Mapping an uncertainty zone between interpolated types of a categorical variable

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Mapping an uncertainty zone between interpolated types of a categorical variable

Categorical data cannot be interpolated directly because they are outcomes of discrete random variables. Thus, types of categorical variables are transformed into indicator functions that can be handled by interpolation methods. Interpolated indicator values are then backtransformed to the original types of categorical variables. However, aspects such as variability and uncertainty of interpolated values of categorical data have never been considered. In this paper we show that the interpolation variance can be used to map an uncertainty zone around boundaries between types of categorical variables. Moreover, it is shown that the interpolation variance is a component of the total variance of the categorical variables, as measured by the coefficient of unalikeability. (C) 2011 Elsevier Ltd. All rights reserved.

Using Bayesian networks with rule extraction to infer the risk of weed infestation in a corn-crop

This paper describes the modeling of a weed infestation risk inference system that implements a collaborative inference scheme based on rules extracted from two Bayesian network classifiers. The first Bayesian classifier infers a categorical variable value for the weed-crop competitiveness using as input categorical variables for the total density of weeds and corresponding proportions of narrow and broad-leaved weeds. The inferred categorical variable values for the weed-crop competitiveness along with three other categorical variables extracted from estimated maps for the weed seed production and weed coverage are then used as input for a second Bayesian network classifier to infer categorical variables values for the risk of infestation. Weed biomass and yield loss data samples are used to learn the probability relationship among the nodes of the first and second Bayesian classifiers in a supervised fashion, respectively. For comparison purposes, two types of Bayesian network structures are considered, namely an expert-based Bayesian classifier and a naive Bayes classifier. The inference system focused on the knowledge interpretation by translating a Bayesian classifier into a set of classification rules. The results obtained for the risk inference in a corn-crop field are presented and discussed. (C) 2009 Elsevier Ltd. All rights reserved.

O efeito da idade da empresa nos salários

Dissertação de Mestrado, Ciências Económicas e Empresariais, 10 de Dezembro de 2015, Universidade dos Açores.

Prophylactic Use of Liposomal Amphotericin B in Preventing Fungal Infections Early After Liver Transplantation: a Retrospective, Single-Center Study

In this study the authors evaluated the efficacy of prophylaxis with liposomal amphotericin B (L-AmB) in the incidence of fungal infections (FI) during the first 3 months after liver transplant (LT). The study was retrospective and accessed a 4-year period from 2008 to 2011. All patients who died in the first 48 hours after LT were excluded. Patients were divided by the risk groups for FI: Group 1, high-risk (at least 1 of the following conditions: urgent LT; serum creatinine >2 mg/dL; early acute kidney injury [AKI] after LT; retransplantation; surgical exploration early post-LT; transfused cellular blood components [>40 U]); and Group 2, low-risk patients. Group 1 patients were further separated into those who received antifungal prophylaxis with L-AmB and those who did not. Prophylaxis with L-AmB consisted of intravenous administration of L-AmB, 100 mg daily for 14 days. Four hundred ninety-two patients underwent LT; 31 died in the first 48 hours after LT. From the remaining 461 patients, 104 presented with high-risk factors for FI (Group 1); of these, 66 patients received antifungal prophylaxis and 38 did not. In this group 8 FI were observed, 5 in patients without antifungal prophylaxis (P = .011). Three more FI were identified in Group 2. By logistic regression analysis, the categorical variable high-risk group was independently related to the occurrence of invasive FI (P = .006). We conclude that prophylaxis with L-AmB after LT was effective in reducing the incidence of FI. No influence on mortality was detected.

Mixing compositions and other scales

Theory of compositional data analysis is often focused on the composition only. However in practical applications we often treat a composition together with covariableswith some other scale. This contribution systematically gathers and develop statistical tools for this situation. For instance, for the graphical display of the dependenceof a composition with a categorical variable, a colored set of ternary diagrams mightbe a good idea for a first look at the data, but it will fast hide important aspects ifthe composition has many parts, or it takes extreme values. On the other hand colored scatterplots of ilr components could not be very instructive for the analyst, if theconventional, black-box ilr is used.Thinking on terms of the Euclidean structure of the simplex, we suggest to set upappropriate projections, which on one side show the compositional geometry and on theother side are still comprehensible by a non-expert analyst, readable for all locations andscales of the data. This is e.g. done by defining special balance displays with carefully-selected axes. Following this idea, we need to systematically ask how to display, explore,describe, and test the relation to complementary or explanatory data of categorical, real,ratio or again compositional scales.This contribution shows that it is sufficient to use some basic concepts and very fewadvanced tools from multivariate statistics (principal covariances, multivariate linearmodels, trellis or parallel plots, etc.) to build appropriate procedures for all these combinations of scales. This has some fundamental implications in their software implementation, and how might they be taught to analysts not already experts in multivariateanalysis

Tea consumption and incidence of type 2 diabetes in Europe: the EPIC-InterAct case-cohort study.

