A Genetic Algorithm for the selection of structural MRI features for classification of Mild Cognitive Impairment and Alzheimer's Disease

This work investigates the problem of feature selection in neuroimaging features from structural MRI brain images for the classification of subjects as healthy controls, suffering from Mild Cognitive Impairment or Alzheimer’s Disease. A Genetic Algorithm wrapper method for feature selection is adopted in conjunction with a Support Vector Machine classifier. In very large feature sets, feature selection is found to be redundant as the accuracy is often worsened when compared to an Support Vector Machine with no feature selection. However, when just the hippocampal subfields are used, feature selection shows a significant improvement of the classification accuracy. Three-class Support Vector Machines and two-class Support Vector Machines combined with weighted voting are also compared with the former and found more useful. The highest accuracy achieved at classifying the test data was 65.5% using a genetic algorithm for feature selection with a three-class Support Vector Machine classifier.

Soft balancing in the Americas: Latin American opposition to U.S. intervention, 1898–1936

In the aftermath of the 2003 U.S.-led invasion of Iraq, scholars of international relations debated how to best characterize the rising tide of global opposition. The concept of “soft balancing” emerged as an influential, though contested, explanation of a new phenomenon in a unipolar world: states seeking to constrain the ability of the United States to deploy military force by using multinational organizations, international law, and coalition building. Soft balancing can also be observed in regional unipolar systems. Multinational archival research reveals how Argentina, Mexico, and other Latin American countries responded to expanding U.S. power and military assertiveness in the early twentieth century through coordinated diplomatic maneuvering that provides a strong example of soft balancing. Examination of this earlier case makes an empirical contribution to the emerging soft-balancing literature and suggests that soft balancing need not lead to hard balancing or open conflict.

Modelling personal thermal sensations using C-Support Vector Classification (C-SVC) algorithm

The personalised conditioning system (PCS) is widely studied. Potentially, it is able to reduce energy consumption while securing occupants’ thermal comfort requirements. It has been suggested that automatic optimised operation schemes for PCS should be introduced to avoid energy wastage and discomfort caused by inappropriate operation. In certain automatic operation schemes, personalised thermal sensation models are applied as key components to help in setting targets for PCS operation. In this research, a novel personal thermal sensation modelling method based on the C-Support Vector Classification (C-SVC) algorithm has been developed for PCS control. The personal thermal sensation modelling has been regarded as a classification problem. During the modelling process, the method ‘learns’ an occupant’s thermal preferences from his/her feedback, environmental parameters and personal physiological and behavioural factors. The modelling method has been verified by comparing the actual thermal sensation vote (TSV) with the modelled one based on 20 individual cases. Furthermore, the accuracy of each individual thermal sensation model has been compared with the outcomes of the PMV model. The results indicate that the modelling method presented in this paper is an effective tool to model personal thermal sensations and could be integrated within the PCS for optimised system operation and control.

A constrained recursive least squares algorithm for adaptive combination of multiple models

In this paper, we develop a novel constrained recursive least squares algorithm for adaptively combining a set of given multiple models. With data available in an online fashion, the linear combination coefficients of submodels are adapted via the proposed algorithm.We propose to minimize the mean square error with a forgetting factor, and apply the sum to one constraint to the combination parameters. Moreover an l1-norm constraint to the combination parameters is also applied with the aim to achieve sparsity of multiple models so that only a subset of models may be selected into the final model. Then a weighted l2-norm is applied as an approximation to the l1-norm term. As such at each time step, a closed solution of the model combination parameters is available. The contribution of this paper is to derive the proposed constrained recursive least squares algorithm that is computational efficient by exploiting matrix theory. The effectiveness of the approach has been demonstrated using both simulated and real time series examples.

