Multiple Kernel Learning of Environmental Data. Case study: Analysis and Mapping of Wind Fields

The paper presents the Multiple Kernel Learning (MKL) approach as a modelling and data exploratory tool and applies it to the problem of wind speed mapping. Support Vector Regression (SVR) is used to predict spatial variations of the mean wind speed from terrain features (slopes, terrain curvature, directional derivatives) generated at different spatial scales. Multiple Kernel Learning is applied to learn kernels for individual features and thematic feature subsets, both in the context of feature selection and optimal parameters determination. An empirical study on real-life data confirms the usefulness of MKL as a tool that enhances the interpretability of data-driven models.

Advanced geostatistical and machine-learning models for spatial data analysis of radioactively contaminated regions

Radioactive soil-contamination mapping and risk assessment is a vital issue for decision makers. Traditional approaches for mapping the spatial concentration of radionuclides employ various regression-based models, which usually provide a single-value prediction realization accompanied (in some cases) by estimation error. Such approaches do not provide the capability for rigorous uncertainty quantification or probabilistic mapping. Machine learning is a recent and fast-developing approach based on learning patterns and information from data. Artificial neural networks for prediction mapping have been especially powerful in combination with spatial statistics. A data-driven approach provides the opportunity to integrate additional relevant information about spatial phenomena into a prediction model for more accurate spatial estimates and associated uncertainty. Machine-learning algorithms can also be used for a wider spectrum of problems than before: classification, probability density estimation, and so forth. Stochastic simulations are used to model spatial variability and uncertainty. Unlike regression models, they provide multiple realizations of a particular spatial pattern that allow uncertainty and risk quantification. This paper reviews the most recent methods of spatial data analysis, prediction, and risk mapping, based on machine learning and stochastic simulations in comparison with more traditional regression models. The radioactive fallout from the Chernobyl Nuclear Power Plant accident is used to illustrate the application of the models for prediction and classification problems. This fallout is a unique case study that provides the challenging task of analyzing huge amounts of data ('hard' direct measurements, as well as supplementary information and expert estimates) and solving particular decision-oriented problems.

Analysis and spatial modeling of the indoor radon Swiss data

The present research deals with an important public health threat, which is the pollution created by radon gas accumulation inside dwellings. The spatial modeling of indoor radon in Switzerland is particularly complex and challenging because of many influencing factors that should be taken into account. Indoor radon data analysis must be addressed from both a statistical and a spatial point of view. As a multivariate process, it was important at first to define the influence of each factor. In particular, it was important to define the influence of geology as being closely associated to indoor radon. This association was indeed observed for the Swiss data but not probed to be the sole determinant for the spatial modeling. The statistical analysis of data, both at univariate and multivariate level, was followed by an exploratory spatial analysis. Many tools proposed in the literature were tested and adapted, including fractality, declustering and moving windows methods. The use of Quan-tité Morisita Index (QMI) as a procedure to evaluate data clustering in function of the radon level was proposed. The existing methods of declustering were revised and applied in an attempt to approach the global histogram parameters. The exploratory phase comes along with the definition of multiple scales of interest for indoor radon mapping in Switzerland. The analysis was done with a top-to-down resolution approach, from regional to local lev¬els in order to find the appropriate scales for modeling. In this sense, data partition was optimized in order to cope with stationary conditions of geostatistical models. Common methods of spatial modeling such as Κ Nearest Neighbors (KNN), variography and General Regression Neural Networks (GRNN) were proposed as exploratory tools. In the following section, different spatial interpolation methods were applied for a par-ticular dataset. A bottom to top method complexity approach was adopted and the results were analyzed together in order to find common definitions of continuity and neighborhood parameters. Additionally, a data filter based on cross-validation was tested with the purpose of reducing noise at local scale (the CVMF). At the end of the chapter, a series of test for data consistency and methods robustness were performed. This lead to conclude about the importance of data splitting and the limitation of generalization methods for reproducing statistical distributions. The last section was dedicated to modeling methods with probabilistic interpretations. Data transformation and simulations thus allowed the use of multigaussian models and helped take the indoor radon pollution data uncertainty into consideration. The catego-rization transform was presented as a solution for extreme values modeling through clas-sification. Simulation scenarios were proposed, including an alternative proposal for the reproduction of the global histogram based on the sampling domain. The sequential Gaussian simulation (SGS) was presented as the method giving the most complete information, while classification performed in a more robust way. An error measure was defined in relation to the decision function for data classification hardening. Within the classification methods, probabilistic neural networks (PNN) show to be better adapted for modeling of high threshold categorization and for automation. Support vector machines (SVM) on the contrary performed well under balanced category conditions. In general, it was concluded that a particular prediction or estimation method is not better under all conditions of scale and neighborhood definitions. Simulations should be the basis, while other methods can provide complementary information to accomplish an efficient indoor radon decision making.