BACKGROUND In previous meta-analyses, tea consumption has been associated with lower incidence of type 2 diabetes. It is unclear, however, if tea is associated inversely over the entire range of intake. Therefore, we investigated the association between tea consumption and incidence of type 2 diabetes in a European population. METHODOLOGY/PRINCIPAL FINDINGS The EPIC-InterAct case-cohort study was conducted in 26 centers in 8 European countries and consists of a total of 12,403 incident type 2 diabetes cases and a stratified subcohort of 16,835 individuals from a total cohort of 340,234 participants with 3.99 million person-years of follow-up. Country-specific Hazard Ratios (HR) for incidence of type 2 diabetes were obtained after adjustment for lifestyle and dietary factors using a Cox regression adapted for a case-cohort design. Subsequently, country-specific HR were combined using a random effects meta-analysis. Tea consumption was studied as categorical variable (0, >0-<1, 1-<4, ≥ 4 cups/day). The dose-response of the association was further explored by restricted cubic spline regression. Country specific medians of tea consumption ranged from 0 cups/day in Spain to 4 cups/day in United Kingdom. Tea consumption was associated inversely with incidence of type 2 diabetes; the HR was 0.84 [95%CI 0.71, 1.00] when participants who drank ≥ 4 cups of tea per day were compared with non-drinkers (p(linear trend) = 0.04). Incidence of type 2 diabetes already tended to be lower with tea consumption of 1-<4 cups/day (HR = 0.93 [95%CI 0.81, 1.05]). Spline regression did not suggest a non-linear association (p(non-linearity) = 0.20). CONCLUSIONS/SIGNIFICANCE A linear inverse association was observed between tea consumption and incidence of type 2 diabetes. People who drink at least 4 cups of tea per day may have a 16% lower risk of developing type 2 diabetes than non-tea drinkers.

Body mass index and health-related quality of life among young Swiss men.

BACKGROUND: Studies about the association between body mass index (BMI) and health-related quality of life (HRQOL) are often limited, because they 1) did not include a broad range of health-risk behaviors as covariates; 2) relied on clinical samples, which might lead to biased results; and 3) did not incorporate underweight individuals. Hence, this study aims to examine associations between BMI (from being underweight through obesity) and HRQOL in a population-based sample, while considering multiple health-risk behaviors (low physical activity, risky alcohol consumption, daily cigarette smoking, frequent cannabis use) as well as socio-demographic characteristics. METHODS: A total of 5 387 young Swiss men (mean age = 19.99; standard deviation = 1.24) of a cross-sectional population-based study were included. BMI was calculated (kg/m²) based on self-reported height and weight and divided into 'underweight' (<18.5), 'normal weight' (18.5-24.9), 'overweight' (25.0-29.9) and 'obese' (≥30.0). Mental and physical HRQOL was assessed via the SF-12v2. Self-reported information on physical activity, substance use (alcohol, cigarettes, and cannabis) and socio-demographic characteristics also was collected. Logistic regression analyses were conducted to study the associations between BMI categories and below average mental or physical HRQOL. Substance use variables and socio-demographic variables were used as covariates. RESULTS: Altogether, 76.3% were normal weight, whereas 3.3% were underweight, 16.5% overweight and 3.9% obese. Being overweight or obese was associated with reduced physical HRQOL (adjusted OR [95% CI] = 1.58 [1.18-2.13] and 2.45 [1.57-3.83], respectively), whereas being underweight predicted reduced mental HRQOL (adjusted OR [95% CI] = 1.49 [1.08-2.05]). Surprisingly, obesity decreased the likelihood of experiencing below average mental HRQOL (adjusted OR [95% CI] = 0.66 [0.46-0.94]). Besides BMI, expressed as a categorical variable, all health-risk behaviors and socio-demographic variables were associated with reduced physical and/or mental HRQOL. CONCLUSIONS: Deviations from normal weight are, even after controlling for important health-risk behaviors and socio-demographic characteristics, associated with compromised physical or mental HRQOL among young men. Hence, preventive programs should aim to preserve or re-establish normal weight. The self-appraised positive mental well-being of obese men noted here, which possibly reflects a response shift, might complicate such efforts.