Gaze stabilization testing: The efficiency and test-retest reliability of user operated control versus the parameter estimation (PEST) algorithm

The primary objective of this research study is to determine which form of testing, the PEST algorithm or an operator-controlled condition is most accurate and time efficient for administration of the gaze stabilization test

Digital image correlation analysis of the load transfer by implant-supported restorations

This study compared splinted and non-splinted implant-supported prosthesis with and without a distal proximal contact using a digital image correlation method. An epoxy resin model was made with acrylic resin replicas of a mandibular first premolar and second molar and with threaded implants replacing the second premolar and first molar. Splinted and non-splinted metal-ceramic screw-retained crowns were fabricated and loaded with and without the presence of the second molar. A single-camera measuring system was used to record the in-plane deformation on the model surface at a frequency of 1.0 Hz under a load from 0 to 250 N. The images were then analyzed with specialist software to determine the direct (horizontal) and shear strains along the model. Not splinting the crowns resulted in higher stress transfer to the supporting implants when the second molar replica was absent. The presence of a second molar and an effective interproximal contact contributed to lower stress transfer to the supporting structures even for non-splinted restorations. Shear strains were higher in the region between the molars when the second molar was absent, regardless of splinting. The opposite was found for the region between the implants, which had higher shear strain values when the second molar was present. When an effective distal contact is absent, non-splinted implant-supported restorations introduce higher direct strains to the supporting structures under loading. Shear strains appear to be dependent also on the region within the model, with different regions showing different trends in strain changes in the absence of an effective distal contact. (C) 2011 Elsevier Ltd. All rights reserved.

Balancing the role of the dental school in teaching, research and patient care; including care for underserved areas

Inequalities within dentistry are common and are reflected in wide differences in the levels of oral health and the standard of care available both within and between countries and communities. Furthermore there are patients, particularly those with special treatment needs, who do not have the same access to dental services as the general public. The dental school should aim to recruit students from varied backgrounds into all areas covered by the oral healthcare team and to train students to treat the full spectrum of patients including those with special needs. It is essential, however, that the dental student achieves a high standard of clinical competence and this cannot be gained by treating only those patients with low expectations for care. Balancing these aspects of clinical education is difficult. Research is an important stimulus to better teaching and better clinical care. It is recognized that dental school staff should be active in research, teaching, clinical work and frequently administration. Maintaining a balance between the commitments to clinical care, teaching and research while also taking account of underserved areas in each of these categories is a difficult challenge but one that has to be met to a high degree in a successful, modern dental school.

DC traction load flow including AC distribution network

In the first part some information and characterisation about an AC distribution network that feeds traction substations and their possible influences on the DC traction load flow are presented. Those influences are investigated and mathematically modelled. To corroborate the mathematical model, an example is presented and their results are confronted with real measurements.

The SARS algorithm: detrending CoRoT light curves with Sysrem using simultaneous external parameters

Surveys for exoplanetary transits are usually limited not by photon noise but rather by the amount of red noise in their data. In particular, although the CoRoT space-based survey data are being carefully scrutinized, significant new sources of systematic noises are still being discovered. Recently, a magnitude-dependant systematic effect was discovered in the CoRoT data by Mazeh et al. and a phenomenological correction was proposed. Here we tie the observed effect to a particular type of effect, and in the process generalize the popular Sysrem algorithm to include external parameters in a simultaneous solution with the unknown effects. We show that a post-processing scheme based on this algorithm performs well and indeed allows for the detection of new transit-like signals that were not previously detected.

Genetic algorithm inversion of the average 1D crustal structure using local and regional earthquakes

Knowing the best 1D model of the crustal and upper mantle structure is useful not only for routine hypocenter determination, but also for linearized joint inversions of hypocenters and 3D crustal structure, where a good choice of the initial model can be very important. Here, we tested the combination of a simple GA inversion with the widely used HYPO71 program to find the best three-layer model (upper crust, lower crust, and upper mantle) by minimizing the overall P- and S-arrival residuals, using local and regional earthquakes in two areas of the Brazilian shield. Results from the Tocantins Province (Central Brazil) and the southern border of the Sao Francisco craton (SE Brazil) indicated an average crustal thickness of 38 and 43 km, respectively, consistent with previous estimates from receiver functions and seismic refraction lines. The GA + HYPO71 inversion produced correct Vp/Vs ratios (1.73 and 1.71, respectively), as expected from Wadati diagrams. Tests with synthetic data showed that the method is robust for the crustal thickness, Pn velocity, and Vp/Vs ratio when using events with distance up to about 400 km, despite the small number of events available (7 and 22, respectively). The velocities of the upper and lower crusts, however, are less well constrained. Interestingly, in the Tocantins Province, the GA + HYPO71 inversion showed a secondary solution (local minimum) for the average crustal thickness, besides the global minimum solution, which was caused by the existence of two distinct domains in the Central Brazil with very different crustal thicknesses. (C) 2010 Elsevier Ltd. All rights reserved.

