.C. Berkeley Division of Biostatistics Working Paper Series Year 24 Paper 153 A Note on Empirical Likelihood Inference of Residual Life Regression Ying Qing Chen Yichuan

More information

ADVANCED STATISTICAL ANALYSIS OF EPIDEMIOLOGICAL STUDIES. Cox s regression analysis Time dependent explanatory variables

ADVANCED STATISTICAL ANALYSIS OF EPIDEMIOLOGICAL STUDIES. Cox s regression analysis Time dependent explanatory variables ADVANCED STATISTICAL ANALYSIS OF EPIDEMIOLOGICAL STUDIES Cox s regression analysis Time dependent explanatory variables Henrik Ravn Bandim Health Project, Statens Serum Institut 4 November 2011 1 / 53

More information

Survival Analysis for Case-Cohort Studies

Survival Analysis for Case-Cohort Studies Survival Analysis for ase-ohort Studies Petr Klášterecký Dept. of Probability and Mathematical Statistics, Faculty of Mathematics and Physics, harles University, Prague, zech Republic e-mail: petr.klasterecky@matfyz.cz

More information

Understanding the Cox Regression Models with Time-Change Covariates

Understanding the Cox Regression Models with Time-Change Covariates Understanding the Cox Regression Models with Time-Change Covariates Mai Zhou University of Kentucky The Cox regression model is a cornerstone of modern survival analysis and is widely used in many other

More information

Analysis of Time-to-Event Data: Chapter 4 - Parametric regression models

Analysis of Time-to-Event Data: Chapter 4 - Parametric regression models Analysis of Time-to-Event Data: Chapter 4 - Parametric regression models Steffen Unkel Department of Medical Statistics University Medical Center Göttingen, Germany Winter term 2018/19 1/25 Right censored

More information

Multi-state Models: An Overview

Multi-state Models: An Overview Multi-state Models: An Overview Andrew Titman Lancaster University 14 April 2016 Overview Introduction to multi-state modelling Examples of applications Continuously observed processes Intermittently observed

More information

ST745: Survival Analysis: Nonparametric methods

ST745: Survival Analysis: Nonparametric methods ST745: Survival Analysis: Nonparametric methods Eric B. Laber Department of Statistics, North Carolina State University February 5, 2015 The KM estimator is used ubiquitously in medical studies to estimate

More information

Survival Analysis. 732G34 Statistisk analys av komplexa data. Krzysztof Bartoszek

Survival Analysis. 732G34 Statistisk analys av komplexa data. Krzysztof Bartoszek Survival Analysis 732G34 Statistisk analys av komplexa data Krzysztof Bartoszek (krzysztof.bartoszek@liu.se) 10, 11 I 2018 Department of Computer and Information Science Linköping University Survival analysis

More information

Lecture 5 Models and methods for recurrent event data

Lecture 5 Models and methods for recurrent event data Lecture 5 Models and methods for recurrent event data Recurrent and multiple events are commonly encountered in longitudinal studies. In this chapter we consider ordered recurrent and multiple events.

More information

Survival Analysis I (CHL5209H)

Survival Analysis I (CHL5209H) Survival Analysis Dalla Lana School of Public Health University of Toronto olli.saarela@utoronto.ca January 7, 2015 31-1 Literature Clayton D & Hills M (1993): Statistical Models in Epidemiology. Not really

More information

Logistic regression model for survival time analysis using time-varying coefficients

Logistic regression model for survival time analysis using time-varying coefficients Logistic regression model for survival time analysis using time-varying coefficients Accepted in American Journal of Mathematical and Management Sciences, 2016 Kenichi SATOH ksatoh@hiroshima-u.ac.jp Research

More information

Survival Analysis I (CHL5209H)

Survival Analysis I (CHL5209H) Survival Analysis Dalla Lana School of Public Health University of Toronto olli.saarela@utoronto.ca February 3, 2015 21-1 Time matching/risk set sampling/incidence density sampling/nested design 21-2 21-3

More information

You know I m not goin diss you on the internet Cause my mama taught me better than that I m a survivor (What?) I m not goin give up (What?

