Categorical Explanatory Variables Research Articles

Syphilis among HIV-positive men who have sex with men in Ghana: The 2023 biobehavioral survey.

Apart from HIV acquisition, men who have sex with men are at a higher risk of sexually transmitted infections, especially syphilis. Although the intersection between syphilis and HIV poses a formidable challenge among key populations who are known to be vulnerable to many health threats, there is little known about syphilis infection among MSM living with HIV in Ghana. This study seeks to investigate the burden of syphilis and address the existing knowledge gap by exploring behavioral, healthcare access, and structural factors influencing the syphilis burden within the HIV-positive MSM population. This study was conducted in 2023 as part of the bio-behavioral survey (BBS) among men who have sex with men (MSM) in Ghana. A cross-sectional survey that used a respondent-driven sampling (RDS) approach was conducted in the old ten regions of Ghana. Data was collected on 3,420 participants, however, 857 HIV-positive MSM were included in this study since it focused on syphilis among HIV-positive MSM in Ghana. The study estimated the prevalence of syphilis among MSM living with HIV and provided a 95% confidence interval across different categories of explanatory variables. Bivariate and multivariable logistic regression models were used to identify factors associated with overall syphilis prevalence. All other analyses were weighted due to the complex design of the study. The overall prevalence of syphilis was 23.83% (95% CI: 20.44, 27.58). HIV-positive men who only had sex with men had a 29.77% (95% CI: 23.90, 36.40) prevalence of syphilis compared to a prevalence of 9.50% (95% CI: 2.56, 29.53) recorded by HIV-positive MSM who were attracted to mostly females. Participants who ever had receptive anal sex recorded a higher prevalence 26.79% (95% CI: 22.78, 31.23) than those who never had receptive anal sex 12.86% (95% CI: 8.02, 19.99). The odds of syphilis among HIV-positive MSM who never used condoms with regular sex partners in the last 6 months were 41.08 (aOR: 41.08 at 95% CI 1.24-136.42; p = 0.038). There is a high prevalence of syphilis among HIV-positive men who have sex with men in Ghana, especially among those who engage in unprotected anal sex and have multiple sexual partners. Our findings underscore the dire need for targeted interventions to address the dual brunt of HIV and syphilis among the MSM population in Ghana.

A high‐dimensional multinomial logit model

SummaryThe number of parameters in a standard multinomial logit model increases linearly with the number of choice alternatives and number of explanatory variables. Because many modern applications involve large choice sets with categorical explanatory variables, which enter the model as large sets of binary dummies, the number of parameters in a multinomial logit model is often large. This paper proposes a new method for data‐driven two‐way parameter clustering over outcome categories and explanatory dummy categories in a multinomial logit model. A Bayesian Dirichlet process mixture model encourages parameters to cluster over the categories, which reduces the number of unique model parameters and provides interpretable clusters of categories. In an empirical application, we estimate the holiday preferences of 11 household types over 49 holiday destinations and identify a small number of household segments with different preferences across clusters of holiday destinations.

Machine learning predictions of code-based seismic vulnerability for reinforced concrete and masonry buildings: Insights from a 300-building database

This paper proposes a data-driven model for predicting the code-based seismic vulnerability index calibrated on a dataset comprising almost 300 buildings. The vulnerability index, estimated following the Italian Seismic Code, involved rigorous investigations, including geometric surveys, experimental tests, and numerical modelling. Leveraging data from these investigations, encompassing approximately 15 categorical and numerical explanatory variables, the authors developed several regression and classification predictive models, such as Logistic Regression and Artificial Neural Networks (ANN). The optimal models perform binary classification to determine the categorization into two macro-classes, defined by an arbitrary vulnerability threshold. The ANN model stands out as the best performer. When adjusting the vulnerability threshold to obtain a balanced dataset, such a model achieves an accuracy higher than 85%. The paper also discusses the importance of each feature by calculating the SHapley Additive exPlanations (SHAP) values. The proposed model can aid decision-makers in allocating resources effectively to mitigate seismic risks of built environments.

