A method for determining groups in cumulative incidence curves in competing risk data.

The cumulative incidence function is the standard method for estimating the marginal probability of a given event in the presence of competing risks. One basic but important goal in the analysis of competing risk data is the comparison of these curves, for which limited literature exists. We proposed a new procedure that lets us not only test the equality of these curves but also group them if they are not equal. The proposed method allows determining the composition of the groups as well as an automatic selection of their number. Simulation studies show the good numerical behavior of the proposed methods for finite sample size. The applicability of the proposed method is illustrated using real data.

A method for determining groups in nonparametric regression curves: Application to prefrontal cortex neural activity analysis.

Generalized additive models provide a flexible and easily-interpretable method for uncovering a nonlinear relationship between response and covariates. In many situations, the effect of a continuous covariate on the response varies across groups defined by the levels of a categorical variable. When confronted with a considerable number of groups defined by the levels of the categorical variable and a factor-by-curve interaction is detected in the model, it then becomes important to compare these regression curves. When the null hypothesis of equality of curves is rejected, leading to the clear conclusion that at least one curve is different, we may assume that individuals can be grouped into a number of classes whose members all share the same regression function. We propose a method that allows determining such groups with an automatic selection of their number by means of bootstrapping. The validity and behavior of the proposed method were evaluated through simulation studies. The applicability of the proposed method is illustrated using real data from an experimental study in neurology.

Effectiveness and safety of tetanus vaccine administration by intramuscular vs. subcutaneous route in anticoagulated patients: Randomized clinical trial in primary care.

Design: Prospective, double-blind clinical trial comparing tetanus-diphtheria vaccine administration routes, intramuscular (IM) vs. subcutaneous (SC) injection, in patients with oral anticoagulants. ISRCTN69942081. Study population: Patients treated with oral anticoagulants, 15 health centers, Vigo (Spain). Sample size, 117 in each group. Outcome variables: Safety analysis: systemic reactions and, at the vaccine administration site, erythematic, swelling, hematoma, granuloma, pain.Effectiveness analysis: differences in tetanus toxoid antibody titers.Independent variables: route, sex, age, baseline serology, number of doses administered. Analysis: Following the CONSORT guidelines, we performed an intention-to-treat analysis. We conducted a descriptive study of the variables included in both groups (117 in each group) and a bivariate analysis. Fewer than 5% of missing values. Imputation in baseline and final serology with the median was performed. Lost values were assumed to be values missing at random. We conducted a descriptive study of the variables and compared routes. For safety, multivariate logistic regression was applied, with each safety criterion as outcome and the independent variables. Odds ratios (ORs) were calculated. For effectiveness, a generalized additive mixed model, with the difference between final and initial antibody titers as outcome. Due to the bimodal distribution of the outcome, the normal mixture fitting with gamlssMX was used. All statistical analyses were performed with the gamlss.mx and texreg packages of the R free software environment. Results: A previously published protocol was used across the 6-year study period. The breakdown by sex and route showed: 102 women and 132 men; and 117 IM and 117 SC, with one dose administered in over 80% of participants. There were no differences between groups in any independent variable. The second and third doses administered were not analyzed, due to the low number of cases. In terms of safety, there were no severe general reactions. Locally, significant adjusted differences were observed: in pain, by sex (male, OR: 0.39) and route (SC, OR: 0.55); in erythema, by sex (male, OR: 0.34) and route (SC, OR: 5.21); and in swelling, by sex (male, OR: 0.37) and route (SC, OR: 2.75). In terms of effectiveness, the model selected was the one adjusted for baseline serology.

Impact of New Systemic Therapies in Overall Survival in Non-Metastatic Castration Resistant Prostate Cancer: Systematic Review and Meta-Analysis.