You know I m not goin diss you on the internet Cause my mama taught me better than that I m a survivor (What?) I m not goin give up (What? You know I m not goin diss you on the internet Cause my mama taught me better than that I m a survivor (What?) I m not goin give up (What?) I m not goin stop (What?) I m goin work harder (What?) Sir David

More information

1 Introduction. 2 Residuals in PH model

1 Introduction. 2 Residuals in PH model Supplementary Material for Diagnostic Plotting Methods for Proportional Hazards Models With Time-dependent Covariates or Time-varying Regression Coefficients BY QIQING YU, JUNYI DONG Department of Mathematical

More information

Variable Selection in Competing Risks Using the L1-Penalized Cox Model

Variable Selection in Competing Risks Using the L1-Penalized Cox Model Virginia Commonwealth University VCU Scholars Compass Theses and Dissertations Graduate School 2008 Variable Selection in Competing Risks Using the L1-Penalized Cox Model XiangRong Kong Virginia Commonwealth

More information

Lecture 7. Proportional Hazards Model - Handling Ties and Survival Estimation Statistics Survival Analysis. Presented February 4, 2016

Lecture 7. Proportional Hazards Model - Handling Ties and Survival Estimation Statistics Survival Analysis. Presented February 4, 2016 Proportional Hazards Model - Handling Ties and Survival Estimation Statistics 255 - Survival Analysis Presented February 4, 2016 likelihood - Discrete Dan Gillen Department of Statistics University of

More information

Chapter 4 Regression Models

Chapter 4 Regression Models 23.August 2010 Chapter 4 Regression Models The target variable T denotes failure time We let x = (x (1),..., x (m) ) represent a vector of available covariates. Also called regression variables, regressors,

More information

Philosophy and Features of the mstate package

Philosophy and Features of the mstate package Introduction Mathematical theory Practice Discussion Philosophy and Features of the mstate package Liesbeth de Wreede, Hein Putter Department of Medical Statistics and Bioinformatics Leiden University

More information

STAT 331. Accelerated Failure Time Models. Previously, we have focused on multiplicative intensity models, where

STAT 331. Accelerated Failure Time Models. Previously, we have focused on multiplicative intensity models, where STAT 331 Accelerated Failure Time Models Previously, we have focused on multiplicative intensity models, where h t z) = h 0 t) g z). These can also be expressed as H t z) = H 0 t) g z) or S t z) = e Ht

More information

ST745: Survival Analysis: Cox-PH!

ST745: Survival Analysis: Cox-PH! ST745: Survival Analysis: Cox-PH! Eric B. Laber Department of Statistics, North Carolina State University April 20, 2015 Rien n est plus dangereux qu une idee, quand on n a qu une idee. (Nothing is more

More information

if n is large, Z i are weakly dependent 0-1-variables, p i = P(Z i = 1) small, and Then n approx i=1 i=1 n i=1

if n is large, Z i are weakly dependent 0-1-variables, p i = P(Z i = 1) small, and Then n approx i=1 i=1 n i=1 Count models A classical, theoretical argument for the Poisson distribution is the approximation Binom(n, p) Pois(λ) for large n and small p and λ = np. This can be extended considerably to n approx Z

More information

Lecture 7. Poisson and lifetime processes in risk analysis

Lecture 7. Poisson and lifetime processes in risk analysis Lecture 7. Poisson and lifetime processes in risk analysis Jesper Rydén Department of Mathematics, Uppsala University jesper.ryden@math.uu.se Statistical Risk Analysis Spring 2014 Example: Life times of

More information

Model Selection in Cox regression

Model Selection in Cox regression Model Selection in Cox regression Suppose we have a possibly censored survival outcome that we want to model as a function of a (possibly large) set of covariates. How do we decide which covariates to

More information

DAGStat Event History Analysis.

DAGStat Event History Analysis. DAGStat 2016 Event History Analysis Robin.Henderson@ncl.ac.uk 1 / 75 Schedule 9.00 Introduction 10.30 Break 11.00 Regression Models, Frailty and Multivariate Survival 12.30 Lunch 13.30 Time-Variation and

More information

n =10,220 observations. Smaller samples analyzed here to illustrate sample size effect.

n =10,220 observations. Smaller samples analyzed here to illustrate sample size effect. Chapter 7 Parametric Likelihood Fitting Concepts: Chapter 7 Parametric Likelihood Fitting Concepts: Objectives Show how to compute a likelihood for a parametric model using discrete data. Show how to compute

More information

Approximation of Survival Function by Taylor Series for General Partly Interval Censored Data

Approximation of Survival Function by Taylor Series for General Partly Interval Censored Data Malaysian Journal of Mathematical Sciences 11(3): 33 315 (217) MALAYSIAN JOURNAL OF MATHEMATICAL SCIENCES Journal homepage: http://einspem.upm.edu.my/journal Approximation of Survival Function by Taylor

More information

Frailty Models and Copulas: Similarities and Differences

Frailty Models and Copulas: Similarities and Differences Frailty Models and Copulas: Similarities and Differences KLARA GOETHALS, PAUL JANSSEN & LUC DUCHATEAU Department of Physiology and Biometrics, Ghent University, Belgium; Center for Statistics, Hasselt