One-step closed-form estimator for generalized linear model with categorical explanatory variables

One-step closed-form estimator for generalized linear model with categorical explanatory variables

Specific and non-specific effects of Mycobacterium bovis BCG vaccination in dairy calves.

Bovine tuberculosis (bTB) is a chronic disease mainly caused by Mycobacterium bovis, a zoonotic pathogen with economic significance as it leads to reduced milk and meat production, and high costs for control measures. The Bacillus Calmette-Guérin (BCG) vaccine, primarily used to prevent tuberculosis in humans, has also been studied for controlling bTB. While showing effectiveness in preventing M. bovis infection and disease in cattle, the BCG vaccine can induce non-specific effects on the immune system, enhancing responses to infections caused by unrelated pathogens, and also having non-specific effects on lactation. The aim of this study is to describe both the specific and non-specific effects of BCG vaccination in calves from a commercial dairy herd in central Chile. Diagnosis of M. bovis infection was performed through the IFNγ release assay (IGRA) using ESAT6/CFP-10 and Rv3615c antigens. The records of milk production, somatic cell count (SCC), clinical mastitis (CM) and retained placenta (RP) during the first lactation were compared between vaccinated and non-vaccinated animals. The breed (Holstein Friesian [HF] v/s HF × Swedish Red crossbred [HFSR]) and the season (warm v/s cold) were also analyzed as categorical explanatory variables. Results of IGRA showed significant differences between vaccinated and control groups, indicating a vaccine efficacy of 58.5% at 18 months post vaccination in HFSR crossbred animals. Although milk production did not vary, SCC and CM showed differences between groups, associated to the breed and the season, respectively. When analyzing CM and RP as a whole entity of disease, BCG showed protection in all but the cold season variables. Overall, the BCG vaccine induced protective specific and non-specific effects on health parameters, which may be influenced by the breed of animals and the season. These results provide new features of BCG protection, supporting initiatives for its implementation as a complementary tool in bTB control.

Survival status and predictors of mortality among adult Stroke patients admitted to Jimma University Medical Center, South west Ethiopia: A retrospective Cohort study.

Stroke is the leading cause of cardiovascular disease death in sub-Saharan Africa and the second leading cause of mortality worldwide. In 2016, 6.23% of all fatalities in Ethiopia were stroke-related. To assess survival status and predictors of mortality among adult stroke patients admitted to Jimma University Medical Center from April 1/2017 to March 31/2022. A retrospective cohort study was conducted on 480 adult stroke patients selected by simple random sampling from patients admitted to the Jimma University Medical Center Stroke Unit from April 1, 2017 to March 31, 2022. Data were extracted from May to June 2022 and entered Epi-data v.3.1 and analyzed by R v.4.2. The Kaplan-Meier curve with Log rank test was used to estimate survival time and to compare survival experience between categories of explanatory variables. The Cox regression model was computed to identify predictors of survival status in stroke patients. Then the 95% CI of the hazard ratio was set with corresponding p-value < 0.05 to declare statistical significance. During 4350 person-days of follow-up; 88 (18.33%) patients died; resulting in an incidence mortality of 20.23 per 1000 person-days, with a median survival time of 38 days. Glasgow coma score <8 on admission (AHR = 7.71; 95% CI: 3.78, 15.69), dyslipidemia (AHR = 3.96; 95% CI: 2.04, 7.69), aspiration pneumonia (AHR 2.30; 95% CI: 1.23-4.26), and increased intracranial pressure (AHR = 4.27; 95% CI: 2.33, 7.81), were the independent predictors of the time until death. The incidence of stroke mortality was higher at the seven and fourteen days. Glasgow Coma Scale, increased intracranial pressure, dyslipidemia, and aspiration pneumonia were independent predictors of mortality.