There was a high medical need for patients with non-metastatic castration-resistant prostate cancer (nmCRPC) when several next-generation anti-androgens (apalutamide, enzalutamide, and darolutamide) demonstrated clinically relevant delays in metastasis onset. However, to date, few publications have assessed the pooled effect of these treatments on overall survival (OS). We performed a systematic review and meta-analysis of all randomized, placebo-controlled studies investigating a systemic treatment in nmCRPC. Publications were identified by searching several databases on April 7, 2021. The primary objective of this analysis was to determine the OS benefit. Secondary outcomes included the relative risk (RR) of adverse events (AEs) and grade 3-4 AEs. A sensitivity analysis with simulated data was also conducted to examine the influence of the study designs on the results. Three randomized controlled studies (SPARTAN, PROSPER, ARAMIS) met our inclusion criteria. Pooled meta-analyses showed a significant benefit in OS with the active agents versus placebo (hazard ratio [HR] 0.74, 95% confidence interval [CI] 0.65-0.83), as well as increased risk of any grade (RR 1.09, 95% CI 1.01-1.17) and grade 3-4 AEs (RR 1.50, 95% CI 1.23-1.83). The sensitivity analysis with SPARTAN-like simulated populations demonstrated that when using ARAMIS statistical design, OS would be statistically significant in 98.1% of the cases, at a shorter follow-up and with lower number of events. First-line treatment of nmCRPC patients with anti-androgens increased OS with an acceptable safety profile. In light of the different study designs and follow-up, results should be interpreted separately.

Analysis of the impact of social determinants and primary care morbidity on population health outcomes by combining big data: A research protocol.

Background: In recent years, different tools have been developed to facilitate analysis of social determinants of health (SDH) and apply this to health policy. The possibility of generating predictive models of health outcomes which combine a wide range of socioeconomic indicators with health problems is an approach that is receiving increasing attention. Our objectives are twofold: (1) to predict population health outcomes measured as hospital morbidity, taking primary care (PC) morbidity adjusted for SDH as predictors; and (2) to analyze the geographic variability of the impact of SDH-adjusted PC morbidity on hospital morbidity, by combining data sourced from electronic health records and selected operations of the National Statistics Institute (Instituto Nacional de Estadística/INE). Methods: The following will be conducted: a qualitative study to select socio-health indicators using RAND methodology in accordance with SDH frameworks, based on indicators published by the INE in selected operations; and a quantitative study combining two large databases drawn from different Spain's Autonomous Regions (ARs) to enable hospital morbidity to be ascertained, i.e., PC electronic health records and the minimum basic data set (MBDS) for hospital discharges. These will be linked to socioeconomic indicators, previously selected by geographic unit. The outcome variable will be hospital morbidity, and the independent variables will be age, sex, PC morbidity, geographic unit, and socioeconomic indicators. Analysis: To achieve the first objective, predictive models will be used, with a test-and-training technique, fitting multiple logistic regression models. In the analysis of geographic variability, penalized mixed models will be used, with geographic units considered as random effects and independent predictors as fixed effects. Discussion: This study seeks to show the relationship between SDH and population health, and the geographic differences determined by such determinants. The main limitations are posed by the collection of data for healthcare as opposed to research purposes, and the time lag between collection and publication of data, sampling errors and missing data in registries and surveys. The main strength lies in the project's multidisciplinary nature (family medicine, pediatrics, public health, nursing, psychology, engineering, geography).

Modeling conditional reference regions: Application to glycemic markers.

Many clinical decisions are taken based on the results of continuous diagnostic tests. Usually, only the results of one single test is taken into consideration, the interpretation of which requires a reference range for the healthy population. However, the use of two different tests, can be necessary in the diagnosis of certain diseases. This obliges a bivariate reference region be available for their interpretation. It should also be remembered that reference regions may depend on patient variables (eg, age and sex) independent of the suspected disease. However, few proposals have been made regarding the statistical modeling of such reference regions, and those put forward have always assumed a Gaussian distribution, which can be rather restrictive. The present work describes a new statistical method that allows such reference regions to be estimated with no insistence on the results being normally distributed. The proposed method is based on a bivariate location-scale model that provides probabilistic regions covering a specific percentage of the bivariate data, dependent on certain covariates. The reference region is estimated nonparametrically and the nonlinear effects of continuous covariates via polynomial kernel smoothers in additive models. The bivariate model is estimated using a backfitting algorithm, and the optimal smoothing parameters of the kernel smoothers selected by cross-validation. The model performed satisfactorily in simulation studies under the assumption of non-Gaussian conditions. Finally, the proposed methodology was found to be useful in estimating a reference region for two continuous diagnostic tests for diabetes (fasting plasma glucose and glycated hemoglobin), taking into account the age of the patient.