Robust vaccine-elicited cellular immune responses in breast milk following systemic Simian Immunodeficiency Virus DNA prime and live virus vector boost vaccination of lactating rhesus monkeys.

Breast milk transmission of HIV remains an important mode of infant HIV acquisition. Enhancement of mucosal HIV-specific immune responses in milk of HIV-infected mothers through vaccination may reduce milk virus load or protect against virus transmission in the infant gastrointestinal tract. However, the ability of HIV/SIV strategies to induce virus-specific immune responses in milk has not been studied. In this study, five uninfected, hormone-induced lactating, Mamu A*01(+) female rhesus monkey were systemically primed and boosted with rDNA and the attenuated poxvirus vector, NYVAC, containing the SIVmac239 gag-pol and envelope genes. The monkeys were boosted a second time with a recombinant Adenovirus serotype 5 vector containing matching immunogens. The vaccine-elicited immunodominant epitope-specific CD8(+) T lymphocyte response in milk was of similar or greater magnitude than that in blood and the vaginal tract but higher than that in the colon. Furthermore, the vaccine-elicited SIV Gag-specific CD4(+) and CD8(+) T lymphocyte polyfunctional cytokine responses were more robust in milk than in blood after each virus vector boost. Finally, SIV envelope-specific IgG responses were detected in milk of all monkeys after vaccination, whereas an SIV envelope-specific IgA response was only detected in one vaccinated monkey. Importantly, only limited and transient increases in the proportion of activated or CCR5-expressing CD4(+) T lymphocytes in milk occurred after vaccination. Therefore, systemic DNA prime and virus vector boost of lactating rhesus monkeys elicits potent virus-specific cellular and humoral immune responses in milk and may warrant further investigation as a strategy to impede breast milk transmission of HIV.

Detection and chemical profiling of medicine counterfeits by Raman spectroscopy and chemometrics

Raman spectroscopy combined with chemometrics has recently become a widespread technique for the analysis of pharmaceutical solid forms. The application presented in this paper is the investigation of counterfeit medicines. This increasingly serious issue involves networks that are an integral part of industrialized organized crime. Efficient analytical tools are consequently required to fight against it. Quick and reliable authentication means are needed to allow the deployment of measures from the company and the authorities. For this purpose a method in two steps has been implemented here. The first step enables the identification of pharmaceutical tablets and capsules and the detection of their counterfeits. A nonlinear classification method, the Support Vector Machines (SVM), is computed together with a correlation with the database and the detection of Active Pharmaceutical Ingredient (API) peaks in the suspect product. If a counterfeit is detected, the second step allows its chemical profiling among former counterfeits in a forensic intelligence perspective. For this second step a classification based on Principal Component Analysis (PCA) and correlation distance measurements is applied to the Raman spectra of the counterfeits.

2

The quality of environmental data analysis and propagation of errors are heavily affected by the representativity of the initial sampling design [CRE 93, DEU 97, KAN 04a, LEN 06, MUL07]. Geostatistical methods such as kriging are related to field samples, whose spatial distribution is crucial for the correct detection of the phenomena. Literature about the design of environmental monitoring networks (MN) is widespread and several interesting books have recently been published [GRU 06, LEN 06, MUL 07] in order to clarify the basic principles of spatial sampling design (monitoring networks optimization) based on Support Vector Machines was proposed. Nonetheless, modelers often receive real data coming from environmental monitoring networks that suffer from problems of non-homogenity (clustering). Clustering can be related to the preferential sampling or to the impossibility of reaching certain regions.

