Papers by Carmen Cadarso-suárez
The receiver operating characteristic (ROC) curve is the most widely used measure for evaluating ... more The receiver operating characteristic (ROC) curve is the most widely used measure for evaluating the accuracy of diagnostic tests in terms of differentiating between two conditions. It is well known that, in certain circumstances, the characteristics of the patient or the place where the diagnostic test is performed can modify the test’s accuracy. This study presents a new estimator for the conditional ROC curve, based on direct modeling. In this approach, the effect of the covariates and false positive fraction on the ROC curve is modeled nonparametrically using generalized additive models (GAM) combined with local polynomial kernel smoothers. Our method allows for incorporation of more than one covariate in the regression model for the ROC curve, and the possible interaction between them. The proposed model’s performance is examined in an in-depth simulation study. Lastly, endocrine data are analyzed with the aim of assessing the performance of several anthropometric measures in p...
Computer Methods and Programs in Biomedicine, 2007
The aim of this paper is to present an R library, called tdc.msm, developed to analyze multi-stat... more The aim of this paper is to present an R library, called tdc.msm, developed to analyze multi-state survival data. In this library, the time-dependent regression model and multi-state models are included as two possible approaches for such data. For the multi-state modelling five different models are considered, allowing the user to choose between Markov and semi-Markov property, as well as
Calculating odds ratios (OR) and corresponding confidence intervals (CI) for exposures that have ... more Calculating odds ratios (OR) and corresponding confidence intervals (CI) for exposures that have been measured using a continuous scale, presents important limitations in the traditional practice of epidemiology. Our objective is to describe an S-Plus function, that we called np.OR, that allows the computation of the pointwise estimates of the ORs as well as their corresponding CIs of continuous predictors introduced nonlinearly in a generalised model. The function can also be generalised to compute the pointwise estimates of categorical predictors that were introduced nonlinearly. To illustrate usage of the program we analyse the relationship between ambient temperature and total mortality in Barcelona for the period 1991-1995, while controlling for observed and unobserved confounders.
Institute of Mathematical Statistics Lecture Notes - Monograph Series, 1997
... Jacobo de U?a-Alvarez University of Vigo, Spain Wenceslao Gonz?lez-Manteiga and Carmen Cadars... more ... Jacobo de U?a-Alvarez University of Vigo, Spain Wenceslao Gonz?lez-Manteiga and Carmen Cadarso-Su?rez University of Santiago de Compostela, Spain ... tests about such a model (see, for example, Herbst, 1992; Henze, 1993). Under (1) it is true that 1 - F = (1 - ?)T, with ? ...
... Marıa P. Pata1, Marıa Xosé Rodrıguez-´Alvarez2,3, Vicente Lustres-Pérez1 Eugenio Fernández-Pu... more ... Marıa P. Pata1, Marıa Xosé Rodrıguez-´Alvarez2,3, Vicente Lustres-Pérez1 Eugenio Fernández-Pulpeiro1, Carmen Cadarso-Suárez2,3 ... of species is essential in order to understand the ecological processes that have generated such processes (Underwood, Chapman and ...
Atencion Primaria, 2006
Objetivo. Determinar los factores asociados con la incidencia y la duración de la incapacidad tem... more Objetivo. Determinar los factores asociados con la incidencia y la duración de la incapacidad temporal (IT) en un área sanitaria. Diseño. Descriptivo, retrospectivo. Emplazamiento. Área Sanitaria Sur de la provincia de Lugo. Participantes. Una muestra de 1.513 episodios de IT seleccionada aleatoriamente entre el total de éstos, durante un período de 3 años. Mediciones principales. Se analizaron las características sociodemográficas del paciente, el régimen de la seguridad social (SS), el diagnóstico que justifica la IT y la fecha de la prescripción; del médico prescriptor se analizaron la edad, el sexo, la formación especializada, la antigüedad en la plaza y los años de ejercicio. La comparación de medias se realizó mediante el análisis de la varianza y el test de Kruskal-Wallis. El efecto relativo de cada variable sobre la probabilidad de volver al trabajo se estimó mediante modelos de regresión de Cox. Resultados. La duración media de los episodios de IT fue de 74 ± 103 días. Los diagnósticos más frecuentes fueron los del sistema osteomioarticular (SOMA), las lesiones y envenenamientos (LYE) y las enfermedades respiratorias (NML). Se reduce la probabilidad de volver al trabajo con el incremento de la edad, en los regímenes de seguridad social autónomos y agrarios por cuenta propia, en los diagnósticos de enfermedades mentales y del aparato circulatorio, y cuando el médico prescriptor es de mayor edad o menos antiguo en la plaza. Conclusiones. La duración media de los episodios de IT es superior a la de otros estudios españoles. Los factores que más influyen en la reincorporación al trabajo son la edad del paciente, el régimen de la seguridad social y la enfermedad diagnosticada.
