Bayesian network data imputation with application to survival tree analysis

Retrospective clinical datasets are often characterized by a relatively small sample size and many missing data. In this case, a common way for handling the missingness consists in discarding from the analysis patients with missing covariates, further reducing the sample size. Alternatively, if the mechanism that generated the missing allows, incomplete data can be imputed on the basis of the observed data, avoiding the reduction of the sample size and allowing methods to deal with complete data later on. Moreover, methodologies for data imputation might depend on the particular purpose and might achieve better results by considering specific characteristics of the domain. The problem of missing data treatment is studied in the context of survival tree analysis for the estimation of a prognostic patient stratification. Survival tree methods usually address this problem by using surrogate splits, that is, splitting rules that use other variables yielding similar results to the original ones. Instead, our methodology consists in modeling the dependencies among the clinical variables with a Bayesian network, which is then used to perform data imputation, thus allowing the survival tree to be applied on the completed dataset. The Bayesian network is directly learned from the incomplete data using a structural expectation–maximization (EM) procedure in which the maximization step is performed with an exact anytime method, so that the only source of approximation is due to the EM formulation itself. On both simulated and real data, our proposed methodology usually outperformed several existing methods for data imputation and the imputation so obtained improved the stratification estimated by the survival tree (especially with respect to using surrogate splits).

Evaluating Train Protection Systems

This paper arose from the work carried out for the Cullen/Uff Joint Inquiry into Train Protection Systems. It is concerned with the problem of evaluating the benefits of safety enhancements in order to avoid rare, but catastrophic accidents, and the role of Operations Research in the process. The problems include both input values and representation of outcomes. A key input is the value of life. This paper briefly discusses why the value of life might vary from incident to incident and reviews alternative estimates before producing a 'best estimate' for rail. When the occurrence of an event is uncertain, the normal method is to apply a single 'expected' value. This paper argues that a more effective method of representing such situations is through Monte-Carlo simulation and demonstrates the use of the methodology on a case study of the decision as to whether or not advanced train protection (ATP) should have been installed on a route to the west of London. This paper suggests that the output is more informative than traditional cost-benefit appraisals or engineering event tree approaches. It also shows that, unlike the results from utilizing the traditional approach, the value of ATP on this route would be positive over 50% of the time.

A Note on Race, Ethnicity and Nativity Differences in Remarriage in the United States

The objectives of this study are to produce up-to-date estimates of race/ethnic/nativity differentials for remarriage and repartnership among women in the United States and to see if these differences are due to across-group differences in demographic characteristics. First, we produce lifetable estimates of remarriage and repartnering for white, black, U.S. born Latina and foreign born Latina women. Next, we estimate race/ethnic/nativity differentials for remarriage and repartnership using event-history analysis with and without controls for demographic characteristics. The results suggest a continued overall decline in remarriage rates, while many women repartner by cohabitating. Whites are more likely than blacks or Latinas to remarry and they are also more likely to repartner. Race/ethnic/nativity differentials remain even after accounting for variations in demographic characteristics. This suggests that race/ethnic/nativity differentials in remarriage and repartnering rates, rather than ameliorating disadvantages associated with divorce, reinforce these differentials.

Gender-specific genomic profiling in metastatic colorectal cancer patients treated with 5-fluorouracil and oxaliplatin

AIMS: Survival and response rates in metastatic colorectal cancer remain poor, despite advances in drug development. There is increasing evidence to suggest that gender-specific differences may contribute to poor clinical outcome. We tested the hypothesis that genomic profiling of metastatic colorectal cancer is dependent on gender.

MATERIALS & METHODS: A total of 152 patients with metastatic colorectal cancer who were treated with oxaliplatin and continuous infusion 5-fluorouracil were genotyped for 21 polymorphisms in 13 cancer-related genes by PCR. Classification and regression tree analysis tested for gender-related association of polymorphisms with overall survival, progression-free survival and tumor response.

RESULTS: Classification and regression tree analysis of all polymorphisms, age and race resulted in gender-specific predictors of overall survival, progression-free survival and tumor response. Polymorphisms in the following genes were associated with gender-specific clinical outcome: estrogen receptor β, EGF receptor, xeroderma pigmentosum group D, voltage-gated sodium channel and phospholipase A2.

CONCLUSION: Genetic profiling to predict the clinical outcome of patients with metastatic colorectal cancer may depend on gender.

Dynamics of North Patagonian rainforests from fine-resolution pollen, charcoal and tree-ring analysis, Chonos Archipelago, Southern Chile

Fine-resolution palaeoecological and dendrochronological methods were used to investigate the impacts of climate change, and natural and anthropogenic disturbances on vegetation in the North Patagonian rainforest of southern Chile at decadal to century timescales during the late Holocene. A lake sediment mud–water interface core was collected from the northern Chonos Archipelago and analysed for pollen and charcoal. Dendrochronological analysis of tree cores collected from stands of Pilgerodendron uviferum close to the lake site was incorporated into the study. The combined analysis showed that the present mosaic of vegetation types in this region is a function of environmental changes across a range of timescales: millennial climate change, more recent natural and anthropogenic disturbances, and possibly short-term climatic variations. Of particular interest is the spatiotemporal distribution of Pilgerodendron uviferum dieback/burning in the Chonos Archipelago region.