Canonical correspondence analysis in social science research

The use of simple and multiple correspondence analysis is well-established in socialscience research for understanding relationships between two or more categorical variables.By contrast, canonical correspondence analysis, which is a correspondence analysis with linearrestrictions on the solution, has become one of the most popular multivariate techniques inecological research. Multivariate ecological data typically consist of frequencies of observedspecies across a set of sampling locations, as well as a set of observed environmental variablesat the same locations. In this context the principal dimensions of the biological variables aresought in a space that is constrained to be related to the environmental variables. Thisrestricted form of correspondence analysis has many uses in social science research as well,as is demonstrated in this paper. We first illustrate the result that canonical correspondenceanalysis of an indicator matrix, restricted to be related an external categorical variable, reducesto a simple correspondence analysis of a set of concatenated (or stacked ) tables. Then weshow how canonical correspondence analysis can be used to focus on, or partial out, aparticular set of response categories in sample survey data. For example, the method can beused to partial out the influence of missing responses, which usually dominate the results of amultiple correspondence analysis.

Pretherapeutic gamma-glutamyltransferase is an independent prognostic factor for patients with renal cell carcinoma.

BACKGROUND: Gamma-glutamyltransferase (GGT) regulates apoptotic balance and promotes cancer progression and invasion. Higher pretherapeutic GGT serum levels have been associated with worse outcomes in various malignancies, but there are no data for renal cell carcinoma (RCC). METHODS: Pretherapeutic GGT serum levels and clinicopathological parameters were retrospectively evaluated in 921 consecutive RCC patients treated with nephrectomy at a single institution between 1998 and 2013. Gamma-glutamyltransferase was analysed as continuous and categorical variable. Associations with RCC-specific survival were assessed with Cox proportional hazards models. Discrimination was measured with the C-index. Decision-curve analysis was used to evaluate the clinical net benefit. The median postoperative follow-up was 45 months. RESULTS: Median pretherapeutic serum GGT level was 25 U l(-1). Gamma-glutamyltransferase levels increased with advancing T (P<0.001), N (P=0.006) and M stages (P<0.001), higher grades (P<0.001), and presence of tumour necrosis (P<0.001). An increase of GGT by 10 U l(-1) was associated with an increase in the risk of death from RCC by 4% (HR 1.04, P<0.001). Based on recursive partitioning-based survival tree analysis, we defined four prognostic categories of GGT: normal low (<17.5 U l(-1)), normal high (17.5 to <34.5 U l(-1)), elevated (34.5 to <181.5 U l(-1)), and highly elevated (⩾181.5 U l(-1)). In multivariable analyses that adjusted for the effect of standard features, both continuously and categorically coded GGT were independent prognostic factors. Adding GGT to a model that included standard features increased the discrimination by 0.9% to 1.8% and improved the clinical net benefit. CONCLUSIONS: Pretherapeutic serum GGT is a novel and independent prognostic factor for patients with RCC. Stratifying patients into prognostic subgroups according to GGT may be used for patient counselling, tailoring surveillance, individualised treatment planning, and clinical trial design.

Prevalence of overweight and obesity in children from urban and rural Niagara

Purpose: The influence of environment in the development of overweight and obesity is an ongoing concern. This investigation examined the influence of urbanization on the rates of childhood overweight and obesity. Method: 2167 (1090M, 1077F) grade four children from 75 schools in Ontario's Niagara Region were sampled. A sophisticated algorithm overlaying electoral boundaries, population densities, and the knowledge of community members was used to classify schools into one of three location categories: urban {N= 1588), urban fringe {N= 379), and rural (A^= 234). Each subject was measured for: height, weight, and aerobic performance (Leger). Physical activity was evaluated with the self-report Participation Questionnaire (free-time and organized sport activities), and teacher's evaluations of student activity. Overweight (overweight and obesity combined) was measured both as a continuous (BMI) and categorical variable (BMI category), to evaluate the prevalence by location. A multivariate analysis was used to test for a suppression effect. Results: BMI and BMI category did not differ significantly by location or gender, and no evidence of a gender interaction existed. According to both a linear and logistic regression, physical activity or fitness levels did not suppress the influence of location on BMI and BMI category. Age, gender, free-time activity, organized sports, fitness level, and number of siblings, were all found to significantly influence overweight. Conclusions: It is plausible that the prevalence of overweight does not differ in urban and rural children from the Niagara Region. Further investigation is recommended, examining subjects by individual location of residence, in multiple regions throughout Ontario.

