Search

In this paper, we introduce a Bayesian approach for segmented Weibull distributions which could be a good alternative to analyze medical survival data in the presence of censored observations and covariates. With the obtained Bayesian estimated change-points we could get an excellent fit of the proposed model to any data sets. With the proposed methodology, it is also possible to identify survival times intervals where a covariate could have significantly different effects when compared to other lifetime intervals, an important point under a clinical view. The obtained Bayesian estimates are obtained using standard Markov Chain Monte Carlo methods. Some examples with real data sets illustrate the proposed methodology and its potential clinical value. Resumen En este artículo introducimos un nuevo modelo Bayesiano para distribuciones Weibull segmentadas, que puede ser una buena alternativa en el análisis de datos aplicados a la investigación en salud, con la presencia de censuras y covariables. Con este método basado en la estimación de puntos de cambio, hemos obtenido un excelente ajuste a los datos utilizados como ejemplos. De acuerdo con el modelo propuesto, fue posible identificar rangos de valores en las series temporales en que una variable independiente podría tener diferentes efectos. Este es un resultado importante desde el punto de vista clínico. Los estimados bayesianos fueron obtenidos usando métodos de Monte Carlo en Cadenas de Markov. Ejemplos basados en conjuntos de datos reales fueran usados para ilustrar el uso de los modelos propuestos y sus potenciales aplicaciones en investigaciones clínicas.

To make informed decisions in natural environments that change over time, humans must update their beliefs as new observations are gathered. Studies exploring human inference as a dynamical process that unfolds in time have focused on situations in which the statistics of observations are history-independent. Yet temporal structure is everywhere in nature, and yields history-dependent observations. Do humans modify their inference processes depending on the latent temporal statistics of their observations? We investigate this question experimentally and theoretically using a change-point inference task. We show that humans adapt their inference process to fine aspects of the temporal structure in the statistics of stimuli. As such, humans behave qualitatively in a Bayesian fashion, but, quantitatively, deviate away from optimality. Perhaps more importantly, humans behave suboptimally in that their responses are not deterministic, but variable. We show that this variability itself is modulated by the temporal statistics of stimuli. To elucidate the cognitive algorithm that yields this behavior, we investigate a broad array of existing and new models that characterize different sources of suboptimal deviations away from Bayesian inference. While models with 'output noise' that corrupts the response-selection process are natural candidates, human behavior is best described by sampling-based inference models, in which the main ingredient is a compressed approximation of the posterior, represented through a modest set of random samples and updated over time. This result comes to complement a growing literature on sample-based representation and learning in humans., Comment: 59 pages, 21 figures, and 6 tables

Background The nucleus of eukaryotic cells spatially packages chromosomes into a hierarchical and distinct segregation that plays critical roles in maintaining transcription regulation. High-throughput methods of chromosome conformation capture, such as Hi-C, have revealed topologically associating domains (TADs) that are defined by biased chromatin interactions within them. Results We introduce a novel method, HiCKey, to decipher hierarchical TAD structures in Hi-C data and compare them across samples. We first derive a generalized likelihood-ratio (GLR) test for detecting change-points in an interaction matrix that follows a negative binomial distribution or general mixture distribution. We then employ several optimal search strategies to decipher hierarchical TADs with p values calculated by the GLR test. Large-scale validations of simulation data show that HiCKey has good precision in recalling known TADs and is robust against random collisions of chromatin interactions. By applying HiCKey to Hi-C data of seven human cell lines, we identified multiple layers of TAD organization among them, but the vast majority had no more than four layers. In particular, we found that TAD boundaries are significantly enriched in active chromosomal regions compared to repressed regions. Conclusions HiCKey is optimized for processing large matrices constructed from high-resolution Hi-C experiments. The method and theoretical result of the GLR test provide a general framework for significance testing of similar experimental chromatin interaction data that may not fully follow negative binomial distributions but rather more general mixture distributions.

Building on a narrative approach to identify episodes of fiscal consolidation, data for a group of 17 industrial countries over the period 1978-2009 and continuous-time duration models, we find evidence suggesting that the likelihood of a fiscal consolidation ending increases over time, but only for programs that last less than six years. Additionally, fiscal consolidations tend to last longer in non-European than in European countries. Our results emphasize that chronic fiscal imbalances might lead to a vicious austerity cycle, while discipline in the behaviour of fiscal authorities is a means of achieving credible and shorter adjustment measures. Therefore, fiscal fatigue is likely to compromise the implementation and successfulness of fiscal consolidation programmes., COMPETE, QREN, FEDER, Fundação para a Ciência e a Tecnologia (FCT)