Multiproxy evidence for terrestrial and aquatic ecosystem responses during the 8.2 ka cold event as recorded at Hojby So, Denmark

A sediment succession from Hojby So, a lake in eastern Denmark, covering the time period 9400-7400 cal yr BP was studied using high-resolution geochemistry, magnetic susceptibility, pollen, macrofossil, diatom, and algal pigment analysis to investigate responses of the terrestrial and aquatic ecosystems to the 8.2 ka cold event. A reduced pollen production by thermophilous deciduous tree taxa in the period c. 8250-8000 cal yr BP reveal that the forest ecosystem was affected by low temperatures during the summer and winter/early-spring seasons. This finding is consistent with the timing of the 8.2 ka cold event as registered in the Greenland ice cores. At Hojby So, the climate anomaly appears to have started 200-250 yr earlier than the 8.2 ka cold event as the lake proxy data provide strong evidence for a precipitation-induced distinct increase in catchment soil erosion beginning around 8500 cal yr BP. Alteration of the terrestrial environment then resulted in a major aquatic ecosystem change with nutrient enrichment of the lake and enhanced productivity, which lasted until c. 7900 cal yr BP. (C) 2009 University of Washington. Published by Elsevier Inc. All rights reserved.

Hydroclimatic changes during the 8.2 ka event revealed by Irish pines

A pivotal cold event, deduced from the Greenland ice cores, took place between 8200 and 8000 cal. BP. Modelling of this climatic episode suggests that higher northern latitudes would have also experienced substantial reduction in rainfall and that Ireland would have observed a notable decline. No well-dated proxy record exists from the British Isles to test the model results. We present significant independent data for a phase of increased Scots pine initiation on Irish bogs at around 8150 cal. BP. Dendrochronological dating of sub-fossil Scots pine trees from three locations reveals synchronicity in germination across the area, indicative of a regional forcing, and allows for high-precision radiocarbon based dates. The starting rings of 40% of all samples from the north of Ireland dating to the period 8500-7500 cal. BP fall within a period of 25 years. The present colonisation model of Scots pine on peatland is interpreted as increasing drier conditions in the region and provides the first meaningful proxy data in support of a significant hydrological change in the north of Ireland accompanying the 8.2 ka event. The dating uncertainties associated with the Irish Scots pine record and the Greenland Ice Core Chronology 2005 (GICC05) do not allow for any overlap between the two. The results indicate that the discrepancy could be a result of dating inaccuracy that could have affected analysis of prior proxy alignments.

Low alanine aminotransferase levels and higher number of cardiovascular events in people with Type 2 diabetes: analysis of the Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) study.

AIMS: To determine whether alanine aminotransferase or gamma-glutamyltransferase levels, as markers of liver health and non-alcoholic fatty liver disease, might predict cardiovascular events in people with Type 2 diabetes.

METHODS: Data from the Fenofibrate Intervention and Event Lowering in Diabetes study were analysed to examine the relationship between liver enzymes and incident cardiovascular events (non-fatal myocardial infarction, stroke, coronary and other cardiovascular death, coronary or carotid revascularization) over 5 years.

RESULTS: Alanine aminotransferase level had a linear inverse relationship with the first cardiovascular event occurring in participants during the study period. After adjustment, for every 1 sd higher baseline alanine aminotransferase value (13.2 U/l), the risk of a cardiovascular event was 7% lower (95% CI 4-13; P=0.02). Participants with alanine aminotransferase levels below and above the reference range 8-41 U/l for women and 9-59 U/l for men, had hazard ratios for a cardiovascular event of 1.86 (95% CI 1.12-3.09) and 0.65 (95% CI 0.49-0.87), respectively (P=0.001). No relationship was found for gamma-glutamyltransferase.

CONCLUSIONS: The data may indicate that in people with Type 2 diabetes, which is associated with higher alanine aminotransferase levels because of prevalent non-alcoholic fatty liver disease, a low alanine aminotransferase level is a marker of hepatic or systemic frailty rather than health. This article is protected by copyright. All rights reserved.

Phylogeographic analysis of the red seaweed Palmaria palmata reveals a Pleistocene marine glacial refugium in the English Channel

Phylogeography has provided a new approach to the analysis of the postglacial history of a wide range of taxa but, to date, little is known about the effect of glacial periods on the marine biota of Europe. We have utilized a combination of nuclear, plastid and mitochondrial genetic markers to study the biogeographic history of the red seaweed Palmaria palmata in the North Atlantic. Analysis of the nuclear rDNA operon (ITS1-5.8S-ITS2), the plastid 16S-trnI-trnA-23S-5S, rbcL-rbcS and rpl12-rps31-rpl9 regions and the mitochondrial cox2–3 spacer has revealed the existence of a previously unidentified marine refugium in the English Channel, along with possible secondary refugia off the southwest coast of Ireland and in northeast North America and/or Iceland. Coalescent and mismatch analyses date the expansion of European populations from approximately 128 000 bp and suggest a continued period of exponential growth since then. Consequently, we postulate that the penultimate (Saale) glacial maximum was the main event in shaping the biogeographic history of European P. palmata populations which persisted throughout the last (Weichselian) glacial maximum (c. 20 000 bp) in the Hurd Deep, an enigmatic trench in the English Channel.

