Quantile-Based Optimization of Noisy Computer Experiments with Tunable Precision

This article addresses the issue of kriging-based optimization of stochastic simulators. Many of these simulators depend on factors that tune the level of precision of the response, the gain in accuracy being at a price of computational time. The contribution of this work is two-fold: first, we propose a quantile-based criterion for the sequential design of experiments, in the fashion of the classical expected improvement criterion, which allows an elegant treatment of heterogeneous response precisions. Second, we present a procedure for the allocation of the computational time given to each measurement, allowing a better distribution of the computational effort and increased efficiency. Finally, the optimization method is applied to an original application in nuclear criticality safety. This article has supplementary material available online. The proposed criterion is available in the R package DiceOptim.

A computer-aided diagnostic system to characterize CT focal liver lesions: design and optimization of a neural network classifier

In this paper, a computer-aided diagnostic (CAD) system for the classification of hepatic lesions from computed tomography (CT) images is presented. Regions of interest (ROIs) taken from nonenhanced CT images of normal liver, hepatic cysts, hemangiomas, and hepatocellular carcinomas have been used as input to the system. The proposed system consists of two modules: the feature extraction and the classification modules. The feature extraction module calculates the average gray level and 48 texture characteristics, which are derived from the spatial gray-level co-occurrence matrices, obtained from the ROIs. The classifier module consists of three sequentially placed feed-forward neural networks (NNs). The first NN classifies into normal or pathological liver regions. The pathological liver regions are characterized by the second NN as cyst or "other disease." The third NN classifies "other disease" into hemangioma or hepatocellular carcinoma. Three feature selection techniques have been applied to each individual NN: the sequential forward selection, the sequential floating forward selection, and a genetic algorithm for feature selection. The comparative study of the above dimensionality reduction methods shows that genetic algorithms result in lower dimension feature vectors and improved classification performance.

Comparing growth trajectories of risk behaviors from late adolescence through young adulthood: An accelerated design

Risk behaviors such as substance use or deviance are often limited to the early stages of the life course. Whereas the onset of risk behavior is well studied, less is currently known about the decline and timing of cessation of risk behaviors of different domains during young adulthood. Prevalence and longitudinal developmental patterning of alcohol use, drinking to the point of drunkenness, smoking, cannabis use, deviance, and HIV-related sexual risk behavior were compared in a Swiss community sample (N = 2,843). Using a longitudinal cohort-sequential approach to link multiple assessments with 3 waves of data for each individual, the studied period spanned the ages of 16 to 29 years. Although smoking had a higher prevalence, both smoking and drinking up to the point of drunkenness followed an inverted U-shaped curve. Alcohol consumption was also best described by a quadratic model, though largely stable at a high level through the late 20s. Sexual risk behavior increased slowly from age 16 to age 22 and then remained largely stable. In contrast, cannabis use and deviance linearly declined from age 16 to age 29. Young men were at higher risk for all behaviors than were young women, but apart from deviance, patterning over time was similar for both sexes. Results about the timing of increase and decline as well as differences between risk behaviors may inform tailored prevention programs during the transition from late adolescence to adulthood.

The FReedom from Ischemic Events-New Dimensions for Survival (FRIENDS) registry: design of a prospective cohort study of patients with advanced peripheral artery disease.

BACKGROUND Advanced lower extremity peripheral artery disease (PAD), whether presenting as acute limb ischemia (ALI) or chronic critical limb ischemia (CLI), is associated with high rates of cardiovascular ischemic events, amputation, and death. Past research has focused on strategies of revascularization, but few data are available that prospectively evaluate the impact of key process of care factors (spanning pre-admission, acute hospitalization, and post-discharge) that might contribute to improving short and long-term health outcomes. METHODS/DESIGN The FRIENDS registry is designed to prospectively evaluate a range of patient and health system care delivery factors that might serve as future targets for efforts to improve limb and systemic outcomes for patients with ALI or CLI. This hypothesis-driven registry was designed to evaluate the contributions of: (i) pre-hospital limb ischemia symptom duration, (ii) use of leg revascularization strategies, and (iii) use of risk-reduction pharmacotherapies, as pre-specified factors that may affect amputation-free survival. Sequential patients would be included at an index "vascular specialist-defined" ALI or CLI episode, and patients excluded only for non-vascular etiologies of limb threat. Data including baseline demographics, functional status, co-morbidities, pre-hospital time segments, and use of medical therapies; hospital-based use of revascularization strategies, time segments, and pharmacotherapies; and rates of systemic ischemic events (e.g., myocardial infarction, stroke, hospitalization, and death) and limb ischemic events (e.g., hospitalization for revascularization or amputation) will be recorded during a minimum of one year follow-up. DISCUSSION The FRIENDS registry is designed to evaluate the potential impact of key factors that may contribute to adverse outcomes for patients with ALI or CLI. Definition of new "health system-based" therapeutic targets could then become the focus of future interventional clinical trials for individuals with advanced PAD.