More information

Proportional hazards regression

Proportional hazards regression Proportional hazards regression Patrick Breheny October 8 Patrick Breheny Survival Data Analysis (BIOS 7210) 1/28 Introduction The model Solving for the MLE Inference Today we will begin discussing regression

More information

Typical Survival Data Arising From a Clinical Trial. Censoring. The Survivor Function. Mathematical Definitions Introduction

Typical Survival Data Arising From a Clinical Trial. Censoring. The Survivor Function. Mathematical Definitions Introduction Outline CHL 5225H Advanced Statistical Methods for Clinical Trials: Survival Analysis Prof. Kevin E. Thorpe Defining Survival Data Mathematical Definitions Non-parametric Estimates of Survival Comparing

More information

Credit risk and survival analysis: Estimation of Conditional Cure Rate

Credit risk and survival analysis: Estimation of Conditional Cure Rate Stochastics and Financial Mathematics Master Thesis Credit risk and survival analysis: Estimation of Conditional Cure Rate Author: Just Bajželj Examination date: August 30, 2018 Supervisor: dr. A.J. van

More information

Modelling geoadditive survival data

Modelling geoadditive survival data Modelling geoadditive survival data Thomas Kneib & Ludwig Fahrmeir Department of Statistics, Ludwig-Maximilians-University Munich 1. Leukemia survival data 2. Structured hazard regression 3. Mixed model

More information

Continuous Time Survival in Latent Variable Models

Continuous Time Survival in Latent Variable Models Continuous Time Survival in Latent Variable Models Tihomir Asparouhov 1, Katherine Masyn 2, Bengt Muthen 3 Muthen & Muthen 1 University of California, Davis 2 University of California, Los Angeles 3 Abstract

More information

Reduced-rank hazard regression

Reduced-rank hazard regression Chapter 2 Reduced-rank hazard regression Abstract The Cox proportional hazards model is the most common method to analyze survival data. However, the proportional hazards assumption might not hold. The

More information

18.465, further revised November 27, 2012 Survival analysis and the Kaplan Meier estimator

18.465, further revised November 27, 2012 Survival analysis and the Kaplan Meier estimator 18.465, further revised November 27, 2012 Survival analysis and the Kaplan Meier estimator 1. Definitions Ordinarily, an unknown distribution function F is estimated by an empirical distribution function

More information

FULL LIKELIHOOD INFERENCES IN THE COX MODEL: AN EMPIRICAL LIKELIHOOD APPROACH

FULL LIKELIHOOD INFERENCES IN THE COX MODEL: AN EMPIRICAL LIKELIHOOD APPROACH FULL LIKELIHOOD INFERENCES IN THE COX MODEL: AN EMPIRICAL LIKELIHOOD APPROACH Jian-Jian Ren 1 and Mai Zhou 2 University of Central Florida and University of Kentucky Abstract: For the regression parameter

More information

Part III. Hypothesis Testing. III.1. Log-rank Test for Right-censored Failure Time Data

Part III. Hypothesis Testing. III.1. Log-rank Test for Right-censored Failure Time Data 1 Part III. Hypothesis Testing III.1. Log-rank Test for Right-censored Failure Time Data Consider a survival study consisting of n independent subjects from p different populations with survival functions

More information

4 Testing Hypotheses. 4.1 Tests in the regression setting. 4.2 Non-parametric testing of survival between groups

4 Testing Hypotheses. 4.1 Tests in the regression setting. 4.2 Non-parametric testing of survival between groups 4 Testing Hypotheses The next lectures will look at tests, some in an actuarial setting, and in the last subsection we will also consider tests applied to graduation 4 Tests in the regression setting )

More information

Classification 2: Linear discriminant analysis (continued); logistic regression

Classification 2: Linear discriminant analysis (continued); logistic regression Classification 2: Linear discriminant analysis (continued); logistic regression Ryan Tibshirani Data Mining: 36-462/36-662 April 4 2013 Optional reading: ISL 4.4, ESL 4.3; ISL 4.3, ESL 4.4 1 Reminder:

More information

Full likelihood inferences in the Cox model: an empirical likelihood approach

Full likelihood inferences in the Cox model: an empirical likelihood approach Ann Inst Stat Math 2011) 63:1005 1018 DOI 10.1007/s10463-010-0272-y Full likelihood inferences in the Cox model: an empirical likelihood approach Jian-Jian Ren Mai Zhou Received: 22 September 2008 / Revised:

