Temporal and statistical information in causal structure learning

Three experiments examined children’s and adults’ abilities to use statistical and temporal information to distinguish between common cause and causal chain structures. In Experiment 1, participants were provided with conditional probability information and/or temporal information and asked to infer the causal structure of a three-variable mechanical system that operated probabilistically. Participants of all ages preferentially relied on the temporal pattern of events in their inferences, even if this conflicted with statistical information. In Experiments 2 and 3, participants observed a series of interventions on the system, which in these experiments operated deterministically. In Experiment 2, participants found it easier to use temporal pattern information than statistical information provided as a result of interventions. In Experiment 3, in which no temporal pattern information was provided, children from 6-7 years, but not younger children, were able to use intervention information to make causal chain judgments, although they had difficulty when the structure was a common cause. The findings suggest that participants, and children in particular, may find it more difficult to use statistical information than temporal pattern information because of its demands on information processing resources. However, there may also be an inherent preference for temporal information.

Post-diagnostic prescriptions for low-dose aspirin and breast cancer-specific survival: a nested case-control study in a breast cancer cohort from the UK Clinical Practice Research Datalink

INTRODUCTION: Recent observational studies indicate that post-diagnostic use of aspirin in breast cancer patients may protect against cancer progression perhaps by inhibiting cyclooxygenase-2 dependent mechanisms. Evidence also supports a crucial role for interactions between tumour cells and circulating platelets in cancer growth and dissemination, therefore, use of low-dose aspirin may reduce the risk of death from cancer in breast cancer patients.

METHODS: A cohort of newly diagnosed breast cancer patients (1998 to 2006) were identified in the UK Clinical Practice Research Datalink (and confirmed by cancer registry linkage). Cancer-specific deaths were identified up to 2011 from Office for National Statistics mortality data. A nested case-control analysis was conducted using conditional logistic regression to compare post-diagnostic aspirin exposure using General Practice prescription data in 1,435 cases (breast cancer deaths) with 5,697 controls (matched by age and year of diagnosis).

RESULTS: After breast cancer diagnosis, 18.3% of cancer-specific deaths and 18.5% of matched controls received at least one prescription for low-dose aspirin, corresponding to an odds ratio (OR) of 0.98 (95% CI 0.83, 1.15). Adjustment for potential confounders (including stage and grade) had little impact on this estimate. No dose response relationship was observed when the number of tablets was investigated and no associations were seen when analyses were stratified by receipt of prescriptions for aspirin in the pre-diagnostic period, by stage at diagnosis or by receipt of prescriptions for hormone therapy.

CONCLUSIONS: Overall, in this large population-based cohort of breast cancer patients, there was little evidence of an association between receipt of post-diagnostic prescriptions for low-dose aspirin and breast cancer-specific death. However, information was not available on medication compliance or over-the-counter use of aspirin, which may have contributed to the null findings.

Genetic pleiotropy between multiple sclerosis and schizophrenia but not bipolar disorder: differential involvement of immune-related gene loci

Converging evidence implicates immune abnormalities in schizophrenia (SCZ), and recent genome-wide association studies (GWAS) have identified immune-related single-nucleotide polymorphisms (SNPs) associated with SCZ. Using the conditional false discovery rate (FDR) approach, we evaluated pleiotropy in SNPs associated with SCZ (n=21 856) and multiple sclerosis (MS) (n=43 879), an inflammatory, demyelinating disease of the central nervous system. Because SCZ and bipolar disorder (BD) show substantial clinical and genetic overlap, we also investigated pleiotropy between BD (n=16 731) and MS. We found significant genetic overlap between SCZ and MS and identified 21 independent loci associated with SCZ, conditioned on association with MS. This enrichment was driven by the major histocompatibility complex (MHC). Importantly, we detected the involvement of the same human leukocyte antigen (HLA) alleles in both SCZ and MS, but with an opposite directionality of effect of associated HLA alleles (that is, MS risk alleles were associated with decreased SCZ risk). In contrast, we found no genetic overlap between BD and MS. Considered together, our findings demonstrate genetic pleiotropy between SCZ and MS and suggest that the MHC signals may differentiate SCZ from BD susceptibility.Molecular Psychiatry advance online publication, 28 January 2014; doi:10.1038/mp.2013.195.