Mixing compositions and other scales

Theory of compositional data analysis is often focused on the composition only. However in practical applications we often treat a composition together with covariables with some other scale. This contribution systematically gathers and develop statistical tools for this situation. For instance, for the graphical display of the dependence of a composition with a categorical variable, a colored set of ternary diagrams might be a good idea for a first look at the data, but it will fast hide important aspects if the composition has many parts, or it takes extreme values. On the other hand colored scatterplots of ilr components could not be very instructive for the analyst, if the conventional, black-box ilr is used. Thinking on terms of the Euclidean structure of the simplex, we suggest to set up appropriate projections, which on one side show the compositional geometry and on the other side are still comprehensible by a non-expert analyst, readable for all locations and scales of the data. This is e.g. done by defining special balance displays with carefully- selected axes. Following this idea, we need to systematically ask how to display, explore, describe, and test the relation to complementary or explanatory data of categorical, real, ratio or again compositional scales. This contribution shows that it is sufficient to use some basic concepts and very few advanced tools from multivariate statistics (principal covariances, multivariate linear models, trellis or parallel plots, etc.) to build appropriate procedures for all these combinations of scales. This has some fundamental implications in their software implementation, and how might they be taught to analysts not already experts in multivariate analysis

Crossbred cow adoption and milk market participation in a multivariate count data framework

Cross-bred cow adoption is an important and potent policy variable precipitating subsistence household entry into emerging milk markets. This paper focuses on the problem of designing policies that encourage and sustain milkmarket expansion among a sample of subsistence households in the Ethiopian highlands. In this context it is desirable to measure households’ ‘proximity’ to market in terms of the level of deficiency of essential inputs. This problem is compounded by four factors. One is the existence of cross-bred cow numbers (count data) as an important, endogenous decision by the household; second is the lack of a multivariate generalization of the Poisson regression model; third is the censored nature of the milk sales data (sales from non-participating households are, essentially, censored at zero); and fourth is an important simultaneity that exists between the decision to adopt a cross-bred cow, the decision about how much milk to produce, the decision about how much milk to consume and the decision to market that milk which is produced but not consumed internally by the household. Routine application of Gibbs sampling and data augmentation overcome these problems in a relatively straightforward manner. We model the count data from two sites close to Addis Ababa in a latent, categorical-variable setting with known bin boundaries. The single-equation model is then extended to a multivariate system that accommodates the covariance between crossbred-cow adoption, milk-output, and milk-sales equations. The latent-variable procedure proves tractable in extension to the multivariate setting and provides important information for policy formation in emerging-market settings

Construct validity of a figure rating scale for Brazilian adolescents

Background: Figure rating scales were developed as a tool to determine body dissatisfaction in women, men, and children. However, it lacks in the literature the validation of the scale for body silhouettes previously adapted. We aimed to obtain evidence for construct validity of a figure rating scale for Brazilian adolescents.Methods: The study was carried out with adolescent students attending three public schools in an urban region of the municipality of Florianopolis in the State of Santa Catarina (SC). The sample comprised 232 10-19-year-old students, 106 of whom are boys and 126 girls, from the 5th series (i.e. year) of Primary School to the 3rd year of Secondary School. Data-gathering involved the application of an instrument containing 8 body figure drawings representing a range of children's and adolescents' body shapes, ranging from very slim (contour 1) to obese (contour 8). Weights and heights were also collected, and body mass index (BMI) was calculated later. BMI was analyzed as a continuous variable, using z-scores, and as a dichotomous categorical variable, representing a diagnosis of nutritional status (normal and overweight including obesity).Results: Results showed that both males and females with larger BMI z-scores chose larger body contours. Girls with higher BMI z-scores also show higher values of body image dissatisfaction.Conclusion: We provided the first evidence of validity for a figure rating scale for Brazilian adolescents.

Modified radical mastectomy: A pilot clinical trial comparing the use of conventional electric scalpel and harmonic scalpel

Background: The aim of this study was to compare the rates of local postoperative complications among women undergoing modified radical mastectomy with an electric scalpel (ES) or a harmonic scalpel (HS). It is thought that HS use has less postoperative complications, mainly seroma formation. Methods: This study was a prospective non-randomised clinical trial (NCT01391988) among consecutive patients, performed in parallel. Patients underwent modified radical mastectomy using an HS or ES. We analysed the following operative variables: time, blood loss and seroma volume drainage. Postoperative complications, including seroma, flap necrosis, haematoma and infection were evaluated on the 7th and 14th days. Results: Forty-six patients underwent a MRM with ES and 49 with HS; no differences were observed between the groups. The rate of local complications was 29% in the HS group and 52% in the ES group (p=0.024). The rates of seroma (16.3% versus 28.3%; p=0.161), necrosis (4.1% vs. 21.7%; p=0.013; OR=0.15), haematoma (2.0% vs. 8.7%; p=0.195) and infection (2.0% vs. 6.5%; p=0.351) were lower in the HS group. Adding the findings of all comparative studies using HSs in MRM to the seroma rates in the current study, the seroma rate, expressed as a categorical variable, did not decrease with HS. Seroma was present in 60/219 cases using an HS and in 69/239 cases utilising an ES (p=0.72). Based on a multivariate analysis, HS decreased the risk of skin necrosis (p=0.015). Conclusions: HSs do not decrease the seroma rate. However, this method may be useful in skin sparing mastectomy because it decreases skin flap necrosis. © 2013 Surgical Associates Ltd.