Receiver operating characteristic (ROC) curves are frequently used in biomedical informatics research to evaluate classification and prediction models for decision support, diagnosis, and prognosis. ROC analysis investigates the accuracy of a model's ability to separate positive from negative cases (such as predicting the presence or absence of disease), and the results are independent of the prevalence of positive cases in the study population. It is especially useful in evaluating predictive models or other tests that produce output values over a continuous range, since it captures the trade-off between sensitivity and specificity over that range. There are many ways to conduct an ROC analysis. The best approach depends on the experiment; an inappropriate approach can easily lead to incorrect conclusions. In this article, we review the basic concepts of ROC analysis, illustrate their use with sample calculations, make recommendations drawn from the literature, and list readily available software.

The main variables associated with incomplete tumor resection in 101 patients were identified by using statistical predictive analyses.

To optimize opacification and distention of the normal urinary collecting system, contrast material-enhanced multi-detector row CT urography may be supplemented with intravenous furosemide alone.

Two experiments explored the ability of 18-month-old infants to form an abstract categorical representation of tight-fit spatial relations in a visual habituation task. In Experiment 1, infants formed an abstract spatial category when hearing a familiar word (tight) during habituation but not when viewing the events in silence or when hearing a novel word. In Experiment 2, infants were given experience viewing and producing tight-fit relations while an experimenter labeled them with a novel word. Following this experience, infants formed the tight-fit spatial category in the visual habituation task, particularly when hearing the novel word again during habituation. Results suggest that even brief experience with a label and tight-fit relations can aid infants in forming an abstract categorical representation of tight-fit relations. Keywords infant categorization; spatial cognition; spatial languageProviding labels for the objects and events in an infant's visual world can facilitate his or her ability to learn about the environment. Infants of 12 months attend longer to objects that have been labeled relative to those that remained unlabeled (Baldwin, 1991). Labels also motivate infants to abstract commonalities to form categories that they might not otherwise form, with respect to both objects (e.g., Balaban & Waxman, 1997;Booth & Waxman, 2002;Roberts & Jacob, 1991;Waxman & Markow, 1995) and dynamic events (Casasola, 2005a;Casasola & Bhagwat, 2007; Pruden & Hirsh-Pasek, 2006). However, the extent to which labels are needed in the formation of specific categories remains unclear and is an issue that has received much discussion, particularly with respect to the spatial categories that infants form (e.g., Bowerman & Choi, 2001;Choi & Bowerman, 1991;Hespos & Spelke, 2004;Mandler, 1996). In the present two experiments, we explored this issue by examining 18-month-old English-learning infants' formation of a spatial category of tight-fitting relations, a distinction that is not lexically marked in English but which forms the basis of the Korean semantic category of kkita (further described below). Infants were tested on their ability to form a spatial category of tight fit when provided with a specific label or when viewing the events in silence. The studies also explored whether the effect of a specific label on infants' spatial categorization differed as a function of label familiarity and, in a second study, as a function of firsthand experience with tight-fit relations. Together, these studies represent an effort to begin to outline how infants' experiences with particular labels and tight-fit relations may differentially influence the spatial categories that they learn to form.

Vitamin A supplementation has consistently reduced infant mortality and the severity of pathogen-induced diarrhea. The mechanism by which vitamin A modulates the mucosal immune response to produce these effects remains poorly defined. To address this issue, stools collected during the summer months from 127 Mexican children 5-15 mo old enrolled in a larger, randomized, double-blind, placebo-controlled, vitamin A supplementation trial were screened for interleukin (IL)-4, IL-6, interferon-gamma (IFN-gamma), and gastrointestinal pathogens. Fecal cytokine values were categorized into 3 levels (undetectable, or =median). Multinomial regression models were used to determine the probability that vitamin A-supplemented children had higher categorical values of a cytokine than children in the placebo group. Differences in categorical values were also analyzed after stratification by gastrointestinal pathogen infections and diarrheal symptoms. Overall, fecal cytokine categorical levels did not differ between children randomized to the 2 arms. Vitamin A-supplemented children infected with enteropathogenic E. coli (EPEC) had reduced IL-4 and IFN-gamma levels [odds ratio (OR) = 0.3, 95% CI 0.13-0.67 and OR = 0.34, 95% CI 0.14-0.83, respectively] compared with children in the placebo group. Vitamin A-supplemented children had increased IL-4 levels when infected with A. lumbricoides (OR = 12.06, 95% CI 0.95-153.85). In contrast, IL-4 levels increased (OR = 2.14, 95% CI 0.94-4.87) and IFN-gamma levels decreased (OR = 0.51, 95% CI 0.26-0.99) among vitamin A-supplemented children with diarrhea compared with children in the placebo group. These findings suggest that the regulation of the mucosal immune response by vitamin A may depend on the type of enteric pathogen infecting the child and the presence of clinical symptoms.