http://hdl.handle.net/2099/3719 Wed, 22 May 2013 11:39:57 GMT 2013-05-22T11:39:57Z webmaster.bupc@upc.edu Universitat Politècnica de Catalunya. Servei de Biblioteques i Documentació no http://hdl.handle.net/2099/3735 Title: Likelihood for interval-censored observations from multi-state models Authors: Commenges, Daniel Abstract: We consider the mixed dicrete-continuous pattern of observation in a multi-state model; this is a classical pattern because very often clinical status is assessed at discrete visit times while time of death is observed exactly. The likelihood can easily be written heuristically for such models. However a formal proof is not easy in such observational patterns. We give a rigorous derivation of the likelihood for the illness-death model based on applying Jacod’s formula to an observed bivariate counting process. Wed, 01 Jan 2003 00:00:00 GMT http://hdl.handle.net/2099/3735 2003-01-01T00:00:00Z Commenges, Daniel no Inference, Survival Analysis, Statistics, Mathematical Bioscience Institute We consider the mixed dicrete-continuous pattern of observation in a multi-state model; this is a classical pattern because very often clinical status is assessed at discrete visit times while time of death is observed exactly. The likelihood can easily be written heuristically for such models. However a formal proof is not easy in such observational patterns. We give a rigorous derivation of the likelihood for the illness-death model based on applying Jacod’s formula to an observed bivariate counting process. http://hdl.handle.net/2099/3734 Title: Cumulative processes related to event histories Authors: Cook, Richard J.; Lawless, Jerald F.; Lee, Ker-Ai Abstract: Costs or benefits which accumulate for individuals over time are of interest in many life history processes. Familiar examples include costs of health care for persons with chronic medical conditions, the payments to insured persons during periods of disability, and quality of life which is sometimes used in the evaluation of treatments in terminally ill patients. For convenience, here we use the term costs to refer to cost or other cumulative measures. Two important scenarios are (i) where costs are associated with the occurrence of certain events, so that total cost accumulates as a step function, and (ii) where individuals may move between various states over time, with cost accumulating at a constant rate determined by the state occupied. In both cases, there is frequently a random variable T that represents the duration of the process generating the costs. Here we consider estimation of the mean cumulative cost over a period of interest using methods based upon marginal features of the cost process and intensity based models. Robustness to adaptive censoring is discussed in the context of the multi-state methods. Data from a quality of life study of breast cancer patients are used to illustrate the methods. Wed, 01 Jan 2003 00:00:00 GMT http://hdl.handle.net/2099/3734 2003-01-01T00:00:00Z Cook, Richard J.; Lawless, Jerald F.; Lee, Ker-Ai no Survival Analysis, Statistics Costs or benefits which accumulate for individuals over time are of interest in many life history processes. Familiar examples include costs of health care for persons with chronic medical conditions, the payments to insured persons during periods of disability, and quality of life which is sometimes used in the evaluation of treatments in terminally ill patients. For convenience, here we use the term costs to refer to cost or other cumulative measures. Two important scenarios are (i) where costs are associated with the occurrence of certain events, so that total cost accumulates as a step function, and (ii) where individuals may move between various states over time, with cost accumulating at a constant rate determined by the state occupied. In both cases, there is frequently a random variable T that represents the duration of the process generating the costs. Here we consider estimation of the mean cumulative cost over a period of interest using methods based upon marginal features of the cost process and intensity based models. Robustness to adaptive censoring is discussed in the context of the multi-state methods. Data from a quality of life study of breast cancer patients are used to illustrate the methods. http://hdl.handle.net/2099/3733 Title: A sensitivity analysis for causal parameters in structural proportional hazards models Authors: Goetghebeur, E.; Loeys, T. Abstract: Deviations from assigned treatment occur often in clinical trials. In such a setting, the traditional intent-to-treat analysis does not measure biological efficacy but rather programmatic effectiveness. For all-or-nothing compliance situation, Loeys and Goetghebeur (2003) recently proposed a Structural Proportional Hazards method. It allows for causal estimation in the complier subpopulation provided the exclusion restriction holds: randomization per se has no effect unless exposure has changed. This assumption is typically made with structural models for noncompliance but questioned when the trial is not blinded. In this paper we extend the structural PH model to allow for an effect of randomization per se. This enables analyzing sensitivity of conclusions to