We propose a novel Bayesian methodology for analyzing nonstationary time series that exhibit oscillatory behaviour. We approximate the time series using a piecewise oscillatory model with unknown periodicities, where our goal is to estimate the change-points while simultaneously identifying the potentially changing periodicities in the data. Our proposed methodology is based on a trans-dimensional Markov chain Monte Carlo (MCMC) algorithm that simultaneously updates the change-points and the periodicities relevant to any segment between them. We show that the proposed methodology successfully identifies time changing oscillatory behaviour in two applications which are relevant to e-Health and sleep research, namely the occurrence of ultradian oscillations in human skin temperature during the time of night rest, and the detection of instances of sleep apnea in plethysmographic respiratory traces., Comment: Received 23 Oct 2018, Accepted 12 May 2019

International audience; In this paper, we propose a novel modeling and a new methodology for estimating the location of block boundaries in a random matrix consisting of a block-wise constant matrix corrupted with white noise. Our method consists in rewriting this problem as a variable selection issue. A penalized least-squares criterion with an l1-type penalty is used for dealing with this problem. Firstly, some theoretical results ensuring the consistency of our block boundaries estimators are provided. Secondly, we explain how to implement our approach in a very efficient way. This implementation is available in the R package blockseg which can be found in the Comprehensive R Archive Network. Thirdly, we provide some numerical experiments to illustrate the statistical and numerical performance of our package, as well as a thorough comparison with existing methods. Fourthly, an empirical procedure is proposed for estimating the number of blocks. Finally, our approach is applied to HiC data which are used in molecular biology for better understanding the influence of the chromosomal conformation on the cells functioning.

The relation between change points in multivariate surveillance is important but seldom considered. The sufficiency principle is here used to clarify the structure of some problems, to find efficient methods, and to determine appropriate evaluation metrics. We study processes where the changes occur simultaneously or with known time lags. The surveillance of spatial data is one example where known time lags can be of interest. A general version of a theorem for the sufficient reduction of processes that change with known time lags is given. A simulation study illustrates the benefits or the methods based on the sufficient statistics.

To segment a sequence of independent random variables at an unknown number of change-points, we introduce new procedures that are based on thresholding the likelihood ratio statistic. We also study confidence regions based on the likelihood ratio statistic for the change-points and joint confidence regions for the change-points and the parameter values. Applications to segment array CGH data are discussed., Comment: 48 pages

In this paper, we consider an estimation problem of the regression coefficients in multiple regression models with several unknown change-points. Under some realistic assumptions, we propose a class of estimators which includes as a special cases shrinkage estimators (SEs) as well as the unrestricted estimator (UE) and the restricted estimator (RE). We also derive a more general condition for the SEs to dominate the UE. To this end, we generalize some identities for the evaluation of the bias and risk functions of shrinkage-type estimators. As illustrative example, our method is applied to the "gross domestic product" data set of 10 countries whose USA, Canada, UK, France and Germany. The simulation results corroborate our theoretical findings., Comment: Published at http://dx.doi.org/10.3150/14-BEJ642 in the Bernoulli (http://isi.cbs.nl/bernoulli/) by the International Statistical Institute/Bernoulli Society (http://isi.cbs.nl/BS/bshome.htm)

BACKGROUND: Although oral cavity, pharyngeal, oesophageal and gastric cancers share some risk factors, no comparative analysis of mortality rate trends in these illnesses has been undertaken in Spain. This study aimed to evaluate the independent effects of age, death period and birth cohort on the mortality rates of these tumours. METHODS: Specific and age-adjusted mortality rates by tumour and sex were analysed. Age-period-cohort log-linear models were fitted separately for each tumour and sex, and segmented regression models were used to detect changes in period- and cohort-effect curvatures. RESULTS: Among men, the period-effect curvatures for oral cavity/pharyngeal and oesophageal cancers displayed a mortality trend that rose until 1995 and then declined. Among women, oral cavity/pharyngeal cancer mortality increased throughout the study period whereas oesophageal cancer mortality decreased after 1970. Stomach cancer mortality decreased in both sexes from 1965 onwards. Lastly, the cohort-effect curvature showed a certain degree of similarity for all three tumours in both sexes, which was greater among oral cavity, pharyngeal and oesophageal cancers, with a change point in evidence, after which risk of death increased in cohorts born from the 1910-1920s onwards and decreased among the 1950-1960 cohorts and successive generations. This latter feature was likewise observed for stomach cancer. CONCLUSIONS: While the similarities of the cohort effects in oral cavity/pharyngeal, oesophageal and gastric tumours support the implication of shared risk factors, the more marked changes in cohort-effect curvature for oral cavity/pharyngeal and oesophageal cancer could be due to the greater influence of some risk factors in their aetiology, such as smoking and alcohol consumption. The increase in oral cavity/pharyngeal cancer mortality in women deserves further study. The study was funded by a research grant from the Spanish Health Research Fund (FIS PI11/00871). DSM was funded by the Galician Health Service; NA, EF, JGP, PFN, RPB and GLA were funded by the Carlos III Institute of Health, Madrid, Spain; and MCA was funded by the CIBER in Epidemiology and Public Health, Spain (CIBERESP). The study design and research findings are the responsibility of the authors. Sí

