Machine learning algorithms for spatial data. Case studies: Environmental pollution, natural hazards, renewable resources

This paper presents general problems and approaches for the spatial data analysis using machine learning algorithms. Machine learning is a very powerful approach to adaptive data analysis, modelling and visualisation. The key feature of the machine learning algorithms is that they learn from empirical data and can be used in cases when the modelled environmental phenomena are hidden, nonlinear, noisy and highly variable in space and in time. Most of the machines learning algorithms are universal and adaptive modelling tools developed to solve basic problems of learning from data: classification/pattern recognition, regression/mapping and probability density modelling. In the present report some of the widely used machine learning algorithms, namely artificial neural networks (ANN) of different architectures and Support Vector Machines (SVM), are adapted to the problems of the analysis and modelling of geo-spatial data. Machine learning algorithms have an important advantage over traditional models of spatial statistics when problems are considered in a high dimensional geo-feature spaces, when the dimension of space exceeds 5. Such features are usually generated, for example, from digital elevation models, remote sensing images, etc. An important extension of models concerns considering of real space constrains like geomorphology, networks, and other natural structures. Recent developments in semi-supervised learning can improve modelling of environmental phenomena taking into account on geo-manifolds. An important part of the study deals with the analysis of relevant variables and models' inputs. This problem is approached by using different feature selection/feature extraction nonlinear tools. To demonstrate the application of machine learning algorithms several interesting case studies are considered: digital soil mapping using SVM, automatic mapping of soil and water system pollution using ANN; natural hazards risk analysis (avalanches, landslides), assessments of renewable resources (wind fields) with SVM and ANN models, etc. The dimensionality of spaces considered varies from 2 to more than 30. Figures 1, 2, 3 demonstrate some results of the studies and their outputs. Finally, the results of environmental mapping are discussed and compared with traditional models of geostatistics.

Primary spinal epidural lymphoma: patients' profile, outcome, and prognostic factors: a multicenter Rare Cancer Network study

RESUME Objectif : Les lymphomes épiduraux primaires représentent moins de 10% des tumeurs épidurales et de 0,1 à 3,3% de tous les lymphomes. Le but de cette étude a été d'évaluer le profil clinique de cette maladie rare, son traitement, ses résultats ainsi que ses facteurs de pronostic. Matériel et méthode : Entre 1982 et 2002, 52 patients présentant un lymphome épidural primaire ont été traités dans neuf institutions membres du Rare Cancer Network. Les critères d'inclusion comprenaient : une biopsie confirmant le lymphome non-hodgkinien, un stade IE et IIE selon la classification de Ann Arbor, un traitement à visée curative de radiothérapie combinée ou non à une chimiothérapie et un suivi d'au moins six mois. Selon la Working Formulation, 12 patients (23%) présentaient un lymphome de bas grade, 28 (54%) un grade intermédiaire et 12 (23%) un haut grade. Les hommes étaient atteints 1.9 fois plus fréquemment que les femmes. L'âge moyen était de 61 ans (intervalle : 21 à 96). Le bilan incluait un Ct-scan spinal (98%), une IRM (52%), un CT-scan thoraco-abdominal (77%) et une aspiration ou biopsie de moelle osseuse (96%). Les symptômes les plus fréquents comprenaient des douleurs dorsales (79% des patients), une faiblesse musculaire (92%) et des déficits sensoriels (71 %). Quarante-huit patients ont subi une laminectomie de décompression avec résection partielle ou complète (42% et 13% des cas respectivement), tous ont reçu une radiothérapie seule (20 patients) ou en combinaison avec une chimiothérapie (32 patients). La dose médiane totale était de 36 Gy (intervalle 6-50 Gy) avec une moyenne de 20 Gy par fraction (intervalle : 1-25). Le suivi moyen était de 71 mois (intervalle : 22-165 mois). Résultats : Suite au traitement, une progression locale a été observée chez 6 patients après un temps de latence moyen de 6 mois. Le taux de rechute systémique a été de 42% (22 patients) le plus souvent dans les ganglions lymphatiques (n=9) après un intervalle de temps moyen de 20 mois. Lors du dernier contrôle, 28 patients étaient vivants et 24 patients étaient décédés. Le taux de survie à 5 ans, le taux de survie sans maladie et le contrôle local étaient de 69%, 57% et 88% respectivement. En analyse univariée, les facteurs pronostics favorables statistiquement significatifs concernant la survie sans maladie étaient un âge inférieur à 63 ans, ainsi qu'une réponse neurologique complète. Pour la survie à 5 ans, les facteurs favorables étaient un âge inférieur à 63 ans. En analyse multivariée, les facteurs pronostics favorables pour la survie globale à 5 ans étaient une réponse neurologique complète, un traitement combiné, un volume de radiothérapie plus que focal, une dose totale de radiothérapie supérieure à 36 Gy et une résection partielle ou complète de la tumeur. En ce qui concerne la survie sans maladie, les facteurs pronostics favorables étáient un âge inférieur à 63 ans et un traitement combiné. Conclusion : Ce qui ressort de cette analyse est que le bilan diagnostic devrait inclure une IRM ou un CT-scan, un échantillon de tissu pour poser le diagnostic pathologique définitif de la lésion, une histoire médicale et un examen physique complet, une chimie sanguine, un CTscan thoraco-abdominal et une biopsie de la moelle osseuse, un PET-scan devrait également faire partie du bilan. Le traitement devrait consister, dans la phase aiguë, en une chirurgie de décompression avec ou sans résection, suivie d'une radiothérapie d'au moins 36Gy en 2 Gy par fraction et d'une chimiothérapie. Tous les patients présentant un lymphome de haut grade ou de grade intermédiaire devraient pouvoir bénéficier d'un traitement combiné.