deviations from the exclusion estriction. In a colo-rectal cancer trial we find the causal estimator of the effect of an arterial device implantation to be remarkably insensitive to such deviations. Wed, 01 Jan 2003 00:00:00 GMT http://hdl.handle.net/2099/3733 2003-01-01T00:00:00Z Goetghebeur, E.; Loeys, T. no Survival Analysis, Statistics, Mathematical Bioscience Institute Deviations from assigned treatment occur often in clinical trials. In such a setting, the traditional intent-to-treat analysis does not measure biological efficacy but rather programmatic effectiveness. For all-or-nothing compliance situation, Loeys and Goetghebeur (2003) recently proposed a Structural Proportional Hazards method. It allows for causal estimation in the complier subpopulation provided the exclusion restriction holds: randomization per se has no effect unless exposure has changed. This assumption is typically made with structural models for noncompliance but questioned when the trial is not blinded. In this paper we extend the structural PH model to allow for an effect of randomization per se. This enables analyzing sensitivity of conclusions to deviations from the exclusion estriction. In a colo-rectal cancer trial we find the causal estimator of the effect of an arterial device implantation to be remarkably insensitive to such deviations. http://hdl.handle.net/2099/3732 Title: Survival analysis with coarsely observed covariates Authors: Nielsen, Søren Feodor Abstract: In this paper we consider analysis of survival data with incomplete covariate information. We model the incomplete covariates as a random coarsening of the complete covariate, and an overview of the theory of coarsening at random is given. Various ways of estimating the parameters of the model for the survival data given the covariates are discussed and compared. Wed, 01 Jan 2003 00:00:00 GMT http://hdl.handle.net/2099/3732 2003-01-01T00:00:00Z Nielsen, Søren Feodor no Survival Analysis In this paper we consider analysis of survival data with incomplete covariate information. We model the incomplete covariates as a random coarsening of the complete covariate, and an overview of the theory of coarsening at random is given. Various ways of estimating the parameters of the model for the survival data given the covariates are discussed and compared. http://hdl.handle.net/2099/3731 Title: Aspects of the analysis of multivariative failure time data Authors: Prentice, Ross L.; Kalbfleisch, J. D. Abstract: Multivariate failure time data arise in various forms including recurrent event data when individuals are followed to observe the sequence of occurrences of a certain type of event; correlated failure time when an individual is followed for the occurrence of two or more types of events for which the individual is simultaneously at risk, or when distinct individuals have dependent event times; or more complicated multistate processes when individuals may move among a number of discrete states over the course of a follow-up study and the states and associated sojourn times are recorded. Here we provide a critical review of statistical models and data analysis methods for the analysis of recurrent event data and correlated failure time data. This review suggests a valuable role for partially marginalized intensity models for the analysis of recurrent event data, and points to the usefulness of marginal hazard rate models and nonparametric estimates of pairwise dependencies for the analysis of correlated failure times. Areas in need of further methodology development are indicated. Wed, 01 Jan 2003 00:00:00 GMT http://hdl.handle.net/2099/3731 2003-01-01T00:00:00Z Prentice, Ross L.; Kalbfleisch, J. D. no Multivariate analysis, Survival Analysis, Mathematical Bioscience Institute Multivariate failure time data arise in various forms including recurrent event data when individuals are followed to observe the sequence of occurrences of a certain type of event; correlated failure time when an individual is followed for the occurrence of two or more types of events for which the individual is simultaneously at risk, or when distinct individuals have dependent event times; or more complicated multistate processes when individuals may move among a number of discrete states over the course of a follow-up study and the states and associated sojourn times are recorded. Here we provide a critical review of statistical models and data analysis methods for the analysis of recurrent event data and correlated failure time data. This review suggests a valuable role for partially marginalized intensity models for the analysis of recurrent event data, and points to the usefulness of marginal hazard rate models and nonparametric estimates of pairwise dependencies for the analysis of correlated failure times. Areas in need of further methodology development are indicated. http://hdl.handle.net/2099/3730 Title: Indirect inference for survival data Authors: Turnbull, Bruce W.; Jiang, Wenxin Abstract: In this paper we describe the so-called “indirect” method of inference, originally developed from the econometric literature, and apply it to survival analyses of two data sets with repeated events. This method is often more convenient computationally than