We consider the problem of multiple change-point estimation in the mean of a Gaussian AR(1) process. Taking into account the dependence structure does not allow us to use the dynamic programming algorithm, which is the only algorithm giving the optimal solution in the independent case. We propose a robust estimator of the autocorrelation parameter, which is consistent and satisfies a central limit theorem. Then, we propose to follow the classical inference approach, by plugging this estimator in the criteria used for change-points estimation. We show that the asymptotic properties of these estimators are the same as those of the classical estimators in the independent framework. The same plug-in approach is then used to approximate the modified BIC and choose the number of segments. This method is implemented in the R package AR1seg and is available from the Comprehensive R Archive Network (CRAN). This package is used in the simulation section in which we show that for finite sample sizes taking into account the dependence structure improves the statistical performance of the change-point estimators and of the selection criterion., Comment: 42 pages, 13 figures

Agradecimentos: We are grateful to two anonymous reviewers for their helpful, thought-provoking comments on our original manuscript. We thank Professor Wensheng Guo for initially suggesting the smoothing spline approach, Professor Eli Glatstein for sharing his wealth of knowledge regarding PDT with us, and Professor ArjunYodh for his insights into DCS. Dr Andrew Pole contributed valuable insights into the original analyses of these data.We thank Joann Miller and Steven S. Schenkel for help with data collection. Research supported by NIH-NCI 5-P01-CA-087971, R01 CA85831, and CA087971-S1 and CA129554. Code for implementing the programs is available from the first author upon request Abstract: In a smoothing spline model with unknown change-points, the choice of the smoothing parameter strongly influences the estimation of the change-point locations, and the function at the changepoints. In a tumor biology example, where change-points in blood flow in response to treatment were of interest, choosing the smoothing parameter based on minimizing generalized cross validation, GCV, gave unsatisfactory estimates of the change-points. We propose a new method, aGCV, that re-weights the residual sum of squares and generalized degrees of freedom terms from GCV. The weight is chosen to maximize the decrease in the generalized degrees of freedom as a function of the weight value, while simultaneously minimizing aGCV as a function of the smoothing parameter and the change-points. Compared to GCV, simulation studies suggest that the aGCV method yields improved estimates of the change-point and the value of the function at the change-point Aberto

We consider Bayesian nonparametric inference for continuous-valued partially exchangeable data, when the partition of the observations into groups is unknown. This includes change-point problems and mixture models. As the prior, we consider a mixture of products of Dirichlet processes. We show that the discreteness of the Dirichlet process can have a large effect on inference (posterior distributions and Bayes factors), leading to conclusions that can be different from those that result from a reasonable parametric model. When the observed data are all distinct, the effect of the prior on the posterior is to favor more evenly balanced partitions, and its effect on Bayes factors is to favor more groups. In a hierarchical model with a Dirichlet process as the second-stage prior, the prior can also have a large effect on inference, but in the opposite direction, towards more unbalanced partitions.

BACKGROUND: In contrast to other haematological cancers, mortality from non-Hodgkin's lymphoma and multiple myeloma increased dramatically during the second half of the 20th century in most developed countries. This widespread upward trend remains controversial, as it may be attributable either to progressive improvements in diagnosis and certification or to increasing exposures to little-known but relevant risk factors. METHODS: To assess the relative contribution of these factors, we analysed the independent effects of age, death period, and birth cohort on haematological cancer mortality rates in Spain across the period 1952-2006. Weighted joinpoint regression analyses were performed to detect and estimate changes in period and cohort curvatures. RESULTS: Although mortality rates were consistently higher among men, trends across periods and cohorts were virtually identical in both sexes. There was an early period trend reversal in the 1960s for Hodgkin's disease and leukaemia, which was delayed to the 1980s for multiple myeloma and the 1990s for non-Hodgkin's lymphoma. Birth cohort patterns showed a first downturn for generations born in the 1900s and 1910s for all haematological cancers, and a second trend reversal for more recent cohorts born in the 1950s and 1960s for non-Hodgkin's lymphoma and leukaemia. CONCLUSIONS: The sustained decline in Hodgkin's disease mortality and the levelling off in leukaemia seem to be driven by an early period effect linked to improvements in disease treatment, whereas the steep upward trends in non-Hodgkin's lymphoma and multiple myeloma mortality in Spain are more likely explained by a cohort effect linked to better diagnosis and death certification in the elderly. The consistent male excess mortality across all calendar periods and age groups points to the importance of possible sex-related genetic markers of susceptibility in haematological cancers. The study was supported in part by a research grant from the Spanish Health Research Fund (FIS PI11/00871). The study sponsor had no role in the study design; the collection, analysis, and interpretation of data; the writing of the manuscript; and the decision to submit the manuscript for publication. Sí