Evaluation of organ-specific peripheral doses after 2-dimensional, 3-dimensional and hybrid intensity modulated radiation therapy for breast cancer based on Monte Carlo and convolution/superposition algorithms: implications for secondary cancer risk assessment.

To make a comprehensive evaluation of organ-specific out-of-field doses using Monte Carlo (MC) simulations for different breast cancer irradiation techniques and to compare results with a commercial treatment planning system (TPS). Three breast radiotherapy techniques using 6MV tangential photon beams were compared: (a) 2DRT (open rectangular fields), (b) 3DCRT (conformal wedged fields), and (c) hybrid IMRT (open conformal+modulated fields). Over 35 organs were contoured in a whole-body CT scan and organ-specific dose distributions were determined with MC and the TPS. Large differences in out-of-field doses were observed between MC and TPS calculations, even for organs close to the target volume such as the heart, the lungs and the contralateral breast (up to 70% difference). MC simulations showed that a large fraction of the out-of-field dose comes from the out-of-field head scatter fluence (>40%) which is not adequately modeled by the TPS. Based on MC simulations, the 3DCRT technique using external wedges yielded significantly higher doses (up to a factor 4-5 in the pelvis) than the 2DRT and the hybrid IMRT techniques which yielded similar out-of-field doses. In sharp contrast to popular belief, the IMRT technique investigated here does not increase the out-of-field dose compared to conventional techniques and may offer the most optimal plan. The 3DCRT technique with external wedges yields the largest out-of-field doses. For accurate out-of-field dose assessment, a commercial TPS should not be used, even for organs near the target volume (contralateral breast, lungs, heart).

Quantifying network properties in multi-electrode recordings: spatiotemporal characterization and inter-trial variation of evoked gamma oscillations in mouse somatosensory cortex in vitro.

Linking the structural connectivity of brain circuits to their cooperative dynamics and emergent functions is a central aim of neuroscience research. Graph theory has recently been applied to study the structure-function relationship of networks, where dynamical similarity of different nodes has been turned into a "static" functional connection. However, the capability of the brain to adapt, learn and process external stimuli requires a constant dynamical functional rewiring between circuitries and cell assemblies. Hence, we must capture the changes of network functional connectivity over time. Multi-electrode array data present a unique challenge within this framework. We study the dynamics of gamma oscillations in acute slices of the somatosensory cortex from juvenile mice recorded by planar multi-electrode arrays. Bursts of gamma oscillatory activity lasting a few hundred milliseconds could be initiated only by brief trains of electrical stimulations applied at the deepest cortical layers and simultaneously delivered at multiple locations. Local field potentials were used to study the spatio-temporal properties and the instantaneous synchronization profile of the gamma oscillatory activity, combined with current source density (CSD) analysis. Pair-wise differences in the oscillation phase were used to determine the presence of instantaneous synchronization between the different sites of the circuitry during the oscillatory period. Despite variation in the duration of the oscillatory response over successive trials, they showed a constant average power, suggesting that the rate of expenditure of energy during the gamma bursts is consistent across repeated stimulations. Within each gamma burst, the functional connectivity map reflected the columnar organization of the neocortex. Over successive trials, an apparently random rearrangement of the functional connectivity was observed, with a more stable columnar than horizontal organization. This work reveals new features of evoked gamma oscillations in developing cortex.