Combining discrete choice models and neural networks through embeddings: Formulation, interpretability and performance

This paper proposes a novel approach that combines theory- and data-driven choice models using Artificial Neural Networks (ANNs). In particular, we use continuous vector representations, called embeddings, for encoding categorical or discrete explanatory variables with a special focus on interpretability and model transparency. Although embedding representations within the logit framework have been conceptualized by Pereira (2019), their dimensions do not have an absolute definitive meaning, hence offering limited behavioral insights in this earlier work. The novelty of our work lies in enforcing interpretability to the embedding vectors by formally associating each of their dimensions to a choice alternative. Thus, our approach brings benefits much beyond a simple parsimonious representation improvement over dummy encoding, as it provides behaviorally meaningful outputs that can be used in travel demand analysis and policy decisions. Additionally, in contrast to previously suggested ANN-based Discrete Choice Models (DCMs) that either sacrifice interpretability for performance or are only partially interpretable, our main model (E-MNL) preserves interpretability of the utility coefficients for all the input variables despite being based on ANN principles. The proposed models are tested on two real world datasets and evaluated against benchmark and baseline models that use dummy-encoding. The results of the experiments indicate that our models deliver state-of-the-art predictive performance, outperforming existing ANN-based models while drastically reducing the number of required network parameters.

Association Between Alcohol Consumption and Cardiovascular Risk Based on Data from the Three Generations for Health Program in Hungary.

BACKGROUND Alcohol consumption has a significant effect on cardiovascular health, and risk factors, such as excessive alcohol use, should be avoided. Although alcohol consumption has decreased over the last decade in Hungary, it is still significantly higher than the average across the European Union. The objective of this study was to describe the patterns of alcohol use based on the Alcohol Use Disorders Identification Test (AUDIT), with a special focus on cardiovascular risk status (low, moderate, high, or very high). MATERIAL AND METHODS The Three Generations for Health program focuses on the development of primary health care in Hungary. One of the key elements of the program is the identification of risk factors of cardiovascular diseases. An ordinal multiple logistic regression analysis was performed with 10 categorical explanatory variables and the outcome was the categorical cardiovascular risk. RESULTS The database consisted of patients aged 40-65 years with a sample size of 11 348. A significant relationship was found between alcohol consumption and cardiovascular risk status; people with high-risk drinking patterns had higher a value of odds ratio (OR=1.306 [1.003-1.701]) for having a more serious cardiovascular status. According to multiple regression analysis, alcohol dependence was associated with cardiovascular risk. CONCLUSIONS From a public health perspective, the results highlight the importance of reducing alcohol consumption with the help of primary care and preventive services in countries with a high cardiovascular risk profile to reduce the cardiovascular disease-related burden.

The Influence of Bank Performance, Market Condition and Economic Growth on Non-Performing Loans

The paper contributes to the literature on the determinants of non-performing loans by applying panel fixed effects and dynamic Generalised Method of Moments (GMM) estimations to a relatively large panel of 80 countries over the period 1999-2017. The paper considers three categories of explanatory variables; bank-level, industry-level, and macroeconomic-level. The bank-level variables highlight the relevance of bank profitability and efficiency to avoid non-performing loans. The industry-level variables, for the entire period and particularly after the onset of the Global Financial Crisis (years 2009-2017), show that bank market concentration promoted non-performing loans, while bank market competition and bank stability did not contribute to increasing non-performing loans. The macroeconomic-level variable, real per capita gross domestic product, provides very convincing evidence that promoting economic growth looks like the best way to avoid non-performing loans. Analyses with different sub-samples show that the determinants of non-performing loans are not particularly dependent on the level of income of the country.

Productivity Prediction of Garment Employees using Multiple Linear Regression

This paper presents a multiple linear curve fitting method for finding how the change of manipulated variables affect the final result using productivity prediction of garment employees data set. To perform fitting, a function is defined, which depends on the parameters that measures the closeness between the data and model. The simplest form of modelling is linear regression, which is the prediction of one variable from another. When the relationship between a few variables are assumed to be linear, then multiple linear regression will be used to model the relationship between a continuous response variable and continuous or categorical explanatory variables.