Expression of Staphylococcus aureus clumping factor A in Lactococcus lactis subsp. cremoris using a new shuttle vector.

Staphylococcus aureus harbors redundant adhesins mediating tissue colonization and infection. To evaluate their intrinsic role outside of the staphylococcal background, a system was designed to express them in Lactococcus lactis subsp. cremoris 1363. This bacterium is devoid of virulence factors and has a known genetic background. A new Escherichia coli-L. lactis shuttle and expression vector was constructed for this purpose. First, the high-copy-number lactococcal plasmid pIL253 was equipped with the oriColE1 origin, generating pOri253 that could replicate in E. coli. Second, the lactococcal promoters P23 or P59 were inserted at one end of the pOri253 multicloning site. Gene expression was assessed by a luciferase reporter system. The plasmid carrying P23 (named pOri23) expressed luciferase constitutively at a level 10,000 times greater than did the P59-containing plasmid. Transcription was absent in E. coli. The staphylococcal clumping factor A (clfA) gene was cloned into pOri23 and used as a model system. Lactococci carrying pOri23-clfA produced an unaltered and functional 130-kDa ClfA protein attached to their cell walls. This was indicated both by the presence of the protein in Western blots of solubilized cell walls and by the ability of ClfA-positive lactococci to clump in the presence of plasma. ClfA-positive lactococci had clumping titers (titer of 4,112) similar to those of S. aureus Newman in soluble fibrinogen and bound equally well to solid-phase fibrinogen. These experiments provide a new way to study individual staphylococcal pathogenic factors and might complement both classical knockout mutagenesis and modern in vivo expression technology and signature tag mutagenesis.

Women's decision making and experience of subsequent pregnancy following stillbirth.

INTRODUCTION: This study sought to increase understanding of women's thoughts and feelings about decision making and the experience of subsequent pregnancy following stillbirth (intrauterine death after 24 weeks' gestation). METHODS: Eleven women were interviewed, 8 of whom were pregnant at the time of the interview. Modified grounded theory was used to guide the research methodology and to analyze the data. RESULTS: A model was developed to illustrate women's experiences of decision making in relation to subsequent pregnancy and of subsequent pregnancy itself. DISCUSSION: The results of the current study have significant implications for women who have experienced stillbirth and the health professionals who work with them. Based on the model, women may find it helpful to discuss their beliefs in relation to healing and health professionals to provide support with this in mind. Women and their partners may also benefit from explanations and support about the potentially conflicting emotions they may experience during this time.

Adhésion thérapeutique aux traitements oncologiques oraux et prise en charge interdisciplinaire [Therapeutic adherence to oral cancer therapy and interdisciplinary management].

Medication adherence is a well-known risk factor in internal medicine. However in oncology this dimension is emerging due to the increasing number of oral formulations. First results in the oral oncology literature suggest that patients' ability to cope with medical prescription decreases with time. This might preclude patients from reaching clinical outcomes. Factors impacting on medication adherence to oral oncology treatments have not been yet extensively described neither strategies to address them and support patient's needs. Oncologists and pharmacists in our University outpatient settings performed a pilot study which aimed at measuring and facilitating adherence to oral oncology treatments and at understanding determinants of patient's adherence. The ultimate purpose of such a patient-centered and interdisciplinary collaboration would be to promote patient self-management and complement the standard medical follow-up.

Automatic Mallampati Classification Using Active Appearance Models

Difficult tracheal intubation assessment is an important research topic in anesthesia as failed intubations are important causes of mortality in anesthetic practice. The modified Mallampati score is widely used, alone or in conjunction with other criteria, to predict the difficulty of intubation. This work presents an automatic method to assess the modified Mallampati score from an image of a patient with the mouth wide open. For this purpose we propose an active appearance models (AAM) based method and use linear support vector machines (SVM) to select a subset of relevant features obtained using the AAM. This feature selection step proves to be essential as it improves drastically the performance of classification, which is obtained using SVM with RBF kernel and majority voting. We test our method on images of 100 patients undergoing elective surgery and achieve 97.9% accuracy in the leave-one-out crossvalidation test and provide a key element to an automatic difficult intubation assessment system.