Due North? Do American Theories of Legislative Committees Apply to Canada?

Several theories of legislative organisation have been proposed to explain committee selection in American legislatures, but do these theories travel outside the United States? This paper tests whether these theories apply to data from the Canadian House of Commons. It was found that the distributive and partisan models of legislative organisation explain committee composition in Canada. In many cases, committees in the House of Commons are made up of preference outliers. As predicted by partisan models, it was also found that the governing party stacks committees with its members, but this is conditional upon the strength of the governing party.

Civil Society Aid to Post-Communist Countries

A major debate within foreign aid literature is whether civil society can be ‘purchased’ through outside assistance.We test this proposition by exploring the influence of aid provided by the United States Agency for International
Development on post-communist civil rights environments. A review of research critical of international assistance highlights the risk of unsustainability, polarization and dependence among recipient civic organizations.We argue that
a more effective stimulant is socio-economic growth, which stimulates committed constituencies, higher citizen expectations and pressure on the state to protect civil freedoms. Using cross-sectional, time-series data from 27
post-communist countries, we find no evidence that aid independently promotes stronger civil rights environments but that economic growth produces substantial improvements. Further, any aid effectiveness appears to be conditional on economic strength.We conclude that developmental organizations should reassess how and where civil society aid is targeted.

Efficient Marginal Likelihood Computation for Gaussian Process Regression

In a Bayesian learning setting, the posterior distribution of a predictive model arises from a trade-off between its prior distribution and the conditional likelihood of observed data. Such distribution functions usually rely on additional hyperparameters which need to be tuned in order to achieve optimum predictive performance; this operation can be efficiently performed in an Empirical Bayes fashion by maximizing the posterior marginal likelihood of the observed data. Since the score function of this optimization problem is in general characterized by the presence of local optima, it is necessary to resort to global optimization strategies, which require a large number of function evaluations. Given that the evaluation is usually computationally intensive and badly scaled with respect to the dataset size, the maximum number of observations that can be treated simultaneously is quite limited. In this paper, we consider the case of hyperparameter tuning in Gaussian process regression. A straightforward implementation of the posterior log-likelihood for this model requires O(N^3) operations for every iteration of the optimization procedure, where N is the number of examples in the input dataset. We derive a novel set of identities that allow, after an initial overhead of O(N^3), the evaluation of the score function, as well as the Jacobian and Hessian matrices, in O(N) operations. We prove how the proposed identities, that follow from the eigendecomposition of the kernel matrix, yield a reduction of several orders of magnitude in the computation time for the hyperparameter optimization problem. Notably, the proposed solution provides computational advantages even with respect to state of the art approximations that rely on sparse kernel matrices.

Kuznetsov independence for interval-valued expectations and sets of probability distributions: Properties and algorithms

Kuznetsov independence of variables X and Y means that, for any pair of bounded functions f(X) and g(Y), E[f(X)g(Y)]=E[f(X)] *times* E[g(Y)], where E[.] denotes interval-valued expectation and *times* denotes interval multiplication. We present properties of Kuznetsov independence for several variables, and connect it with other concepts of independence in the literature; in particular we show that strong extensions are always included in sets of probability distributions whose lower and upper expectations satisfy Kuznetsov independence. We introduce an algorithm that computes lower expectations subject to judgments of Kuznetsov independence by mixing column generation techniques with nonlinear programming. Finally, we define a concept of conditional Kuznetsov independence, and study its graphoid properties.

Inference in credal networks using multilinear programming

A credal network is a graphical tool for representation and manipulation of uncertainty, where probability values may be imprecise or indeterminate. A credal network associates a directed acyclic graph with a collection of sets of probability measures; in this context, inference is the computation of tight lower and upper bounds for conditional probabilities. In this paper we present new algorithms for inference in credal networks based on multilinear programming techniques. Experiments indicate that these new algorithms have better performance than existing ones, in the sense that they can produce more accurate results in larger networks.