Diseases of The Colon & Rectum, 1999
PURPOSE: In many types of cancer, certain morphometric characteristc of tumor cells correlate wit... more PURPOSE: In many types of cancer, certain morphometric characteristc of tumor cells correlate with patient survival. Our observations suggested that the survival of patients with colorectal carcinomas is negatively correlated with tumor-cell nucleus size. METHODS: We investigated relationships between postsurgery survival and nucleus morphometrics in 90 patients who had undergone resection for a colorectal tumor. The nucleus-size variables considered were
Journal of Geriatric Psychiatry and Neurology, 2006
The authors compared the accuracy of Addenbrooke’s Cognitive Examination Clock Drawing Test scori... more The authors compared the accuracy of Addenbrooke’s Cognitive Examination Clock Drawing Test scoring method with 8 standard Clock Drawing Test scoring methods in a Spanish speaking population of low educational level. A clinical group composed of 70 patients affected by dementia and 25 patients with memory complaints without dementia was compared with 72 controls matched for gender, age, and educational
Thyroid, 2007
To identify the current state of iodine nutrition in the adult population of Galicia (Spain), whi... more To identify the current state of iodine nutrition in the adult population of Galicia (Spain), which is considered iodine sufficient based on results from studies carried out on schoolchildren. Urinary iodine concentration (UIC) and its relationship with different socio-demographic variables were assessed. A cross-sectional study was carried out on the population aged above 18 years in the Autonomous Community of Galicia (Spain) during 2004. The UICs were determined in an isolated urine sample using Dunn's colorimetric method. Iodine status was based on World Health Organization/International Council for the Control of Iodine Deficiency Disorders (WHO/ICCIDD) UIC. A total of 2877 urine samples were taken. Median UIC for the total Galician population was 75.6 microg/L. About 30% of the population showed a UIC below 50 microg/L. Educational level, place of residence (coast vs. inland), and consumption of iodized salt were independent variables associated with the iodine nutrition of the adult population of Galicia. There is "mild" iodine deficiency (WHO) in the adult population of Galicia, which affects all the groups analyzed and which is particularly significant in the group of women of a fertile age. The data obtained on the state of iodine nutrition in school-age populations cannot be extended to the adult population.
Statistics in Medicine, 2008
Correct determination of time of death is an important goal in forensic medicine. Numerous method... more Correct determination of time of death is an important goal in forensic medicine. Numerous methods have been described for estimating postmortem interval (PMI), but most are imprecise, poorly reproducible and/or have not been validated with real data. In recent years, however, some progress in PMI estimation has been made, notably through the use of new biochemical methods for quantifying relevant indicator compounds in the vitreous humour. The best, but unverified, results have been obtained with [K+] and hypoxanthine [Hx], using simple linear regression (LR) models. The main aim of this paper is to offer more flexible alternatives to LR, such as generalized additive models (GAMs) and support vector machines (SVMs) in order to obtain improved PMI estimates. The present study, based on detailed analysis of [K+] and [Hx] in more than 200 vitreous humour samples from subjects with known PMI, compared classical LR methodology with GAM and SVM methodologies. Both proved better than LR for estimation of PMI. SVM showed somewhat greater precision than GAM, but GAM offers a readily interpretable graphical output, facilitating understanding of findings by legal professionals; there are thus arguments for using both types of models. R code for these methods is available from the authors, permitting accurate prediction of PMI from vitreous humour [K+], [Hx] and [U], with confidence intervals and graphical output provided.
Statistics in Medicine, 2009
In many biomedical applications, interest lies in being able to distinguish between two possible ... more 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.