maximum likelihood estimation when handling such model complexities as random effects and measurement error, for example; and it can also serve as a basis for robust inference with less stringent assumptions on the data generating mechanism. The first data set concerns recurrence times of mammary tumors in rats and is modeled using a Poisson process model with covariates and frailties. The second data set involves times of recurrences of skin tumors in individual patients in a clinical trial. The methodology is applied in both parametric and semi-parametric regression analyses to accommodate random effects and covariate measurement error. Wed, 01 Jan 2003 00:00:00 GMT http://hdl.handle.net/2099/3730 2003-01-01T00:00:00Z Turnbull, Bruce W.; Jiang, Wenxin no Survival Analysis, Statistics In this paper we describe the so-called “indirect” method of inference, originally developed from the econometric literature, and apply it to survival analyses of two data sets with repeated events. This method is often more convenient computationally than maximum likelihood estimation when handling such model complexities as random effects and measurement error, for example; and it can also serve as a basis for robust inference with less stringent assumptions on the data generating mechanism. The first data set concerns recurrence times of mammary tumors in rats and is modeled using a Poisson process model with covariates and frailties. The second data set involves times of recurrences of skin tumors in individual patients in a clinical trial. The methodology is applied in both parametric and semi-parametric regression analyses to accommodate random effects and covariate measurement error. http://hdl.handle.net/2099/3729 Title: Optimization of touristic distribution netwoorks using genetic algorithms Authors: Medina, Josep R.; Yepes, Víctor Abstract: The eight basic elements to design genetic algorithms (GA) are described and applied to solve a low demand distribution problem of passengers for a hub airport in Alicante and 30 touristic destinations in Northern Africa and Western Europe. The flexibility of GA and the possibility of creating mutually beneficial feed-back processes with human intelligence to solve complex problems as well as the difficulties in detecting erroneous codes embedded in the software are described. A new three-parent edge mapped recombination operator is used to solve the capacitated vehicle routing problem required for estimating associated costs with touristic distribution networks of low demand. GA proved to be very flexible especially in changing business environments and to solve decision-making problems involving ambiguous and sometimes contradictory constraints. Wed, 01 Jan 2003 00:00:00 GMT http://hdl.handle.net/2099/3729 2003-01-01T00:00:00Z Medina, Josep R.; Yepes, Víctor no Operations research The eight basic elements to design genetic algorithms (GA) are described and applied to solve a low demand distribution problem of passengers for a hub airport in Alicante and 30 touristic destinations in Northern Africa and Western Europe. The flexibility of GA and the possibility of creating mutually beneficial feed-back processes with human intelligence to solve complex problems as well as the difficulties in detecting erroneous codes embedded in the software are described. A new three-parent edge mapped recombination operator is used to solve the capacitated vehicle routing problem required for estimating associated costs with touristic distribution networks of low demand. GA proved to be very flexible especially in changing business environments and to solve decision-making problems involving ambiguous and sometimes contradictory constraints. http://hdl.handle.net/2099/3728 Title: An empirical evaluation of five small area estimators Authors: Costa, Àlex; Satorra, A.; Ventura, Eva Abstract: This paper compares five small area estimators. We use Monte Carlo simulation in the context of both artificial and real populations. In addition to the direct and indirect estimators, we consider the optimal composite estimator with population weights, and two composite estimators with estimated weights: one that assumes homogeneity of within area variance and squared bias and one that uses area-specific estimates of variance and squared bias. In the study with real population, we found that among the feasible estimators, the best choice is the one that uses area-specific estimates of variance and squared bias. Wed, 01 Jan 2003 00:00:00 GMT http://hdl.handle.net/2099/3728 2003-01-01T00:00:00Z Costa, Àlex; Satorra, A.; Ventura, Eva no Inference, Multivariate analysis This paper compares five small area estimators. We use Monte Carlo simulation in the context of both artificial and real populations. In addition to the direct and indirect estimators, we consider the optimal composite estimator with population weights, and two composite estimators with estimated weights: one that assumes homogeneity of within area variance and squared bias and one that uses area-specific estimates of variance and squared bias. In the study with real population, we found that among the feasible estimators, the best choice is the one that uses area-specific estimates of variance and squared bias.