A semi-parametric generalization of the Cox proportional hazards regression model: Inference and applications

Division of Population Science, Fox Chase Cancer Center, Philadelphia, PA 19111.
Computational Statistics & Data Analysis (Impact Factor: 1.15). 01/2011; 55(1):667-676. DOI: 10.1016/j.csda.2010.06.010
Source: RePEc

ABSTRACT The assumption of proportional hazards (PH) fundamental to the Cox PH model sometimes may not hold in practice. In this paper, we propose a generalization of the Cox PH model in terms of the cumulative hazard function taking a form similar to the Cox PH model, with the extension that the baseline cumulative hazard function is raised to a power function. Our model allows for interaction between covariates and the baseline hazard and it also includes, for the two sample problem, the case of two Weibull distributions and two extreme value distributions differing in both scale and shape parameters. The partial likelihood approach can not be applied here to estimate the model parameters. We use the full likelihood approach via a cubic B-spline approximation for the baseline hazard to estimate the model parameters. A semi-automatic procedure for knot selection based on Akaike's information criterion is developed. We illustrate the applicability of our approach using real-life data.

1 Follower
  • [Show abstract] [Hide abstract]
    ABSTRACT: Multifactor software reliability modeling with software test metrics data is well known to be useful for predicting the software reliability with higher accuracy, because it utilizes not only software fault count data but also software testing metrics data observed in the development process. In this paper we generalize the existing Cox proportional hazards regression-based software reliability model by introducing more generalized hazards representation, and improve the goodness-of-fit and predictive performances. In numerical examples with real software development project data, we show that our generalized model can significantly outperform several logistic regression-based models as well as the existing Cox proportional hazards regression-based model.
    Proceedings of the 2013 IEEE 19th Pacific Rim International Symposium on Dependable Computing; 12/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: Risk factor models in clinical epidemiology are important for identifying individuals at high risk of poor health outcomes and for guiding intervention strategies to reduce risk. Such models are often based on generalised linear models (GLM) with a multiplicative risk, rate or odds assumption. However, in practice some risk factors may act additively, in which case the use of a multiplicative model will lead to spurious interactions among risk factors. Computational methodology is developed for fitting non-GLM Poisson regression models that have an additive component with multiplicative stratification. These stratified additive Poisson models, which can also be applied to binomial data, provide an additive–multiplicative framework that allows greater flexibility than multiplicative models. Non-negativity constraints and high dimensionality are dealt with using an Expectation-Conditional-Maximisation (ECM) algorithm that oscillates between the multiplicative and additive components of the model. As well as providing highly stable convergence properties in a potentially unstable setting, the method allows flexible modelling features such as unspecified isotonic regression functions. The methodology is illustrated with an analysis of heart attack mortality in a large clinical trial, and it is found that the combination of additive and multiplicative components allows a more parsimonious risk factor model by removing the need for interaction terms. R code to implement the method is provided as supplementary material.
    Computational Statistics & Data Analysis 05/2012; 56(5). DOI:10.1016/j.csda.2011.08.002 · 1.15 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: To determine the expression patterns of NF-κB regulators and target genes in clear cell renal cell carcinoma (ccRCC), their correlation with von Hippel Lindau (VHL) mutational status, and their association with survival outcomes. Meta-analyses were carried out on published ccRCC gene expression datasets by RankProd, a non-parametric statistical method. DEGs with a False Discovery Rate of < 0.05 by this method were considered significant, and intersected with a curated list of NF-κB regulators and targets to determine the nature and extent of NF-κB deregulation in ccRCC. A highly-disproportionate fraction (~40%; p < 0.001) of NF-κB regulators and target genes were found to be up-regulated in ccRCC, indicative of elevated NF-κB activity in this cancer. A subset of these genes, comprising a key NF-κB regulator (IKBKB) and established mediators of the NF-κB cell-survival and pro-inflammatory responses (MMP9, PSMB9, and SOD2), correlated with higher relative risk, poorer prognosis, and reduced overall patient survival. Surprisingly, levels of several interferon regulatory factors (IRFs) and interferon target genes were also elevated in ccRCC, indicating that an 'interferon signature' may represent a novel feature of this disease. Loss of VHL gene expression correlated strongly with the appearance of NF-κB- and interferon gene signatures in both familial and sporadic cases of ccRCC. As NF-κB controls expression of key interferon signaling nodes, our results suggest a causal link between VHL loss, elevated NF-κB activity, and the appearance of an interferon signature during ccRCC tumorigenesis. These findings identify NF-κB and interferon signatures as clinical features of ccRCC, provide strong rationale for the incorporation of NF-κB inhibitors and/or and the exploitation of interferon signaling in the treatment of ccRCC, and supply new NF-κB targets for potential therapeutic intervention in this currently-incurable malignancy.
    PLoS ONE 10/2013; 8(10):e76746. DOI:10.1371/journal.pone.0076746 · 3.53 Impact Factor


Available from