Statistics and Computing, 2011
Continuous diagnostic tests are often used to discriminate between diseased and healthy populatio... more Continuous diagnostic tests are often used to discriminate between diseased and healthy populations. The receiver operating characteristic (ROC) curve is a widely used tool that provides a graphical visualisation of the effectiveness of such tests. The potential performance of the tests in terms of distinguishing diseased from healthy people may be strongly influenced by covariates, and a variety of regression methods for adjusting ROC curves has been developed. Until now, these methodologies have assumed that covariate effects have parametric forms, but in this paper we extend the induced methodology by allowing for arbitrary nonparametric effects of a continuous covariate. To this end, local polynomial kernel smoothers are used in the estimation procedure. Our method allows for covariate effect not only on the mean, but also on the variance of the diagnostic test. We also present a bootstrap-based method for testing for a significant covariate effect on the ROC curve. To illustrate the method, endocrine data were analysed with the aim of assessing the performance of anthropometry for predicting clusters of cardiovascular risk factors in an adult population in Galicia (NW Spain), duly adjusted for age. The proposed methodology has proved useful for providing age-specific M.X. Rodríguez-Álvarez ( ) · C. Cadarso-Suárez thresholds for anthropometric measures in the Galician community.
Statistics and Computing, 2005
In this paper we considered a generalized additive model with second-order interaction terms. A l... more In this paper we considered a generalized additive model with second-order interaction terms. A local scoring algorithm (with backfitting) based on local linear kernel smoothers was used to estimate the model. Our main aim was to obtain procedures for testing second-order interaction terms. Backfitting theory is difficult in this context, and a bootstrap procedure is therefore provided for estimating the
Statistical Methods in Medical Research, 2008
The experience of a patient in a survival study may be thought of as a process that involves two ... more The experience of a patient in a survival study may be thought of as a process that involves two states, with one possible transition from a 'live' state to a 'dead' state. In some studies, however, the state representing those patients 'alive' may be partitioned into two or more intermediate (transient) states, each of which corresponding to a particular stage of the illness. In such studies where at any time point individuals occupy one of a set of discrete states, multi-state models can be used to model the movement of patients between the various states. These models may offer a better understanding of the process of the illness, leading to a better knowledge of the evolution of the disease over time. Issues of interest include the estimation of progression rates, assessing the effects of individual risk factors, survival rates or prognostic forecasting. In this article, we review several modelling approaches following the methodology of the multi-state models, and focusing on the estimation of several quantities such as the transition probabilities and survival probabilities.
Quality of Life Research, 2000
Mathematical and Computer Modelling, 2011
Lifetime Data Analysis, 2006
In this paper we consider nonparametric estimation of transition probabilities for multi-state mo... more In this paper we consider nonparametric estimation of transition probabilities for multi-state models. Specifically, we focus on the illness-death or disability model. The main novelty of the proposed estimators is that they do not rely on the Markov assumption, typically assumed to hold in a multi-state model. We investigate the asymptotic properties of the introduced estimators, such as their consistency and their convergence to a normal law. Simulations demonstrate that the new estimators may outperform Aalen-Johansen estimators (the classical nonparametric tool for estimating the transition probabilities) in non-Markov situation. An illustration through real data analysis is included.
Journal of the Royal Statistical Society: Series C (Applied Statistics), 2006
In many electrophysiological experiments the main objectives include estimation of the firing rat... more In many electrophysiological experiments the main objectives include estimation of the firing rate of a single neuron, as well as a comparison of its temporal evolution across different experimental conditions. To accomplish these two goals, we propose a flexible approach based on the logistic generalized additive model including condition-by-time interactions. If an interaction of this type is detected in the model, we then establish that the use of the temporal odds ratio curves is very useful in discriminating between the conditions under which the firing probability is higher. Bootstrap techniques are used for testing for interactions and constructing pointwise confidence bands for the true odds ratio curves. Finally, we apply the new methodology to assessing relationships between neural response and decision-making in movement-selective neurons in the prefrontal cortex of behaving monkeys.
Journal of the American Statistical Association, 2008
Neurons can transmit information about the characteristics of a stimulus via the spike rate of ne... more Neurons can transmit information about the characteristics of a stimulus via the spike rate of neurons and via synchronization of the neurons. To describe how 'synchronous' two spike trains are, a variety of association measures can be used. We propose a new measure of synchrony, the conditional synchrony measure, which is the probability of firing together given that at least one of the two neurons is active. Focus is on the specification of a flexible marginal model for multivariate correlated binary data together with a pseudo-likelihood estimation approach, to adequately and directly describe the measures of interest. A joint model must allow different time-and covariatedepending firing rates for each neuron, and must account for the association between them. The association between neurons might depend on covariates as well.
Uploads
Papers by Carmen Cadarso-suárez