Perinatal care at the limit of viability between 22 and 26 completed weeks of gestation in Switzerland.

Perinatal care of pregnant women at high risk for preterm delivery and of preterm infants born at the limit of viability (22-26 completed weeks of gestation) requires a multidisciplinary approach by an experienced perinatal team. Limited precision in the determination of both gestational age and foetal weight, as well as biological variability may significantly affect the course of action chosen in individual cases. The decisions that must be taken with the pregnant women and on behalf of the preterm infant in this context are complex and have far-reaching consequences. When counselling pregnant women and their partners, neonatologists and obstetricians should provide them with comprehensive information in a sensitive and supportive way to build a basis of trust. The decisions are developed in a continuing dialogue between all parties involved (physicians, midwives, nursing staff and parents) with the principal aim to find solutions that are in the infant's and pregnant woman's best interest. Knowledge of current gestational age-specific mortality and morbidity rates and how they are modified by prenatally known prognostic factors (estimated foetal weight, sex, exposure or nonexposure to antenatal corticosteroids, single or multiple births) as well as the application of accepted ethical principles form the basis for responsible decision-making. Communication between all parties involved plays a central role. The members of the interdisciplinary working group suggest that the care of preterm infants with a gestational age between 22 0/7 and 23 6/7 weeks should generally be limited to palliative care. Obstetric interventions for foetal indications such as Caesarean section delivery are usually not indicated. In selected cases, for example, after 23 weeks of pregnancy have been completed and several of the above mentioned prenatally known prognostic factors are favourable or well informed parents insist on the initiation of life-sustaining therapies, active obstetric interventions for foetal indications and provisional intensive care of the neonate may be reasonable. In preterm infants with a gestational age between 24 0/7 and 24 6/7 weeks, it can be difficult to determine whether the burden of obstetric interventions and neonatal intensive care is justified given the limited chances of success of such a therapy. In such cases, the individual constellation of prenatally known factors which impact on prognosis can be helpful in the decision making process with the parents. In preterm infants with a gestational age between 25 0/7 and 25 6/7 weeks, foetal surveillance, obstetric interventions for foetal indications and neonatal intensive care measures are generally indicated. However, if several prenatally known prognostic factors are unfavourable and the parents agree, primary non-intervention and neonatal palliative care can be considered. All pregnant women with threatening preterm delivery or premature rupture of membranes at the limit of viability must be transferred to a perinatal centre with a level III neonatal intensive care unit no later than 23 0/7 weeks of gestation, unless emergency delivery is indicated. An experienced neonatology team should be involved in all deliveries that take place after 23 0/7 weeks of gestation to help to decide together with the parents if the initiation of intensive care measures appears to be appropriate or if preference should be given to palliative care (i.e., primary non-intervention). In doubtful situations, it can be reasonable to initiate intensive care and to admit the preterm infant to a neonatal intensive care unit (i.e., provisional intensive care). The infant's clinical evolution and additional discussions with the parents will help to clarify whether the life-sustaining therapies should be continued or withdrawn. Life support is continued as long as there is reasonable hope for survival and the infant's burden of intensive care is acceptable. If, on the other hand, the health care team and the parents have to recognise that in the light of a very poor prognosis the burden of the currently used therapies has become disproportionate, intensive care measures are no longer justified and other aspects of care (e.g., relief of pain and suffering) are the new priorities (i.e., redirection of care). If a decision is made to withhold or withdraw life-sustaining therapies, the health care team should focus on comfort care for the dying infant and support for the parents.

The gamma-crystallins and human cataracts: a puzzle made clearer.