Selecting the number of categories of the lymph node ratio in cancer research: A bootstrap-based hypothesis test.

The high impact of the lymph node ratio as a prognostic factor is widely established in colorectal cancer, and is being used as a categorized predictor variable in several studies. However, the cut-off points as well as the number of categories considered differ considerably in the literature. Motivated by the need to obtain the best categorization of the lymph node ratio as a predictor of mortality in colorectal cancer patients, we propose a method to select the best number of categories for a continuous variable in a logistic regression framework. Thus, to this end, we propose a bootstrap-based hypothesis test, together with a new estimation algorithm for the optimal location of the cut-off points called BackAddFor, which is an updated version of the previously proposed AddFor algorithm. The performance of the hypothesis test was evaluated by means of a simulation study, under different scenarios, yielding type I errors close to the nominal errors and good power values whenever a meaningful difference in terms of prediction ability existed. Finally, the methodology proposed was applied to the CCR-CARESS study where the lymph node ratio was included as a predictor of five-year mortality, resulting in the selection of three categories.

Evaluating Lightning-Caused Fire Occurrence Using Spatial Generalized Additive Models: A Case Study in Central Spain.

It is widely accepted that the relationship between lightning wildfire occurrence and its influencing factors vary depending on the spatial scale of analysis, making the development of models at the regional scale advisable. In this study, we analyze the effects of different biophysical variables and lightning characteristics on lightning-caused forest wildfires in Castilla y León region (Central Spain). The presence/absence of at least one lightning-caused fire in any 4 × 4-km grid cell was used as a dependent variable and vegetation type and structure, terrain, climate, and lightning characteristics were used as possible covariates. Five prediction methods were compared: a generalized linear model (GLM), a random forest model (RFM), a generalized additive model (GAM), a GAM that includes a spatial trend function (GAMs) and a spatial autoregressive model (AUREG). A GAMs with just one covariate, apart from longitude and latitude for each observation included as a combined effect, was considered the most appropriate model in terms of both predictive ability and simplicity. According to our results, the probability of a forest being affected by a lightning-caused fire is positively and nonlinearly associated with the percentage of coniferous woodlands in the landscape, suggesting that occurrence is more closely associated with vegetation type than with topography, climate, or lightning characteristics. The selected GAMs is intended to inform the Regional Government of Castilla y León (the fire and fuel agency in the region) regarding identification of areas at greatest risk so it can design long-term forest fuel and fire management strategies.

Multiscale Supervised Classification of Point Clouds with Urban and Forest Applications.

We analyze the utility of multiscale supervised classification algorithms for object detection and extraction from laser scanning or photogrammetric point clouds. Only the geometric information (the point coordinates) was considered, thus making the method independent of the systems used to collect the data. A maximum of five features (input variables) was used, four of them related to the eigenvalues obtained from a principal component analysis (PCA). PCA was carried out at six scales, defined by the diameter of a sphere around each observation. Four multiclass supervised classification models were tested (linear discriminant analysis, logistic regression, support vector machines, and random forest) in two different scenarios, urban and forest, formed by artificial and natural objects, respectively. The results obtained were accurate (overall accuracy over 80% for the urban dataset, and over 93% for the forest dataset), in the range of the best results found in the literature, regardless of the classification method. For both datasets, the random forest algorithm provided the best solution/results when discrimination capacity, computing time, and the ability to estimate the relative importance of each variable are considered together.

Utilidad diagnóstica del cuestionario STOP-Bang en la apnea del sueño moderada en atención primaria / Diagnostic accuracy of STOP-Bang questionnaire on moderate sleep apnoea in primary care