The role of SOCS3/pSTAT3 pathway in the development of diabetic retinopathy

Purpose: Suppressor of cytokine signalling (SOCS) proteins are feedback inhibitors of the JAK/STAT pathway. SOCS3 critically controls STAT3 activation, cytokine signalling and inflammatory gene expression in macrophages and microglia. In this study, we investigated the role of SOCS3/STAT3 in myeloid cells in the initiation and progression of diabetic retinopathy (DR). 
Methods: Mice with a conditional deletion of SOCS3 in myeloid cells (LysMCre-SOCS3 fl/fl) and C57BL/6J (as control) were rendered diabetic by a low-dose multiple intraperitoneal injections of Stroptozocine. Diabetes related retinal changes, including leukostasis, acellular capilliaries, and microglial activation were assessed at different stages of disease. Bone marrow derived macrophages (BMDMs) from LysMCreSOCS3 fl/fl and C57BL/6J mice were cultured in high glucose (HG) medium, and cell activation was evaluated by real-time RT-PCR.
Results: In C57BL/6J diabetic mice the expression of phosphorylated STAT3 (pSTAT3) was increased and SOCS3 was decreased in the retina. Interleukin 6 (IL-6), the main cytokine that stimulates STAT3 activation, was increased in the plamsa in diabetic mice. Although blood glucose levels and Hbac-1 were comparable between LysMCre-SOCS3fl/fl and WT mice after STZ injection, the LysMCreSOCS3 fl/fl diabetic mice developed severe retinal vasculopathy, including increased leukostasis and microglial activation at one month and enhanced acellular capillary formation at 6 months after diabetes induction. 
Conclusions: our study suggests that the JAK/STAT3 pathway is involved in the initiation and progression of DR, and uncontrolled STAT3 activation results in accelerated DR progression. Targeting the STAT3 pathway may be a novel approach for the management of DR.

A belief revision framework for revising epistemic states with partial epistemic states

Belief revision performs belief change on an agent’s beliefs when new evidence (either of the form of a propositional formula or of the form of a total pre-order on a set of interpretations) is received. Jeffrey’s rule is commonly used for revising probabilistic epistemic states when new information is probabilistically uncertain. In this paper, we propose a general epistemic revision framework where new evidence is of the form of a partial epistemic state. Our framework extends Jeffrey’s rule with uncertain inputs and covers well-known existing frameworks such as ordinal conditional function (OCF) or possibility theory. We then define a set of postulates that such revision operators shall satisfy and establish representation theorems to characterize those postulates. We show that these postulates reveal common characteristics of various existing revision strategies and are satisfied by OCF conditionalization, Jeffrey’s rule of conditioning and possibility conditionalization. Furthermore, when reducing to the belief revision situation, our postulates can induce Darwiche and Pearl’s postulates C1 and C2.

β-Blocker usage and colorectal cancer mortality:a nested case-control study in the UK Clinical Practice Research Datalink cohort

BACKGROUND: Epidemiological and laboratory studies suggest that β-blockers may reduce cancer progression in various cancer sites. The aim of this study was to conduct the first epidemiological investigation of the effect of post-diagnostic β-blocker usage on colorectal cancer-specific mortality in a large population-based colorectal cancer patient cohort.

PATIENTS AND METHODS: A nested case-control analysis was conducted within a cohort of 4794 colorectal cancer patients diagnosed between 1998 and 2007. Patients were identified from the UK Clinical Practice Research Datalink and confirmed using cancer registry data. Patients with a colorectal cancer- specific death (data from the Office of National Statistics death registration system) were matched to five controls. Conditional logistic regression was applied to calculate odds ratios (OR) and 95% confidence intervals (95% CIs) according to β-blocker usage (data from GP-prescribing records).

RESULTS: Post-diagnostic β-blocker use was identified in 21.4% of 1559 colorectal cancer-specific deaths and 23.7% of their 7531 matched controls, with little evidence of an association (OR = 0.89 95% CI 0.78-1.02). Similar associations were found when analysing drug frequency, β-blocker type or specific drugs such as propranolol. There was some evidence of a weak reduction in all-cause mortality in β-blocker users (adjusted OR = 0.88; 95% CI 0.77-1.00; P = 0.04) which was in part due to the marked effect of atenolol on cardiovascular mortality (adjusted OR = 0.62; 95% CI 0.40-0.97; P = 0.04).

CONCLUSIONS: In this novel, large UK population-based cohort of colorectal cancer patients, there was no evidence of an association between post-diagnostic β-blocker use and colorectal cancer-specific mortality.

CLINICAL TRIALS NUMBER: NCT00888797.