On the exploration of regression dependence structures in multidimensional contingency tables with ordinal response variables

In this paper, we propose a new data-driven method to explore complex regression dependence structures in a multi-dimensional contingency table with an ordinal response variable and categorical (ordinal or nominal) explanatory variables. The proposed method is based on a sequential decomposition of the overall regression dependence for the data quantified by the checkerboard copula regression association measure (Wei and Kim, 2021) in an informative and interpretable fashion. It can measure the marginal and conditional contributions of any subset of all available explanatory variables to the overall regression association in a hierarchical manner taking into account the order of the explanatory variables. Thus, the proposed method enables a holistic description of various aspects of regression association in a multivariate contingency table, including marginal and conditional associations between an ordinal response variable and a subset of explanatory variables of interest. We investigate theoretical properties of the proposed decomposition method, and we further illustrate its performance through simulation and two real data examples, one from a randomized controlled trial and the other from a longitudinal epidemiological study.

A Cross-Sectional Study of Professional Nurses’ Knowledge, Attitudes, and Practices Regarding Organ Donation in Critical Care Units of Public and Private Hospitals in the Eastern Cape, South Africa

Globally, there is an overwhelming increase in the number of patients waiting for donated organs for transplantation, with a gross shortage of available organs. Lack of clear practice guidelines and the knowledge and attitudes of health care providers were hypothesized as possible reasons. We aimed to determine the attitudes, level of knowledge, and practices of professional nurses working in critical care units in public and private hospitals in Eastern Cape Province regarding organ donation. The study used a quantitative, non-experimental, descriptive design to investigate the current knowledge, attitude, and practice of organ donation in critical care among 108 professional nurses working in public and private critical care units in Eastern cape. Data were collected between 26 February 2017 until 27 June 2017 using anonymous, self-administered, pretested questionnaires. The means of knowledge, and practice scores were estimated among participants, and their associated categorical explanatory variables were ascertained. A total of 108 nurses participated in the study. Of these, 94 (87.0%) were female, 78 (72.2%) were black, 104 (96.3%) were Christians, 79 (73.2%) worked in an ICU, 79 (73.2%) had a diploma qualification, and 67 (62.0%) worked in a tertiary hospital. About 67% of the respondents had good knowledge, 53% had a positive attitude, and 50.4% had poor practice readiness toward organ donation. Working in renal units (p < 0.001) and practicing in tertiary hospitals (p < 0.001) were significantly associated with a high organ donation knowledge score while being a female nurse (p = 0.036), working in renal units (p < 0.001), and practicing in tertiary hospitals (p < 0.001) were significantly associated with a high organ donation practice score. Differences in organ donation knowledge and practices were noted between the different levels of health care services as the tertiary level outperformed the secondary level institutions. Nurses play a major role in critical and end-of-life care and are closer to patients and relatives. Hence, pre- and in-service education and promotional campaigns among nurses at all levels of care would be a strategic step to scale availability of donated organs and would meet the needs of thousands of individuals who need them to survive.

Determinants of Adapted Improved Sanitation in Rural and Urban of Bangladesh and Pakistan

This study focuses on the socioeconomic and demographic determinants of improved sanitation adaptation in South Asia, mainly in Bangladesh and Pakistan. For the purpose of mass population study, cross-sectional datasets were obtained for both Bangladesh and Pakistan. The survey is based on a two-stage stratified sample of households. This study hypothesized the status of sanitation into three categorically distributed dependent variables which are “improved sanitation facility”, “unimproved sanitation facility” and “no sanitation facility”. According to the World Health Organization, these variables are supported to categorize the situation of sanitation by the Joint Monitoring Program categorization of toilet facilities. The “improved sanitation facilities” includes Flush or pour flushed toilets connected to sewers, Flush or pour flushed toilets or latrines connected to pits or septic tanks, and ventilated improved pit (VIP) latrines, Pit latrines with slab, Composting toilets including twin-pit latrines and container-based systems. The “unimproved sanitation facilities” includes Pit latrines without slab, Bucket latrines and Hanging latrines. And the “no facility” category includes open defecation (bush/field/desert/road/beach etc.). Wealth Index of the household, Gender of head of the household, marital status of the respondent, Education of household head, Locality of the household, and Province/ division are the categorical explanatory variables used in this study; family size of the household and age of the head of the household are two numerical explanatory variables that are taken as important demographic factors. Two candidate models were constructed for predicting no facility, unimproved sanitation, and improved sanitation respectively. In the sample, only 1.26% of households of Bangladesh do not have any toilet facility while Pakistan has 10.1% households without any toilet facility. The indicators, all three component models, and two data sets demonstrate agreement in the analysis. There is no discernible upward trend in model performance year after year. All of the comparison results show that the ordinal logistic model outperforms the multinomial regression model. According to the AIC, BIC, and log probability statistics, the highest education of family members is a more relevant element in a household's avoiding status in Pakistan. The indicators, all three component models, and two data sets demonstrate agreement in the analysis. There is no discernible upward trend in model performance year after year. All of the comparison results show that the ordinal logistic model outperforms the multinomial regression model.