Objetivo: Validar el cuestionario STOP-Bang para la apnea moderada frente al método de referencia (polisomnografía de tipo I) en atención primaria. Método: Estudio de utilidad diagnóstica en atención primaria con una muestra estimada de 85 casos y 85 controles sanos. Con muestreo por conveniencia, 203 pacientes fueron reclutados por sus médicos en seis centros de salud. Se excluyeron 25 y se analizaron 57 mujeres y 121 hombres, de los cuales 74 tenían un índice de hipopnea-apnea (IHA) ≥15. Se evaluaron el STOP-Bang y el IHA observado en la polisomnografía en cada paciente. El tamaño de la muestra, el análisis de la curva ROC y los puntos de corte óptimos se identificaron con los paquetes easyROC, pROC y OptimalCutpoints del software libre R. Resultados: El área bajo la curva en la apnea moderada (IHA ≥15) del STOP-Bang fue 0,737 (0,667-0,808), con puntos de corte óptimos diferentes por sexo (4 en mujeres y 6 en hombres). En la validación cruzada con k=10, el área bajo la curva para el STOP-Bang fue 0,678. Conclusiones: El STOP-Bang tiene una utilidad diagnóstica moderada para un IHA ≥15, pero superior a la de otras escalas, en una población comunitaria. Su desempeño es más adecuado en las mujeres

Objective: We aimed to compare the diagnostic utility of the STOP-Bang questionnaire for moderate apnoea against the gold standard (type I polysomnography) in a primary care setting. Method: Study of diagnostic utility in primary care. Estimated sample: 85 cases and 85 healthy controls. In convenience sampling, 203 patients were recruited by their physicians at six health centres. Twenty-five were excluded, and 57 women and 121 men, of whom 74 had apnoea-hypopnoea index (AHI) ≥15, were analyzed. STOP-Bang was validated by comparing scores in the same patient with the apnoea-hypopnoea index observed in polysomnography, as a gold standard. Sample size, ROC curve analysis and optimal cut-off points were identified with the easyROC, pROC, and OptimalCutpoints packages. Results: The area under the curve in moderate apnoea (AHI ≥15) of the STOP-Bang was 0.777 (0.667-0.808), with optimal cut-off points different by sex (4 in women and 6 in men). In the cross-validation with k=10, the area under the curve for the STOP-Bang was 0.678. Conclusions: The STOP-Bang presents a diagnostic moderate utility for AHI≥15, but superior to other scales, in a community population. Its performance is more appropriate in women

[Diagnostic accuracy of STOP-Bang questionnaire on moderate sleep apnoea in primary care]. / Utilidad diagnóstica del cuestionario STOP-Bang en la apnea del sueño moderada en atención primaria.

OBJECTIVE: We aimed to compare the diagnostic utility of the STOP-Bang questionnaire for moderate apnoea against the gold standard (type I polysomnography) in a primary care setting. METHOD: Study of diagnostic utility in primary care. Estimated sample: 85 cases and 85 healthy controls. In convenience sampling, 203 patients were recruited by their physicians at six health centres. Twenty-five were excluded, and 57 women and 121 men, of whom 74 had apnoea-hypopnoea index (AHI) ≥15, were analyzed. STOP-Bang was validated by comparing scores in the same patient with the apnoea-hypopnoea index observed in polysomnography, as a gold standard. Sample size, ROC curve analysis and optimal cut-off points were identified with the easyROC, pROC, and OptimalCutpoints packages. RESULTS: The area under the curve in moderate apnoea (AHI ≥15) of the STOP-Bang was 0.777 (0.667-0.808), with optimal cut-off points different by sex (4 in women and 6 in men). In the cross-validation with k=10, the area under the curve for the STOP-Bang was 0.678. CONCLUSIONS: The STOP-Bang presents a diagnostic moderate utility for AHI≥15, but superior to other scales, in a community population. Its performance is more appropriate in women.

Bootstrap-based procedures for inference in nonparametric receiver-operating characteristic curve regression analysis.

Prior to using a diagnostic test in a routine clinical setting, the rigorous evaluation of its diagnostic accuracy is essential. The receiver-operating characteristic curve is the measure of accuracy most widely used for continuous diagnostic tests. However, the possible impact of extra information about the patient (or even the environment) on diagnostic accuracy also needs to be assessed. In this paper, we focus on an estimator for the covariate-specific receiver-operating characteristic curve based on direct regression modelling and nonparametric smoothing techniques. This approach defines the class of generalised additive models for the receiver-operating characteristic curve. The main aim of the paper is to offer new inferential procedures for testing the effect of covariates on the conditional receiver-operating characteristic curve within the above-mentioned class. Specifically, two different bootstrap-based tests are suggested to check (a) the possible effect of continuous covariates on the receiver-operating characteristic curve and (b) the presence of factor-by-curve interaction terms. The validity of the proposed bootstrap-based procedures is supported by simulations. To facilitate the application of these new procedures in practice, an R-package, known as npROCRegression, is provided and briefly described. Finally, data derived from a computer-aided diagnostic system for the automatic detection of tumour masses in breast cancer is analysed.