Variety matters: adaptive genetic diversity and parasite load in two mouse opossums from the Brazilian Atlantic forest

The adaptive potential of a species to a changing environment and in disease defence is primarily based on genetic variation. Immune genes, such as genes of the major histocompatibility complex (MHC), may thereby be of particular importance. In marsupials, however, there is very little knowledge about natural levels and functional importance of MHC polymorphism, despite their key role in the mammalian evolution. In a previous study, we discovered remarkable differences in the MHC class II diversity between two species of mouse opossums (Gracilinanus microtarsus, Marmosops incanus) from the Brazilian Atlantic forest, which is one of the most endangered hotspots for biodiversity conservation. Since the main forces in generating MHC diversity are assumed to be pathogens, we investigated in this study gastrointestinal parasite burden and functional associations between the individual MHC constitution and parasite load. We tested two contrasting scenarios, which might explain differences in MHC diversity between species. We predicted that a species with low MHC diversity would either be under relaxed selection pressure by low parasite diversity (`Evolutionary equilibrium` scenario), or there was a recent loss in MHC diversity leading to a lack of resistance alleles and increased parasite burden (`Unbalanced situation` scenario). In both species it became apparent that the MHC class II is functionally important in defence against gastrointestinal helminths, which was shown here for the first time in marsupials. On the population level, parasite diversity did not markedly differ between the two host species. However, we did observe considerable differences in the individual parasite load (parasite prevalence and infection intensity): while M. incanus revealed low MHC DAB diversity and high parasite load, G. microtarsus showed a tenfold higher population wide MHC DAB diversity and lower parasite burden. These results support the second scenario of an unbalanced situation.

Inflammation and adipose tissue: effects of progressive load training in rats

Introduction: Cytokines (IL-6, IL-10 and TNF-alpha) are increased after exhaustive exercise in the rat retroperitoneal (RPAT) and mesenteric adipose tissue (MEAT) pads. On the other hand, these cytokines show decreased expression in these depots in response to a chronic exercise protocol. However, the effect of exercise with overload combined with a short recovery period on pro-and anti-inflammatory cytokine expression is unknown. In the present study, we investigated the regulation of cytokine production in the adipose tissue of rats after an overtraining-inducing exercise protocol. Methods: Male Wistar rats were divided into four groups: Control (C), Trained (Tr), Overtrained (OT) and recovered overtrained (R). Cytokines (IL-6, TNF-alpha and IL-10) levels and Toll Like Receptor 4 (TLR4), Nuclear Factor kBBp65 (NF-kBp65), Hormone Sensitive Lipase (HSL) and, Perilipin protein expression were assessed in the adipose tissue. Furthermore, we analysed plasma lipid profile, insulin, testosterone, corticosterone and endotoxin levels, and liver triacylglycerol, cytokine content, as well as apolipoprotein B (apoB) and TLR4 expression in the liver. Results: OT and R groups exhibited reduced performance accompanied by lower testosterone and increased corticosterone and endotoxin levels when compared with the control and trained groups. IL-6 and IL-10 protein levels were increased in the adipose tissue of the group allowed to recover, in comparison with all the other studied groups. TLR-4 and NF-kBp65 were increased in this same group when compared with both control and trained groups. The protein expression of HSL was increased and that of Perilipin, decreased in the adipose in R in relation to the control. In addition, we found increased liver and serum TAG, along with reduced apoB protein expression and IL-6 and IL-10 levels in the of R in relation to the control and trained groups. Conclusion: In conclusion, we have shown that increases in pro-inflammatory cytokines in the adipose tissue after an overtraining protocol may be mediated via TLR-4 and NF-kBp65 signalling, leading to an inflammatory state in this tissue.