Despite the fact that cataracts constitute the leading cause of blindness worldwide, the mechanisms of lens opacification remain unclear. We recently mapped the aculeiform cataract to the gamma-crystallin locus (CRYG) on chromosome 2q33-35, and mutational analysis of the CRYG-genes cluster identified the aculeiform-cataract mutation in exon 2 of gamma-crystallin D (CRYGD). This mutation occurred in a highly conserved amino acid and could be associated with an impaired folding of CRYGD. During our study, we observed that the previously reported Coppock-like-cataract mutation, the first human cataract mutation, in the pseudogene CRYGE represented a polymorphism seen in 23% of our control population. Further analysis of the original Coppock-like-cataract family identified a missense mutation in a highly conserved segment of exon 2 of CRYGC. These mutations were not seen in a large control population. There is no direct evidence, to date, that up-regulation of a pseudogene causes cataracts. To our knowledge, these findings are the first evidence of an involvement of CRYGC and support the role of CRYGD in human cataract formation.

Uncovering a role for micro-RNAs in sleep homeostasis and energy metabolism

Micro-RNAs (miRNAs) are key, post-transcriptional regulators of gene expression and have been implicated in almost every cellular process investigated thus far. However, their role in sleep, in particular the homeostatic aspect of sleep control, has received little attention. We here assessed the effects of sleep deprivation on the brain miRNA transcriptome in the mouse. Sleep deprivation affected miRNA expression in a brain-region specific manner. The forebrain expression of the miRNA miR-709 was affected the most and in situ analyses confirmed its robust increase throughout the brain, especially in the cerebral cortex and the hippocampus. The hippocampus was a major target of the sleep deprivation affecting 37 miRNAs compared to 52 in the whole forebrain. Moreover, independent from the sleep deprivation condition, miRNA expression was highly region-specific with 45% of all expressed miRNAs showing higher expression in hippocampus and 55% in cortex. Next we demonstrated that down-regulation of miRNAs in Com/c2o-expressing neurons of adult mice, through a conditional and inducible Dicer knockout mice model (cKO), results in an altered homeostatic response after sleep deprivation eight weeks following the tamoxifen-induced recombination. Dicer cKO mice showed a larger increase in the electro-encephalographic (EEG) marker of sleep pressure, EEG delta power, and a reduced Rapid Eye Movement sleep rebound, compared to controls, highlighting a functional role of miRNAs in sleep homeostasis. Beside a sleep phenotype, Dicer cKO mice developed an unexpected, severe obesity phenotype associated with hyperphagia and altered metabolism. Even more surprisingly, after reaching maximum body weight 5 weeks after tamoxifen injection, obese cKO mice spontaneously started losing weight as rapidly as it was gained. Brain transcriptome analyses in obese mice identified several obesity-related pathways (e.g. leptin, somatostatin, and nemo-like kinase signaling), as well as genes involved in feeding and appetite (e.g. Pmch, Neurotensin). A gene cluster with anti-correlated expression in the cerebral cortex of post-obese compared to obese mice was enriched for synaptic plasticity pathways. While other studies have identified a role for miRNAs in obesity, we here present a unique model that allows for the study of processes involved in reversing obesity. Moreover, our study identified the cortex as a brain area important for body weight homeostasis. Together, these observations strongly suggest a role for miRNAs in the maintenance of homeostatic processes in the mouse, and support the hypothesis of a tight relationship between sleep and metabolism at a molecular - Les micro-ARNS (miARNs) sont des régulateurs post-transcriptionnels de l'expression des gènes, impliqués dans la quasi-totalité des processus cellulaires. Cependant, leur rôle dans la régulation du sommeil, et en particulier dans le maintien de l'homéostasie du sommeil, n'a reçu que très peu d'attention jusqu'à présent. Dans cette étude, nous avons étudié les conséquences d'une privation de sommeil sur l'expression cérébrale des miARNs chez la souris, et observé des changements dans l'expression de nombreux miARNs. Dans le cerveau antérieur, miR-709 est le miARN le plus affecté par la perte de sommeil, en particulier dans le cortex cérébral et l'hippocampe. L'hippocampe est la région la plus touchée avec 37 miARNs changés comparés à 52 dans le cerveau entier. Par ailleurs, indépendamment de la privation de sommeil, certains miARNs sont spécifiquement enrichis dans certaines aires cérébrales, 45% des miARNs étant surexprimés dans l'hippocampe contre 55% dans le cortex. Dans une seconde étude, nous avons observé que la délétion de DICER, enzyme essentielle à la biosynthèse des miARNs, et la perte subséquente des miARNs dans les neurones exprimant la protéine CAMK2a altère la réponse homéostatique à une privation de sommeil, 8 semaines après l'induction de la recombinaison génétique par le tamoxifen. Les souris sans Dicer (cKO) ont une plus large augmentation de l'EEG delta power, le principal marqueur électro-encéphalographique du besoin de sommeil, comparée aux contrôles, ainsi qu'un rebond en sommeil paradoxal plus petit. De façon surprenante, les souris Dicer cKO développent une obésité rapide, sévère et transitoire, associée à de l'hyperphagie et une altération de leur métabolisme énergétique. Après avoir atteint un pic maximal d'obésité, les souris cKO entrent spontanément dans une période de perte de poids rapide. L'analyse du transcriptome cérébral des souris obèses nous a permis d'identifier des voies associées à l'obésité (leptine, somatostatine et nemo-like kinase), et à la prise alimentaire (Pmch, Neurotensin), tandis que celui des souris post-obèses a révélé un groupe de gènes liés à la plasticité synaptique. Au-delà des nombreux modèles d'obésité existant chez la souris, notre étude présente un modèle unique permettant d'étudier les mécanismes sous-jacent la perte de poids. De plus, nous avons mis en évidence un rôle important du cortex cérébral dans le maintien de la balance énergétique. En conclusion, toutes ces observations soutiennent l'idée que les miARNs sont des régulateurs cruciaux dans le maintien des processus homéostatiques et confortent l'hypothèse d'une étroite relation moléculaire entre le sommeil et le métabolisme.

