Designs for generalized linear models with several variables and model uncertainty

Standard factorial designs sometimes may be inadequate for experiments that aim to estimate a generalized linear model, for example, for describing a binary response in terms of several variables. A method is proposed for finding exact designs for such experiments that uses a criterion allowing for uncertainty in the link function, the linear predictor, or the model parameters, together with a design search. Designs are assessed and compared by simulation of the distribution of efficiencies relative to locally optimal designs over a space of possible models. Exact designs are investigated for two applications, and their advantages over factorial and central composite designs are demonstrated.

Economic and environmental impacts of pollution control in a system of environment and economic interdependence

The importance of the rate of change of the pollution stock in determining the damage to the environment has been an issue of increasing concern in the literature. This paper uses a three-sector (economy, population and environment), non-linear, discrete time, calibrated model to examine pollution control. The model explicitly links economic growth to the health of the environment. The stock of natural resources is affected by the rate of pollution flows, through their impact on the regenerative capacity of the natural resource stock. This can shed useful insights into pollution control strategies, particularly in developing countries where environmental resources are crucial for production in many sectors of the economy. Simulation exercises suggested that, under plausible assumptions, it is possible to reverse undesirable transient dynamics through pollution control expenditure, but this is dependent upon the strategies used for control. The best strategy is to spend money fostering the development of production technologies that reduce pollution rather than spending money dealing with the effects of the pollution flow into the environment. (C) 2001 Elsevier Science Ltd. All rights reserved.

Parametric signal regeneration

We present an ultra-high bandwidth all-optical digital signal regeneration device concept utilising non-degenerate parametric interaction in a one-dimensional waveguide. Performance is analysed in terms of re-amplification, re-timing, and re-shaping (including centre frequency correction) of time domain multiplexed signals. Bandwidths of 10-100 THz are achievable. (C) 2001 Published by Elsevier Science B.V.

Immunogenicity of recombinant HBsAg/HCV particles in mice pre-immunised with hepatitis B virus-specific vaccine

Due to their spatial structure virus-like particles (VLPs) generally induce effective immune responses. VLPs derived from the small envelope protein (HBsAg-S) of hepatitis B virus (HBV) comprise the HBV vaccine. Modified HBsAs-S VLPs, carrying the immunodominant hypervariable region (HVR1) of the hepatitis C virus (HCV) envelope protein E2 within the exposed 'a'-determinant region (HBsAg/HVR1-VLPs), elicited HVR1-specific antibodies in mice. A high percentage of the human population is positive for anti-HBsAg antibodies (anti-HBs), either through vaccination or natural infection. We, therefore, determined if pre-existing anti-HBs could influence immunisation with modified VLPs. Mice were immunised with a commercial HBV vaccine, monitored to ensure an anti-HBs response, then immunised with HBsAg/HVR1-VLPs. The resulting anti-HVR1 antibody titre was similar in mice with or without pre-existing anti-HBs. This suggests that HBsAg/HVR1-VLPs induce a primary immune response to HVR1 in anti-HBs positive mice and, hence, they may be used successfully in individuals already immunised with the HBV vaccine. (C) 2003 Elsevier Science Ltd. All rights reserved.

Quantum computation with optical coherent states

We show that quantum computation circuits using coherent states as the logical qubits can be constructed from simple linear networks, conditional photon measurements, and "small" coherent superposition resource states.

Immunopathology and Pathophysiology of Experimental Autoimmune Encephalomyelitis

Experimental autoimmune encephalomyelitis (EAE) is an inflammatory demyelinating disease of the central nervous system (CNS), and is widely studied as an animal model of the human CNS demyelinating diseases, including multiple sclerosis (Raine, 1984). EAE can be induced by inoculation with whole CNS tissue, purified myelin basic protein (MBP) or myelin proteolipid protein (PLP), together with adjuvants. It may also be induced by the passive transfer of T cells specifically reactive to these myelin antigens. EAE may have either an acute or a chronic relapsing course. Acute EAE closely resembles the human disease acute disseminated encephalomyelitis, while chronic relapsing EAE resembles multiple sclerosis. EAE is also the prototype for T-cell-mediated autoimmune disease in general. This chapter will focus on the immunopathology and pathophysiology of EAE, which are the subjects of investigation in my laboratory.

G551D CF mice display an abnormal host response and have impaired clearance of Pseudomonas lung disease

Several cystic fibrosis (CF) mouse models demonstrate an increased susceptibility to Pseudomonas aeruginosa lung infection, characterized by excessive inflammation and high rates of mortality. Here we developed a model of chronic P. aeruginosa lung disease in mice homozygous for the murine CF transmembrane conductance regulator G551D mutation that provides an excellent model for CF lung disease. After 3 days of infection with mucoid P. aeruginosa entrapped in agar beads, the G551D animals lost substantially more body weight than non-CF control animals and were less able to control the infection, harboring over 40-fold more bacteria in the lung. The airways of infected G551D animals contained altered concentrations of the inflammatory mediators tumor necrosis factor-alpha, KC/N51, and macrophage inflammatory protein-2 during the first 2 days of infection, suggesting that an ineffective inflammatory response is partly responsible for the clearance defect.

Evaluation of unsupervised semantic mapping of natural language with Leximancer concept mapping

The Leximancer system is a relatively new method for transforming lexical co-occurrence information from natural language into semantic patterns in an unsupervised manner. It employs two stages of co-occurrence information extraction-semantic and relational-using a different algorithm for each stage. The algorithms used are statistical, but they employ nonlinear dynamics and machine learning. This article is an attempt to validate the output of Leximancer, using a set of evaluation criteria taken from content analysis that are appropriate for knowledge discovery tasks.

Prefrontal cortex involvement in selective letter generation: A functional magnetic resonance imaging study

Cerebral responses to alternating periods of a control task and a selective letter generation paradigm were investigated with functional Magnetic Resonance Imaging (fMRI). Subjects selectively generated letters from four designated sets of six letters from the English language alphabet, with the instruction that they were not to produce letters in alphabetical order either forward or backward, repeat or alternate letters. Performance during this condition was compared with that of a control condition in which subjects recited the same letters in alphabetical order. Analyses revealed significant and extensive foci of activation in a number of cerebral regions including mid-dorsolateral frontal cortex, inferior frontal gyrus, precuneus, supramarginal gyrus, and cerebellum during the selective letter generation condition. These findings are discussed with respect to recent positron emission tomography (PET) and fMRI studies of verbal working memory and encoding/retrieval in episodic memory.