More information

Multivariate Survival Analysis

Multivariate Survival Analysis Multivariate Survival Analysis Previously we have assumed that either (X i, δ i ) or (X i, δ i, Z i ), i = 1,..., n, are i.i.d.. This may not always be the case. Multivariate survival data can arise in

More information

TMA 4275 Lifetime Analysis June 2004 Solution

TMA 4275 Lifetime Analysis June 2004 Solution TMA 4275 Lifetime Analysis June 2004 Solution Problem 1 a) Observation of the outcome is censored, if the time of the outcome is not known exactly and only the last time when it was observed being intact,

More information

Missing Covariate Data in Matched Case-Control Studies

Missing Covariate Data in Matched Case-Control Studies Missing Covariate Data in Matched Case-Control Studies Department of Statistics North Carolina State University Paul Rathouz Dept. of Health Studies U. of Chicago prathouz@health.bsd.uchicago.edu with

More information

Lecture 3. Truncation, length-bias and prevalence sampling

Lecture 3. Truncation, length-bias and prevalence sampling Lecture 3. Truncation, length-bias and prevalence sampling 3.1 Prevalent sampling Statistical techniques for truncated data have been integrated into survival analysis in last two decades. Truncation in

More information

Efficient Semiparametric Estimators via Modified Profile Likelihood in Frailty & Accelerated-Failure Models

Efficient Semiparametric Estimators via Modified Profile Likelihood in Frailty & Accelerated-Failure Models NIH Talk, September 03 Efficient Semiparametric Estimators via Modified Profile Likelihood in Frailty & Accelerated-Failure Models Eric Slud, Math Dept, Univ of Maryland Ongoing joint project with Ilia

More information

Log-linearity for Cox s regression model. Thesis for the Degree Master of Science

Log-linearity for Cox s regression model. Thesis for the Degree Master of Science Log-linearity for Cox s regression model Thesis for the Degree Master of Science Zaki Amini Master s Thesis, Spring 2015 i Abstract Cox s regression model is one of the most applied methods in medical

More information

Handling Ties in the Rank Ordered Logit Model Applied in Epidemiological

Handling Ties in the Rank Ordered Logit Model Applied in Epidemiological Handling Ties in the Rank Ordered Logit Model Applied in Epidemiological Settings Angeliki Maraki Masteruppsats i matematisk statistik Master Thesis in Mathematical Statistics Masteruppsats 2016:4 Matematisk

More information

Constrained estimation for binary and survival data

Constrained estimation for binary and survival data Constrained estimation for binary and survival data Jeremy M. G. Taylor Yong Seok Park John D. Kalbfleisch Biostatistics, University of Michigan May, 2010 () Constrained estimation May, 2010 1 / 43 Outline

More information

Logistic Regression. Advanced Methods for Data Analysis (36-402/36-608) Spring 2014

Logistic Regression. Advanced Methods for Data Analysis (36-402/36-608) Spring 2014 Logistic Regression Advanced Methods for Data Analysis (36-402/36-608 Spring 204 Classification. Introduction to classification Classification, like regression, is a predictive task, but one in which the

More information

ST495: Survival Analysis: Maximum likelihood

ST495: Survival Analysis: Maximum likelihood ST495: Survival Analysis: Maximum likelihood Eric B. Laber Department of Statistics, North Carolina State University February 11, 2014 Everything is deception: seeking the minimum of illusion, keeping

More information

Logistic regression. 11 Nov Logistic regression (EPFL) Applied Statistics 11 Nov / 20

Logistic regression. 11 Nov Logistic regression (EPFL) Applied Statistics 11 Nov / 20 Logistic regression 11 Nov 2010 Logistic regression (EPFL) Applied Statistics 11 Nov 2010 1 / 20 Modeling overview Want to capture important features of the relationship between a (set of) variable(s)

More information

USING MARTINGALE RESIDUALS TO ASSESS GOODNESS-OF-FIT FOR SAMPLED RISK SET DATA

USING MARTINGALE RESIDUALS TO ASSESS GOODNESS-OF-FIT FOR SAMPLED RISK SET DATA USING MARTINGALE RESIDUALS TO ASSESS GOODNESS-OF-FIT FOR SAMPLED RISK SET DATA Ørnulf Borgan Bryan Langholz Abstract Standard use of Cox s regression model and other relative risk regression models for

More information

Estimating functions for inhomogeneous spatial point processes with incomplete covariate data

Estimating functions for inhomogeneous spatial point processes with incomplete covariate data Estimating functions for inhomogeneous spatial point processes with incomplete covariate data Rasmus aagepetersen Department of Mathematics Aalborg University Denmark August 15, 2007 1 / 23 Data (Barro