In this paper we explore, analyse and apply the change-points detection and location procedures to conditional heteroskedastic processes. We focus on processes that have constant conditional mean, but present a dynamic behavior in the conditional variance and which can also be affected by structural changes. Thus, the goal is to explore, analyse and apply the change-point detection and estimation methods to the situation when the conditional variance of a univariate process is heteroskedastic and exhibits change-points. Based on the fact that a GARCH process can be expressed as an ARMA model in the squares of the variable, we propose to detect and locate change-points by using the Bayesian Information Criterion as an extension of its application in linear models. The proposed procedure is characterized by its computational simplicity, reducing difficulties of the change-point detection in the complex non-linear processes. We compare this procedure with others available in the literature, which are based on cusum methods (Inclán and Tiao (1994), Kokoszka and Leipus (1999), Lee et al. (2004)), informational approach (Fukuda, 2010), minimum description length principle (Davis and Rodriguez-Yam (2008)), and the time varying spectrum (Ombao et al (2002)). We compute the empirical size and power properties by Monte Carlo simulation experiments considering several scenarios. We obtained a good size and power properties in detecting even small magnitudes of change and for low levels of persistence. The procedures were applied to the S\&P500 log returns time series, in order to compare with the results in Andreou and Ghysels (2002) and Davis and Rodriguez-Yam (2008). Changepoints detected by the proposed procedure were similar to the breaks found by the other procedures, and their location can be related with the Southeast Asia financial crisis and with other known financial events.

Conference of 2013 38th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013 ; Conference Date: 26 May 2013 Through 31 May 2013; Conference Code:101421; International audience; In this paper, we develop a test procedure for non-stationarity for possibly long-memory processes. Contrary to most of the proposed methods, the test procedure has the same distribution for short-range and long-range dependence stationary processes. Such tests have been already proposed in [1], but these authors do not have taken into account the dependence of the wavelet coefficients within scales and between scales. We also propose an application to electric power consumption monitoring.

Building on a narrative approach to identify episodes of fiscal consolidation, data for a group of 17 industrial countries over the period 1978-2009 and both continuous-time and discrete-time duration models, we find evidence suggesting that likelihood of a fiscal consolidation ending increases over time, but only for programs that last less than six years. Additionally, fiscal consolidations tend to last longer in non-European than in European countries. Our results emphasize that chronic fiscal imbalances might lead to a vicious austerity cycle, while discipline in the behavior of fiscal authorities is a means of achieving credible and shorter adjustment measures., Fundação para a Ciência e a Tecnologia (FCT)

In this paper, we introduce an asymptotic test procedure to assess the stability of volatilities and cross-volatilites of linear and nonlinear multivariate time series models. The test is very flexible as it can be applied, for example, to many of the multivariate GARCH models established in the literature, and also works well in the case of high dimensionality of the underlying data. Since it is nonparametric, the procedure avoids the difficulties associated with parametric model selection, model fitting and parameter estimation. We provide the theoretical foundation for the test and demonstrate its applicability via a simulation study and an analysis of financial data. Extensions to multiple changes and the case of infinite fourth moments are also discussed., Comment: Published in at http://dx.doi.org/10.1214/09-AOS707 the Annals of Statistics (http://www.imstat.org/aos/) by the Institute of Mathematical Statistics (http://www.imstat.org)

"The objective of this paper is to assess the possible influence of the 1994 In Trust Agreements (ITAs) on acquisition and distribution of germplasm held by the International Research Rice Institute (IRRI) genebank. The agreements, legally affirmed the ???public good??? status of the collections that were placed ???In Trust??? for the benefit of the world community under agreements with FAO. They initiated a formal system of multilateral access to CGIAR-held ex situ genetic resources. The hypothesis that the consequences of the ITAs lead to an enhancement of CGIAR germplasm utilization is tested here using a basic conceptual framework to infer on factors determining the distribution of germplasm. Subsequently, a Bayesian empirical model is applied to IRRI accessions distribution???s time-series to provide formal evidence to the hypothesis. Results show that there is a discernible ???change??? point that would support a significant drop in germplasm distribution followed by a new growing trend around the establishment of the ITAs. This had followed a period beginning around 1989 and leading up to the establishment of the ITAs of a large number of requests for restoration of germplasm back to countries of origin and a reduction in acquisitions. As a result the number of accessions held by IRRI reached a low point around 1994. The number of accessions might not have been built back up without the establishment of a stable policy environment that was provided by the ITAs."