Assessing neural activity related to decision-making through flexible odds ratio curves and their derivatives.

It is well established that neural activity is stochastically modulated over time. Therefore, direct comparisons across experimental conditions and determination of change points or maximum firing rates are not straightforward. This study sought to compare temporal firing probability curves that may vary across groups defined by different experimental conditions. Odds-ratio (OR) curves were used as a measure of comparison, and the main goal was to provide a global test to detect significant differences of such curves through the study of their derivatives. An algorithm is proposed that enables ORs based on generalized additive models, including factor-by-curve-type interactions to be flexibly estimated. Bootstrap methods were used to draw inferences from the derivatives curves, and binning techniques were applied to speed up computation in the estimation and testing processes. A simulation study was conducted to assess the validity of these bootstrap-based tests. This methodology was applied to study premotor ventral cortex neural activity associated with decision-making. The proposed statistical procedures proved very useful in revealing the neural activity correlates of decision-making in a visual discrimination task.

Selecting variables in non-parametric regression models for binary response. An application to the computerized detection of breast cancer.

In many biomedical applications, interest lies in being able to distinguish between two possible states of a given response variable, depending on the values of certain continuous predictors. If the number of predictors, p, is high, or if there is redundancy among them, it then becomes important to decide on the selection of the best subset of predictors that will be able to obtain the models with greatest discrimination capacity. With this aim in mind, logistic generalized additive models were considered and receiver operating characteristic (ROC) curves were applied in order to determine and compare the discriminatory capacity of such models. This study sought to develop bootstrap-based tests that allow for the following to be ascertained: (a) the optimal number q < or = p of predictors; and (b) the model or models including q predictors, which display the largest AUC (area under the ROC curve). A simulation study was conducted to verify the behaviour of these tests. Finally, the proposed method was applied to a computer-aided diagnostic system dedicated to early detection of breast cancer.

Categorical variables, interactions and generalized additive models. Applications in computer-aided diagnosis systems.

Recently, the generalized additive models (GAMs) have been presented as a novel statistical approach to distinguish lesion/non-lesion in computer-aided diagnosis (CAD) systems. In this paper, we present an extension of the GAM that allows for the introduction of factors and their interactions with continuous variables, for reducing false positives in a CAD system for detecting clustered microcalcifications in digital mammograms. The results obtained have shown an increase in the sensitivity from 83.12% to 85.71%, while the false positive rate was drastically reduced from 1.46 to 0.74 false detections per image.

Neural correlates of memory retrieval in the prefrontal cortex.

Working memory includes short-term representations of information that were recently experienced or retrieved from long-term representations of sensory stimuli. Evidence is presented here that working memory activates the same dorsolateral prefrontal cortex neurons that: (a) maintained recently perceived visual stimuli; and (b) retrieved visual stimuli from long-term memory (LTM). Single neuron activity was recorded in the dorsolateral prefrontal cortex while trained monkeys discriminated between two orientated lines shown sequentially, separated by a fixed interstimulus interval. This visual task required the monkey to compare the orientation of the second line with the memory trace of the first and to decide the relative orientation of the second. When the behavioural task required the monkey to maintain in working memory a first stimulus that continually changed from trial to trial, the discharge in these cells was related to the parameters--the orientation--of the memorized item. Then, what the monkey had to recall from memory was manipulated by switching to another task in which the first stimulus was not shown, and had to be retrieved from LTM. The discharge rates of the same neurons also varied depending on the parameters of the memorized stimuli, and their response was progressively delayed as the monkey performed the task. These results suggest that working memory activates dorsolateral prefrontal cortex neurons that maintain parametrical visual information in short-term and LTM, and that the contents of working memory cannot be limited to what has recently happened in the sensory environment.

Using generalized additive models for construction of nonlinear classifiers in computer-aided diagnosis systems.