Pronostic factors in solitary plasmacytoma of the bone: A Rare Cancer Network study

Contexte Le plasmocytome isolé osseux est une tumeur maligne rare des cellules plasmocytaires. Les données issues de la littérature ne permettent pas de se déterminer sur la dose radiothérapeutique optimale. Dans cette perspective nous avons conduit une vaste étude rétrospective dans le but d'évaluer l'évolution, les facteurs pronostiques aunsi que la dose radiothérapeutique optimale chez les patients présentant un plasmocytome isolé. Méthodes Nous avons réunis les données de 206 patients présentant un plasmocytome isolé osseux sans évidence de myélome multiple. Chaque cas a été documenté histopathologiquement. La majorité des patients (n=169) ont été traités par radiothérapie seule, 32 par une combinaison radiothérapie-chimiothérapie, et 5 par chirurgie. La durée de suivi médiane fut de 54 mois (7-245) Résultats A 5 ans, la survie globale est de 70%, la survie sans maladie de 46% et le contrôle local de 88%. La durée médiane de développement vers une myélome multiple est de 21 mois (2-135) avec une probabilité à 5 ans de 51 %. Les analyses multivariées indiquent l'âge (<60 ans) et la taille de la tumeur (<5cm) comme facteur favorables pour survie. L'âge (<60ans) se dégage comme facteur favorable pour la survie sans maladie. La localisation de la tumeur (vertébrale vs autre) indique la probabilité de contrôle local. L'âge plus avancé (>60 ans) est le seul prédicteur de myélome multiple. Aucune relation dose-réponse n'est mise en évidence pour les doses supérieures à 30 Gy, même pour lés tumeurs les plus étendues. Conclusions Les patients les plus jeunes, principalement ceux présentant une localisation vertébrale, présentent la meilleure évolution sous traitement radiothérapeutique modéré. La progression vers le myélome multiple reste le problème thérapeutique principal. Les futures investigations devraient se focaliser sur les chimiothérapies adjuvantes ainsi que sur les nouveaux agents thérapeutiques.

Steady-state expression of self-regulated genes.

MOTIVATION: Regulatory gene networks contain generic modules such as feedback loops that are essential for the regulation of many biological functions. The study of the stochastic mechanisms of gene regulation is instrumental for the understanding of how cells maintain their expression at levels commensurate with their biological role, as well as to engineer gene expression switches of appropriate behavior. The lack of precise knowledge on the steady-state distribution of gene expression requires the use of Gillespie algorithms and Monte-Carlo approximations. METHODOLOGY: In this study, we provide new exact formulas and efficient numerical algorithms for computing/modeling the steady-state of a class of self-regulated genes, and we use it to model/compute the stochastic expression of a gene of interest in an engineered network introduced in mammalian cells. The behavior of the genetic network is then analyzed experimentally in living cells. RESULTS: Stochastic models often reveal counter-intuitive experimental behaviors, and we find that this genetic architecture displays a unimodal behavior in mammalian cells, which was unexpected given its known bimodal response in unicellular organisms. We provide a molecular rationale for this behavior, and we implement it in the mathematical picture to explain the experimental results obtained from this network.

The Politics of Network Governance in Europe: The Case of Energy Regulation

Networks are considered increasingly important for policy-making. The literature on new modes of governance in Europe suggests that their horizontal coordination capacity and flexible and informal structures are particularly suitable for governing the multilevel architecture of the European polity. However, empirical evidence about the effects of networks on policy-making and public policies is still quite limited. This article uses the case of the European network of energy regulators to explore the determinants of the position of network members and, in turn, the domestic adoption of soft rules developed within this network. The empirical analysis, based on multivariate statistics and semi-directive interviews, supports the expectation that institutional complementarities increase actors' centrality in networks, while arguments based on organisational resources and age are disproved. Furthermore, results show that the overall level of adoption is considerable and that centrality might have a small positive effect on domestic adoption.