Hierarchy of Notch-Delta interactions promoting T cell lineage commitment and maturation.

Notch1 (N1) receptor signaling is essential and sufficient for T cell development, and recently developed in vitro culture systems point to members of the Delta family as being the physiological N1 ligands. We explored the ability of Delta1 (DL1) and DL4 to induce T cell lineage commitment and/or maturation in vitro and in vivo from bone marrow (BM) precursors conditionally gene targeted for N1 and/or N2. In vitro DL1 can trigger T cell lineage commitment via either N1 or N2. N1- or N2-mediated T cell lineage commitment can also occur in the spleen after short-term BM transplantation. However, N2-DL1-mediated signaling does not allow further T cell maturation beyond the CD25(+) stage due to a lack of T cell receptor beta expression. In contrast to DL1, DL4 induces and supports T cell commitment and maturation in vitro and in vivo exclusively via specific interaction with N1. Moreover, comparative binding studies show preferential interaction of DL4 with N1, whereas binding of DL1 to N1 is weak. Interestingly, preferential N1-DL4 binding reflects reduced dependence of this interaction on Lunatic fringe, a glycosyl transferase that generally enhances the avidity of Notch receptors for Delta ligands. Collectively, our results establish a hierarchy of Notch-Delta interactions in which N1-DL4 exhibits the greatest capacity to induce and support T cell development.

Active Learning of Very-High Resolution Optical Imagery with SVM: Entropy vs Margin Sampling

An active learning method is proposed for the semi-automatic selection of training sets in remote sensing image classification. The method adds iteratively to the current training set the unlabeled pixels for which the prediction of an ensemble of classifiers based on bagged training sets show maximum entropy. This way, the algorithm selects the pixels that are the most uncertain and that will improve the model if added in the training set. The user is asked to label such pixels at each iteration. Experiments using support vector machines (SVM) on an 8 classes QuickBird image show the excellent performances of the methods, that equals accuracies of both a model trained with ten times more pixels and a model whose training set has been built using a state-of-the-art SVM specific active learning method