More information

Extensions of Cox Model for Non-Proportional Hazards Purpose

Extensions of Cox Model for Non-Proportional Hazards Purpose PhUSE Annual Conference 2013 Paper SP07 Extensions of Cox Model for Non-Proportional Hazards Purpose Author: Jadwiga Borucka PAREXEL, Warsaw, Poland Brussels 13 th - 16 th October 2013 Presentation Plan

More information

Validation. Terry M Therneau. Dec 2015

Validation. Terry M Therneau. Dec 2015 Validation Terry M Therneau Dec 205 Introduction When I use a word, Humpty Dumpty said, in rather a scornful tone, it means just what I choose it to mean - neither more nor less. The question is, said

More information

Censoring mechanisms

Censoring mechanisms Censoring mechanisms Patrick Breheny September 3 Patrick Breheny Survival Data Analysis (BIOS 7210) 1/23 Fixed vs. random censoring In the previous lecture, we derived the contribution to the likelihood

More information

Chapter 17. Failure-Time Regression Analysis. William Q. Meeker and Luis A. Escobar Iowa State University and Louisiana State University

Chapter 17. Failure-Time Regression Analysis. William Q. Meeker and Luis A. Escobar Iowa State University and Louisiana State University Chapter 17 Failure-Time Regression Analysis William Q. Meeker and Luis A. Escobar Iowa State University and Louisiana State University Copyright 1998-2008 W. Q. Meeker and L. A. Escobar. Based on the authors

More information

A COMPARISON OF POISSON AND BINOMIAL EMPIRICAL LIKELIHOOD Mai Zhou and Hui Fang University of Kentucky

A COMPARISON OF POISSON AND BINOMIAL EMPIRICAL LIKELIHOOD Mai Zhou and Hui Fang University of Kentucky A COMPARISON OF POISSON AND BINOMIAL EMPIRICAL LIKELIHOOD Mai Zhou and Hui Fang University of Kentucky Empirical likelihood with right censored data were studied by Thomas and Grunkmier (1975), Li (1995),

More information

Faculty of Health Sciences. Cox regression. Torben Martinussen. Department of Biostatistics University of Copenhagen. 20. september 2012 Slide 1/51

Faculty of Health Sciences. Cox regression. Torben Martinussen. Department of Biostatistics University of Copenhagen. 20. september 2012 Slide 1/51 Faculty of Health Sciences Cox regression Torben Martinussen Department of Biostatistics University of Copenhagen 2. september 212 Slide 1/51 Survival analysis Standard setup for right-censored survival

More information

Machine Learning Linear Classification. Prof. Matteo Matteucci

Machine Learning Linear Classification. Prof. Matteo Matteucci Machine Learning Linear Classification Prof. Matteo Matteucci Recall from the first lecture 2 X R p Regression Y R Continuous Output X R p Y {Ω 0, Ω 1,, Ω K } Classification Discrete Output X R p Y (X)

More information

Asymptotic equivalence of paired Hotelling test and conditional logistic regression

Asymptotic equivalence of paired Hotelling test and conditional logistic regression Asymptotic equivalence of paired Hotelling test and conditional logistic regression Félix Balazard 1,2 arxiv:1610.06774v1 [math.st] 21 Oct 2016 Abstract 1 Sorbonne Universités, UPMC Univ Paris 06, CNRS

More information

Lecture 5: LDA and Logistic Regression

Lecture 5: LDA and Logistic Regression Lecture 5: and Logistic Regression Hao Helen Zhang Hao Helen Zhang Lecture 5: and Logistic Regression 1 / 39 Outline Linear Classification Methods Two Popular Linear Models for Classification Linear Discriminant

More information

Frailty Modeling for Spatially Correlated Survival Data, with Application to Infant Mortality in Minnesota By: Sudipto Banerjee, Mela. P.

Frailty Modeling for Spatially Correlated Survival Data, with Application to Infant Mortality in Minnesota By: Sudipto Banerjee, Mela. P. Frailty Modeling for Spatially Correlated Survival Data, with Application to Infant Mortality in Minnesota By: Sudipto Banerjee, Melanie M. Wall, Bradley P. Carlin November 24, 2014 Outlines of the talk

More information

Empirical Likelihood in Survival Analysis

Empirical Likelihood in Survival Analysis Empirical Likelihood in Survival Analysis Gang Li 1, Runze Li 2, and Mai Zhou 3 1 Department of Biostatistics, University of California, Los Angeles, CA 90095 vli@ucla.edu 2 Department of Statistics, The

More information