Several investigators have pointed out the possibility of using computer-aided diagnosis (CAD) schemes, as second readers, to help radiologists in the interpretation of images. One of the most important aspects to be considered when the diagnostic imaging systems are analyzed is the evaluation of their diagnostic performance. To perform this task, receiver operating characteristic curves are the method of choice. An important step in nearly all CAD systems is the reduction of false positives, as well as the classification of lesions, using different algorithms, such as neural networks or feature analysis, and several statistical methods. A statistical model more often employed is linear discriminant analysis (LDA). However, LDA implies several limitations in the type of variables that it can analyze. In this work, we have developed a novel approach, based on generalized additive models (GAMs), as an alternative to LDA, which can deal with a broad variety of variables, improving the results produced by using the LDA model. As an application, we have used GAM techniques for reducing the number of false detections in a computerized method to detect clustered microcalcifications, and we have compared this with the results obtained when LDA was applied. Employing LDA, the system achieved a sensitivity of 80.52% at a false-positive rate of 1.90 false detections per image. With the GAM, the sensitivity increased to 83.12% and 1.46 false positives per image.

Effect measures in non-parametric regression with interactions between continuous exposures.

In many biomedical studies, interest is often attached to calculating effect measures in the presence of interactions between two continuous exposures. Traditional approaches based on parametric regression are limited by the degree of arbitrariness involved in transforming these exposures into categorical variables or imposing a parametric form on the regression function. In this paper, we present: (a) a flexible non-parametric method for estimating effect measures through generalized additive models including interactions; and (b) bootstrap techniques for (i) testing the significance of interaction terms, and (ii) constructing confidence intervals for effect measures. The validity of our methodology is supported by simulations, and illustrated using data from a study of possible risk factors for post-operative infection. This application revealed a hitherto unreported effect: for patients with high plasma glucose levels, increased risk is associated, not only with low, but also with high percentages of lymphocytes.

Bootstrap-based methods for testing factor-by-curve interactions in generalized additive models: assessing prefrontal cortex neural activity related to decision-making.

In many situations the effect of a continuous covariate on response varies across groups defined by levels of a categorical variable. This paper addresses generalized additive models incorporating the so-called factor-by-curve interaction. A local scoring algorithm based on local linear kernel smoothers was used to estimate the model. Two different types of bootstrap-based procedures are proposed for testing interaction terms, namely, the likelihood ratio test, and a procedure based on an estimate of the interaction terms. Given the high computational cost involved, binning techniques were used to speed up computation in the estimation and testing processes. A simulation study was conducted to assess the validity of these bootstrap-based tests. This methodology was applied to studying prefrontal cortex neural activity associated with decision-making in monkeys. The proposed statistical procedure proved very useful in revealing the neural activity correlates of decision-making strategies adopted by monkeys in accordance with different behavioural tasks.

A bootstrap method to avoid the effect of concurvity in generalised additive models in time series studies of air pollution.

BACKGROUND: In recent years a great number of studies have applied generalised additive models (GAMs) to time series data to estimate the short term health effects of air pollution. Lately, however, it has been found that concurvity--the non-parametric analogue of multicollinearity--might lead to underestimation of standard errors of the effects of independent variables. Underestimation of standard errors means that for concurvity levels commonly present in the data, the risk of committing type I error rises by over threefold. METHODS: This study developed a conditional bootstrap methology that consists of assuming that the outcome in any observation is conditional upon the values of the set of independent variables used. It then tested this procedure by means of a simulation study using a Poisson additive model. The response variable of this model is a function of an unobserved confounding variable (that introduces trend and seasonality), real black smoke data, and temperature. Scenarios were created with different coefficients and degrees of concurvity. RESULTS: Conditional bootstrap provides confidence intervals with coverages close to nominal (95%), irrespective of the degree of concurvity, number of variables in the model or magnitude of the coefficient to be estimated (for example, for a concurvity of 0.85, bootstrap confidence interval coverage is 95% compared with 71% in the case of the asymptotic interval obtained directly with S-plus gam function). CONCLUSIONS: The bootstrap method avoids the problem of concurvity in time series studies of air pollution, and is easily generalised to non-linear dose-risk effects. All bootstrap calculations described in this paper can be performed using S-Plus gam.boot software.