A closed-form alternative estimator for GLM with categorical explanatory variables

The parameters of generalized linear models (GLMs) are usually estimated by the maximum likelihood estimator (MLE) which is known to be asymptotically efficient. But the MLE is computed using a Newton-Raphson-type algorithm which is time-consuming for a large number of variables or modalities, or a large sample size. An alternative closed-form estimator is proposed in this paper in the case of categorical explanatory variables. Asymptotic properties of the alternative estimator is studied. The performances in terms of both computation time and asymptotic variance of the proposed estimator are compared with the MLE for a Gamma distributed GLM.

What drives credit expansion worldwide?——An empirical investigation with long-term cross-country panel data

This paper examines a broad range of driving factors of credit expansion using long-term cross-country panel data from 1994 to 2018 and the System GMM method. We discuss the role of macroeconomic fundamentals in explaining credit expansion by examining five categories of explanatory variables. Some structural patterns of weak linkages of credit expansion with macroeconomic fundamentals are identified. We further explore the possible causes that follow the theoretical logic of risk-taking in the private and public sectors. Controlling for the influence of macroeconomic fundamentals, we find the effects of various financial system characteristics, such as institutional features of the financial system, financial institutions' performance, asset markets booms, and macroeconomic and financial regulatory environment, on inducing credit expansion over the needs supported by macroeconomic fundamentals. Our study provides strong implications on regulatory policies on the rising credit during the COVID-19 Pandemic.

Model averaging with the hybrid model: An asymptotic study and demonstration.

In this paper, we present a new model averaging technique that can be applied in medical research. The dataset is first partitioned by the values of its categorical explanatory variables. Then for each partition, a model average is determined by minimising some form of squared errors, which could be the leave-one-out cross-validation errors. From our asymptotic optimality study and the results of simulations, we demonstrate under several high-level assumptions and modelling conditions that this model averaging procedure may outperform jackknife model averaging, which is a well-established technique. We also present an example where a cross-validation procedure does not work (that is, a zero-valued cross-validation error is obtained) when determining the weights for model averaging.

Factors associated with Pharmaceutical Benefits Advisory Committee decisions for listing medicines for diabetes and its associated complications.

Objective To retrospectively analyse the key factors associated with listing decisions by the Pharmaceutical Benefits Advisory Committee (PBAC) for medicines for diabetes and its complications on the Pharmaceutical Benefits Scheme. Methods The clinical and economic evidence were retrieved from public summary documents (PSD) of all major submissions between July 2005 and March 2020. A multivariate binary logit regression analysis was conducted to assess the relationship between the categorical explanatory variables and PBAC recommendations. Results We identified a total of 211 PSD of which 118 (56%) were recommended for listing. Clinical and economic uncertainty were significantly and inversely associated with the PBAC recommendation. Submissions with high clinical and economic uncertainty were less likely to be recommended. Conclusion Our findings will enhance the understanding of medical professionals, pharmaceutical companies, and other stakeholders about the rationale of PBAC reimbursement decisions for these medicines and assist prospective applicant sponsor companies in preparing their submissions.

Reversal of hypertensive heart disease: a multiple linear regression model.