Unraveling the complex network of cuticular structure and function.

A hydrophobic cuticle is deposited at the outermost extracellular matrix of the epidermis in primary tissues of terrestrial plants. Besides forming a protective shield against the environment, the cuticle is potentially involved in several developmental processes during plant growth. A high degree of variation in cuticle composition and structure exists between different plant species and tissues. Lots of progress has been made recently in understanding the different steps of biosynthesis, transport, and deposition of cuticular components. However, the molecular mechanisms that underlie cuticular function remain largely elusive.

GLUT2 surface expression and intracellular transport via the constitutive pathway in pancreatic beta cells and insulinoma: evidence for a block in trans-Golgi network exit by brefeldin A.

The biosynthesis, intracellular transport, and surface expression of the beta cell glucose transporter GLUT2 was investigated in isolated islets and insulinoma cells. Using a trypsin sensitivity assay to measure cell surface expression, we determined that: (a) greater than 95% of GLUT2 was expressed on the plasma membrane; (b) GLUT2 did not recycle in intracellular vesicles; and (c) after trypsin treatment, reexpression of the intact transporter occurred with a t1/2 of approximately 7 h. Kinetics of intracellular transport of GLUT2 was investigated in pulse-labeling experiments combined with glycosidase treatment and the trypsin sensitivity assay. We determined that transport from the endoplasmic reticulum to the trans-Golgi network (TGN) occurred with a t1/2 of 15 min and that transport from the TGN to the plasma membrane required a similar half-time. When added at the start of a pulse-labeling experiment, brefeldin A prevented exit of GLUT2 from the endoplasmic reticulum. When the transporter was first accumulated in the TGN during a 15-min period of chase, but not following a low temperature (22 degrees C) incubation, addition of brefeldin A (BFA) prevented subsequent surface expression of the transporter. This indicated that brefeldin A prevented GLUT2 exit from the TGN by acting at a site proximal to the 22 degrees C block. Together, these data demonstrate that GLUT2 surface expression in beta cells is via the constitutive pathway, that transport can be blocked by BFA at two distinct steps and that once on the surface, GLUT2 does not recycle in intracellular vesicles.

Induction of an immune response through the idiotypic network with monoclonal anti-idiotype antibodies in the carcinoembryonic antigen system.

Anti-idiotype antibodies can mimic the conformational epitopes of the original antigen and act as antigen substitutes for vaccination and/or serological purposes. To investigate this possibility concerning the tumor marker carcinoembryonic antigen (CEA), BALB/c mice were immunized with the previously described anti-CEA monoclonal antibody (MAb) 5.D11 (AB1). After cell fusion, 15 stable cloned cell lines secreting anti-Ids (AB2) were obtained. Selected MAbs gave various degrees of inhibition (up to 100%) of the binding of 125I-labeled CEA to MAb 5.D11. Absence of reactivity of anti-Id MAbs with normal mouse IgG was first demonstrated by the fact that anti-Id MAbs were not absorbed by passage through a mouse IgG column, and second because they bound specifically to non-reduced MAb 5.D11 on Western blots. Anti-5.D11 MAbs did not inhibit binding to CEA of MAb 10.B9, another anti-CEA antibody obtained in the same fusion as 5.D11, or that of several anti-CEA MAbs reported in an international workshop, with the exception of two other anti-CEA MAbs, both directed against the GOLD IV epitope. When applied to an Id-anti-Id competitive radioimmunoassay, a sensitivity of 2 ng/ml of CEA was obtained, which is sufficient for monitoring circulating CEA in carcinoma patients. To verify that the anti-Id MAbs have the potential to be used as CEA vaccines, syngeneic BALB/c mice were immunized with these MAbs (AB2). Sera from immunized mice were demonstrated to contain AB3 antibodies recognizing the original antigen, CEA, both in enzyme immunoassay and by immunoperoxidase staining of human colon carcinoma. These results open the perspective of vaccination against colorectal carcinoma through the use of anti-idiotype antibodies as antigen substitutes.