Apoptosis of macrophages during pulmonary Mycobacterium bovis infection: correlation with intracellular bacillary load and cytokine levels

P>Apoptosis of macrophages infected with pathogenic mycobacteria is an alternative host defence capable of removing the environment supporting bacterial growth. In this work the influence of virulence and bacterial load on apoptosis of alveolar macrophages during the initial phase of infection by Mycobacterium bovis was investigated. BALB/c mice were infected intratracheally with high or low doses of the virulent (ATCC19274) or attenuated (bacillus Calmette-Guerin Moreau) strains of M. bovis. The frequency of macrophage apoptosis, the growth of mycobacteria in macrophages, and the in situ levels of the cytokines tumour necrosis factor-alpha (TNF-alpha), interleukin-10 (IL-10) and IL-12 and of the anti-apoptotic protein Bcl-2 were measured at day 3 and day 7 post-infection. An increase of macrophage apoptosis was observed after infection with both strains but the virulent strain induced less apoptosis than the attenuated strain. On the 3rd day after infection with the virulent strain macrophage apoptosis was reduced in the high-dose group, while on the 7th day post-infection macrophage apoptosis was reduced in the low-dose group. Inhibition of apoptosis was correlated with increased production of IL-10, reduced production of TNF-alpha and increased production of Bcl-2. In addition, the production of IL-12 was reduced at points where the lowest levels of macrophage apoptosis were observed. Our results indicate that virulent mycobacteria are able to modulate macrophage apoptosis to an extent dependent on the intracellular bacterial burden, which benefits its intracellular growth and dissemination to adjacent cells.

Clinical correlations of human cytomegalovirus strains and viral load in kidney transplant recipients

Little is known about clinical differences associated with cytomegalovirus (CMV) infection by distinct strains in renal transplant patients. Different clinical pictures may be associated with specific viral genotypes. viral load, as well as host factors. The objective of this study was to identify CMV strains to determine viral load (antigenemia), and their correlation with clinical data in renal transplant recipients. Seventy-one patients were enrolled, comprising 91 samples. After selection, polymorphonuclear cells were used to amplify and sequence the gB region of CMV DNA. The sequences were analyzed to ascertain the frequency of different genotypes. Additionally, the results of this Study showed that the gB coding gene presents a great variability, revealing a variety of patterns: classical gB (1.4%), gB1V (46.4%), classical gB2 (35.2%), gB2V (2.8%), gB3 (1.4%), classical gB4 (4.9%) and gB4V (4.9%). The mean viral load in kidney transplant patient was 75.1 positive cells (1-1000). A higher viral load was observed in patients with genotype 4 infection. Statistically significant differences were detected between gB1 and gB4 (p=0.010), and between gB2 and gB4 (p=0.021). The average numbers of positive cells in relation to clinical presentation were: 34.5 in asymptomatic, 49.5 in CMV associated syndrome and 120.7 in patients with invasive disease (p=0.048). As a group, gB1 was the most frequent strain and revealed a potential risk for developing invasive disease. Viral load also seemed to be important as a marker associated with clinical presentation of the disease. (C) 2008 Elsevier B.V. All rights reserved.

Bayesian network classifiers: Beyond classification accuracy

This work proposes and discusses an approach for inducing Bayesian classifiers aimed at balancing the tradeoff between the precise probability estimates produced by time consuming unrestricted Bayesian networks and the computational efficiency of Naive Bayes (NB) classifiers. The proposed approach is based on the fundamental principles of the Heuristic Search Bayesian network learning. The Markov Blanket concept, as well as a proposed ""approximate Markov Blanket"" are used to reduce the number of nodes that form the Bayesian network to be induced from data. Consequently, the usually high computational cost of the heuristic search learning algorithms can be lessened, while Bayesian network structures better than NB can be achieved. The resulting algorithms, called DMBC (Dynamic Markov Blanket Classifier) and A-DMBC (Approximate DMBC), are empirically assessed in twelve domains that illustrate scenarios of particular interest. The obtained results are compared with NB and Tree Augmented Network (TAN) classifiers, and confinn that both proposed algorithms can provide good classification accuracies and better probability estimates than NB and TAN, while being more computationally efficient than the widely used K2 Algorithm.