Proteomic Analysis of Recurrent Joint Inflammation in Juvenile Idiopathic Arthritis

The synovial fluid proteome in juvenile idiopathic arthritis was investigated to isolate joint-specific biomarkers that are expressed in patients displaying recurrent joint inflammation. To identify the synovial specific proteome, matched synovial fluid and plasma samples were subjected to protein separation by 2-dimension electrophoresis (2DE). Forty-three protein spots, overexpressed in the joint, were identified. Synovial fluids from children with single-event knee joint inflammation were then compared with a group with recurrent knee disease. Nine synovial specific proteins were significantly differentially expressed in the recurrent group. Proteolytic fragments of collagen X, fibrin beta-chain, and T-cell receptor alpha-region have been identified among this protein cluster. Putative biomarkers, overexpressed in the joint and differentially expressed in children with recurrent joint inflammation, have been identified. These proteins may play a significant role determining the pathological state within the chronically inflamed joint and influence disease progression in JIA. This is the first study of the synovial proteome in children.

Numerical analysis of surface and fossil pollen spectra from northern Fennoscandia

Aim Determination of the main directions of variance in an extensive data base of annual pollen deposition, and the relationship between pollen data from modified Tauber traps and palaeoecological data. Location Northern Finland and Norway. Methods Pollen analysis of annual samples from pollen traps and contiguous high-resolution samples from a peat sequence. Numerical analysis (principal components analysis) of the resulting data. Results The main direction of variation in the trap data is due to the vegetation region in which each trap is located. A secondary direction of variation is due to the annual variability of pollen production of some of the tree taxa, especially Betula and Pinus. This annual variability is more conspicuous in ‘absolute’ data than it is in percentage data which, at this annual resolution, becomes more random. There are systematic differences, with respect to peat-forming taxa, between pollen data from traps and pollen data from a peat profile collected over the same period of time. Main conclusions Annual variability in pollen production is rarely visible in fossil pollen samples because these cannot be sampled at precisely a 12-month resolution. At near-annual resolution sampling, it results in erratic percentage values which do not reflect changes in vegetation. Profiles sampled at near annual resolution are better analysed in terms of pollen accumulation rates with the realization that even these do not record changes in plant abundance but changes in pollen abundance. However, at the coarser temporal resolution common in most fossil samples it does not mask the origin of the pollen in terms of its vegetation region. Climate change may not be recognizable from pollen assemblages until the change has persisted in the same direction sufficiently long enough to alter the flowering (pollen production) pattern of the dominant trees.

High levels of genetic structuring as a result of population fragmentation in the tropical tree species Caesalpinia echinata Lam.

We have investigated levels of genetic diversity within and among seven remnant populations of Caesalpinia echinata Lam., an endangered species found as fragmented populations in three major areas around the coastal regions of Brazil. Using amplified fragment length polymorphism (AFLP) genetic markers, we detected levels of within-population genetic diversity ranging from 0.092 to 0.163, with the lowest values generally being found in the smallest populations. Estimates of between-population genetic differentiation were strongly correlated with geographical distance ( r = 0.884, p <0.001), which, along with a neighbour-joining phylogenetic analysis, strongly suggested high levels of genetic isolation by distance. Over half (62%) of the total genetic diversity was partitioned between populations, further highlighting the genetic distinctness of individual populations. Taken together, these results suggest that fragmentation has led to an increase in population differentiation between fragments of C. echinata. These formations will be of great value in the development of conservation plans for species exhibiting high levels of genetic differentiation due to fragmentation, such as indication of conservation unit size, which populations should be chosen as priority in conservation plans and which samples should be introduced in areas with a low number of individuals of brazilwood.

Restricted gene flow in fragmented populations of a wind-pollinated tree

Fragmentation of natural populations can have negative effects at the genetic level, thus threatening their evolutionary potential. Many of the negative genetic impacts of population fragmentation can be ameliorated by gene flow and it has been suggested that in wind-pollinated tree species, high or even increased levels of gene flow are a feature of fragmented populations, although several studies have disputed this. We have used a combination of nuclear microsatellites and allele-specific PCR (AS-PCR) analysis of chloroplast single nucleotide polymorphisms (SNPs) to examine the levels and patterns of genetic diversity and population differentiation in fragmented populations of juniper (Juniperus communis) in Ireland and inform conservation programs for the species. Significant population differentiation was found for both chloroplast and nuclear markers, indicating restricted gene flow, particularly over larger geographic scales. For conservation purposes, the existence of genetically distinct clusters and geographically localised chloroplast haplotypes suggests that the concept of provenance should be taken into account when formulating augmentation or reintroduction strategies. Furthermore, the potential lack of seed dispersal and seedling establishment means that ex-situ approaches to seed and seedling management may have to be considered.