Hui and Walter's latent-class model extended to estimate diagnostic test properties from surveillance data: a latent model for latent data

Diagnostic test sensitivity and specificity are probabilistic estimates with far reaching implications for disease control, management and genetic studies. In the absence of 'gold standard' tests, traditional Bayesian latent class models may be used to assess diagnostic test accuracies through the comparison of two or more tests performed on the same groups of individuals. The aim of this study was to extend such models to estimate diagnostic test parameters and true cohort-specific prevalence, using disease surveillance data. The traditional Hui-Walter latent class methodology was extended to allow for features seen in such data, including (i) unrecorded data (i.e. data for a second test available only on a subset of the sampled population) and (ii) cohort-specific sensitivities and specificities. The model was applied with and without the modelling of conditional dependence between tests. The utility of the extended model was demonstrated through application to bovine tuberculosis surveillance data from Northern and the Republic of Ireland. Simulation coupled with re-sampling techniques, demonstrated that the extended model has good predictive power to estimate the diagnostic parameters and true herd-level prevalence from surveillance data. Our methodology can aid in the interpretation of disease surveillance data, and the results can potentially refine disease control strategies.

Explaining variation in cancer survival between 11 jurisdictions in the International Cancer Benchmarking Partnership: a primary care vignette survey

OBJECTIVES: The International Cancer Benchmarking Partnership (ICBP) is a collaboration between 6 countries and 12 jurisdictions with similar primary care-led health services. This study investigates primary care physician (PCP) behaviour and systems that may contribute to the timeliness of investigating for cancer and subsequently, international survival differences.

DESIGN: A validated survey administered to PCPs via the internet set out in two parts: direct questions on primary care structure and practice relating to cancer diagnosis, and clinical vignettes, assessing management of scenarios relating to the diagnosis of lung, colorectal or ovarian cancer.

PARTICIPANTS: 2795 PCPs in 11 jurisdictions: New South Wales and Victoria (Australia), British Columbia, Manitoba, Ontario (Canada), England, Northern Ireland, Wales (UK), Denmark, Norway and Sweden.

PRIMARY AND SECONDARY OUTCOME MEASURES: Analysis compared the cumulative proportion of PCPs in each jurisdiction opting to investigate or refer at each phase for each vignette with 1-year survival, and conditional 5-year survival rates for the relevant cancer and jurisdiction. Logistic regression was used to explore whether PCP characteristics or system differences in each jurisdiction affected the readiness to investigate.

RESULTS: 4 of 5 vignettes showed a statistically significant correlation (p<0.05 or better) between readiness to investigate or refer to secondary care at the first phase of each vignette and cancer survival rates for that jurisdiction. No consistent associations were found between readiness to investigate and selected PCP demographics, practice or health system variables.

CONCLUSIONS: We demonstrate a correlation between the readiness of PCPs to investigate symptoms indicative of cancer and cancer survival rates, one of the first possible explanations for the variation in cancer survival between ICBP countries. No specific health system features consistently explained these findings. Some jurisdictions may consider lowering thresholds for PCPs to investigate for cancer-either directly, or by specialist referral, to improve outcomes.

, Lexicographic Preferences in Discrete Choice Experiments: Consequences on Individual-Specific Willingness to Pay Estimates

In discrete choice experiments respondents are generally assumed to consider all of the attributes across each of the alternatives, and to choose their most preferred. However, results in this paper indicate that many respondents employ simplified lexicographic decision-making rules, whereby they have a ranking of the attributes, but their choice of an alternative is based solely on the level of their most important attribute(s). Not accounting for these simple decision-making heuristics introduces systemic errors and leads to biased point estimates, as they are a violation of the continuity axiom and a departure from the use of compensatory decision-making. In this paper the implications of lexicographic preferences are examined. In particular, using a mixed logit specification this paper investigates the sensitivity of individual-specific willingness to pay (WTP) estimates conditional on whether lexicographic decision-making rules are accounted for in the modelling of discrete choice responses. Empirical results are obtained from a discrete choice experiment that was carried out to address the value of a number of rural landscape attributes in Ireland

A chemically unlocked binary molecular switch

A highly luminescent and sensitive terbium complex of a ligand comprising of a phthalimide group appended to a DO3A moiety is an active pH sensor that is conditional on its previous pH. © 2012 The Royal Society of Chemistry.