Background: The development of left ventricular hypertrophy in primary hypertension increases cardiovascular mortality and morbidity. Reversal of left ventricular hypertrophy through therapeutic control of blood pressure reduces the risk of adverse cardiovascular incidents. Objective: In our study, we explored for the determinants of left ventricular hypertrophy regression. Methods: A cohort (n=217) of patients with hypertensive left ventricular hypertrophy was identified by screening consecutive patients in medical outpatient unit. The primary inclusion criteria were (i) Blood pressure more than140/90 mm of Hg (ii) Left Ventricular Mass Index more than 115 and 95 gm/m2 for males and females respectively. Left Ventricular Mass Index was determined by echocardiography at the time of recruitment and after 24 weeks of standard pharmacotherapy. The change in Left Ventricular Mass Index was modelled using multiple linear regression with both categorical and continuous explanatory variables. The effect of drug therapy on change in Left Ventricular Mass Index was tested in the model with dummy coded variables for the treatment categories. Results: In simple linear regression, the variables significantly correlating with change in Left Ventricular Mass Index were baseline Left Ventricular Mass Index (r=0.62, p<0.001), change in systolic blood pressure (r=0.22, p=0.001), change in mean blood pressure (r=0.16, p=0.02), baseline systolic blood pressure (r=0.15, p=0.02), age (r=0.12, p=0.09) and diabetes (r=0.12, p=0.09). The best fit model (r2=0.408) retained baseline Left Ventricular Mass Index (β=0.59, p<0.001), change in systolic blood pressure (β=0.14, p=0.01) and diabetes (β=-0.104, p=0.05) as the significant predictors. Introduction of treatment effect into the model non-significantly increased the fit of the model (r2=0.414, p=0.27-0.98). Conclusions: Pre-treatment Left Ventricular Mass Index and reduction in systolic blood pressure were the major determinants of left ventricular hypertrophy regression. We also observed that there is lesser left ventricular hypertrophy regression in diabetic patients, warranting future research to explore glycaemic control as a modifiable factor in left ventricular hypertrophy reversal.

Partial dependence through stratification

Partial dependence curves (FPD) are commonly used to explain feature importance once a supervised learning model has been fitted to data. However, it is common for the same partial dependence algorithm to give meaningfully different curves for different supervised models, even when the algorithm is applied to the same data. As a result, it is difficult to distinguish between model artifacts and true relationships in the data. In this paper, we contribute metods for computing partial dependence curves, for both numerical (StratPD) and categorical explanatory variables (CatStratPD), that work directly from training data rather than the predictions of a fitted model. Our methods provide a direct estimate of partial dependence, and rely on approximating the partial derivative of an unknown regression function. We investigate settings where contemporary partial dependence methods – including FPD, Accumulated Local Effects (ALE), and SHapley Additive exPlanations (SHAP) methods – give biased results. We demonstrate that our approach works correctly on synthetic data and plausibly on real data sets. This work motivates a new line of inquiry into nonparametric partial dependence that provides robust information about the variables considered in a supervised learning task.

On exploratory analytic method for multi-way contingency tables with an ordinal response variable and categorical explanatory variables

In this paper, we propose a new model-free exploratory method for descriptive modeling that identifies and measures the regression dependence between an ordinal response variable and categorical (ordinal or nominal) explanatory variables in a multi-way contingency table. The proposed methodology consists of three parts, checkerboard copula score, checkerboard copula regression, and checkerboard copula association measure. The checkerboard copula score is a new type of score for ordinal variables that preserves the natural ordering of the categorical scale and it will be exploited for developing the methods measuring the association between the variables of interest. The checkerboard copula regression identifies the regression dependence between an ordinal response variable and categorical explanatory variables. It enables delineating the identified dependence in an exploratory manner. The checkerboard copula association measure quantifies the strength of the dependence identified by the checkerboard copula regression. We investigate the properties of checkerboard copula scores, checkerboard copula regression, its association measure, and their estimators. Finally, the performance of the proposed method is illustrated with simulation and real data.