Brunel University London

About: Brunel University London is a education organization based out in London, United Kingdom. It is known for research contribution in the topics: Context (language use) & Large Hadron Collider. The organization has 10918 authors who have published 29515 publications receiving 893330 citations. The organization is also known as: Brunel & University of Brunel.

Researcher Bias: The Use of Machine Learning in Software Defect Prediction

Abstract: Background . The ability to predict defect-prone software components would be valuable. Consequently, there have been many empirical studies to evaluate the performance of different techniques endeavouring to accomplish this effectively. However no one technique dominates and so designing a reliable defect prediction model remains problematic. Objective. We seek to make sense of the many conflicting experimental results and understand which factors have the largest effect onpredictive performance. Method. We conduct a meta-analysis of all relevant, high quality primary studies of defect prediction to determine what factors influence predictive performance. This is based on 42 primary studies that satisfy our inclusion criteria that collectively report 600 sets of empirical prediction results. By reverse engineering a common response variable we build arandom effects ANOVA model to examine the relative contribution of four model building factors (classifier, data set, input metrics and researcher group) to model prediction performance. Results. Surprisingly we find that the choice of classifier has little impact upon performance (1.3 percent) and in contrast the major (31 percent) explanatory factor is the researcher group. It matters more who does the work than what is done. Conclusion. To overcome this high level of researcher bias, defect prediction researchers should (i) conduct blind analysis, (ii) improve reporting protocols and (iii) conduct more intergroup studies in order to alleviate expertise issues. Lastly, research is required to determine whether this bias is prevalent in other applications domains.

A review of health utilities using the EQ-5D in studies of cardiovascular disease.

Abstract: The EQ-5D has been extensively used to assess patient utility in trials of new treatments within the cardiovascular field. The aims of this study were to review evidence of the validity and reliability of the EQ-5D, and to summarise utility scores based on the use of the EQ-5D in clinical trials and in studies of patients with cardiovascular disease. A structured literature search was conducted using keywords related to cardiovascular disease and EQ-5D. Original research studies of patients with cardiovascular disease that reported EQ-5D results and its measurement properties were included. Of 147 identified papers, 66 met the selection criteria, with 10 studies reporting evidence on validity or reliability and 60 reporting EQ-5D responses (VAS or self-classification). Mean EQ-5D index-based scores ranged from 0.24 (SD 0.39) to 0.90 (SD 0.16), while VAS scores ranged from 37 (SD 21) to 89 (no SD reported). Stratification of EQ-5D index scores by disease severity revealed that scores decreased from a mean of 0.78 (SD 0.18) to 0.51 (SD 0.21) for mild to severe disease in heart failure patients and from 0.80 (SD 0.05) to 0.45 (SD 0.22) for mild to severe disease in angina patients. The published evidence generally supports the validity and reliability of the EQ-5D as an outcome measure within the cardiovascular area. This review provides utility estimates across a range of cardiovascular subgroups and treatments that may be useful for future modelling of utilities and QALYs in economic evaluations within the cardiovascular area.

The Turn to Technology in Social Studies of Science.

Abstract: This article examines how the special theoretical significance of the sociology of scientific knowledge (SSK) is affected by attempts to apply relativist-constructivism to technology. The article shows that the failure to confront key analytic ambivalences in the practice of SSK has compromised its original strategic significance. In particular, the construal of SSK as an explanatory formula diminishes its potential for profoundly reconceptualizing epistemic issues. A consideration of critiques of technological determinism, and of some empirical studies, reveals similar analytic ambivalences in the social study of technology (SST). The injunction to consider "technology as text" is critically examined. It is concluded that a reflexive interpretation of this slogan is necessary to recover some of the epistemological significance lost in the constructivist move from SSK to SST.

Predicted-no-effect concentrations for the steroid estrogens estrone, 17β-estradiol, estriol, and 17α-ethinylestradiol

Abstract: The authors derive predicted-no-effect concentrations (PNECs) for the steroid estrogens (estrone [E1], 17β-estradiol [E2], estriol [E3], and 17α-ethinylestradiol [EE2]) appropriate for use in risk assessment of aquatic organisms. In a previous study, they developed a PNEC of 0.35 ng/L for EE2 from a species sensitivity distribution (SSD) based on all available chronic aquatic toxicity data. The present study updates that PNEC using recently published data to derive a PNEC of 0.1 ng/L for EE2. For E2, fish were the most sensitive taxa, and chronic reproductive effects were the most sensitive endpoint. Using the SSD methodology, we derived a PNEC of 2 ng/L for E2. Insufficient data were available to construct an SSD for E1 or E3. Therefore, the authors used in vivo vitellogenin (VTG) induction studies to determine the relative potency of the steroid estrogens to induce VTG. Based on the relative differences between in vivo VTG induction, they derive PNECs of 6 and 60 ng/L for E1 and E3, respectively. Thus, for long-term exposures to steroid estrogens in surface water (i.e., >60 d), the PNECs are 6, 2, 60, and 0.1 ng/L for E1, E2, E3, and EE2, respectively. Higher PNECs are recommended for short-term (i.e., a few days or weeks) exposures.