Statistical learning theory for geospatial data. Case study: Aral sea

In recent years there has been an explosive growth in the development of adaptive and data driven methods. One of the efficient and data-driven approaches is based on statistical learning theory (Vapnik 1998). The theory is based on Structural Risk Minimisation (SRM) principle and has a solid statistical background. When applying SRM we are trying not only to reduce training error ? to fit the available data with a model, but also to reduce the complexity of the model and to reduce generalisation error. Many nonlinear learning procedures recently developed in neural networks and statistics can be understood and interpreted in terms of the structural risk minimisation inductive principle. A recent methodology based on SRM is called Support Vector Machines (SVM). At present SLT is still under intensive development and SVM find new areas of application (www.kernel-machines.org). SVM develop robust and non linear data models with excellent generalisation abilities that is very important both for monitoring and forecasting. SVM are extremely good when input space is high dimensional and training data set i not big enough to develop corresponding nonlinear model. Moreover, SVM use only support vectors to derive decision boundaries. It opens a way to sampling optimization, estimation of noise in data, quantification of data redundancy etc. Presentation of SVM for spatially distributed data is given in (Kanevski and Maignan 2004).

Active learning methods for remote sensing image classification

In this paper, we propose two active learning algorithms for semiautomatic definition of training samples in remote sensing image classification. Based on predefined heuristics, the classifier ranks the unlabeled pixels and automatically chooses those that are considered the most valuable for its improvement. Once the pixels have been selected, the analyst labels them manually and the process is iterated. Starting with a small and nonoptimal training set, the model itself builds the optimal set of samples which minimizes the classification error. We have applied the proposed algorithms to a variety of remote sensing data, including very high resolution and hyperspectral images, using support vector machines. Experimental results confirm the consistency of the methods. The required number of training samples can be reduced to 10% using the methods proposed, reaching the same level of accuracy as larger data sets. A comparison with a state-of-the-art active learning method, margin sampling, is provided, highlighting advantages of the methods proposed. The effect of spatial resolution and separability of the classes on the quality of the selection of pixels is also discussed.

Batch effect confounding leads to strong bias in performance estimates obtained by cross-validation.

BACKGROUND: With the large amount of biological data that is currently publicly available, many investigators combine multiple data sets to increase the sample size and potentially also the power of their analyses. However, technical differences ("batch effects") as well as differences in sample composition between the data sets may significantly affect the ability to draw generalizable conclusions from such studies. FOCUS: The current study focuses on the construction of classifiers, and the use of cross-validation to estimate their performance. In particular, we investigate the impact of batch effects and differences in sample composition between batches on the accuracy of the classification performance estimate obtained via cross-validation. The focus on estimation bias is a main difference compared to previous studies, which have mostly focused on the predictive performance and how it relates to the presence of batch effects. DATA: We work on simulated data sets. To have realistic intensity distributions, we use real gene expression data as the basis for our simulation. Random samples from this expression matrix are selected and assigned to group 1 (e.g., 'control') or group 2 (e.g., 'treated'). We introduce batch effects and select some features to be differentially expressed between the two groups. We consider several scenarios for our study, most importantly different levels of confounding between groups and batch effects. METHODS: We focus on well-known classifiers: logistic regression, Support Vector Machines (SVM), k-nearest neighbors (kNN) and Random Forests (RF). Feature selection is performed with the Wilcoxon test or the lasso. Parameter tuning and feature selection, as well as the estimation of the prediction performance of each classifier, is performed within a nested cross-validation scheme. The estimated classification performance is then compared to what is obtained when applying the classifier to independent data.

'Toxic' and 'Nontoxic': confirming critical terminology concepts and context for clear communication

If 'the dose makes the poison', and if the context of an exposure to a hazard shapes the risk as much as the innate character of the hazard itself, then what is 'toxic' and what is 'nontoxic'? This article is intended to help readers and communicators: anticipate that concepts such as 'toxic' and 'nontoxic' may have different meanings to different stakeholders in different contexts of general use, commerce, science, and the law; recognize specific situations in which terms and related information could potentially be misperceived or misinterpreted; evaluate the relevance, reliability, and other attributes of information for a given situation; control actions, assumptions, interpretations, conclusions, and decisions to avoid flaws and achieve a desired outcome; and confirm that the desired outcome has been achieved. To meet those objectives, we provide some examples of differing toxicology terminology concepts and contexts; a comprehensive decision-making framework for understanding and managing risk; along with a communication and education message and audience-planning matrix to support the involvement of all relevant stakeholders; a set of CLEAR-communication assessment criteria for use by both readers and communicators; example flaws in decision-making; a suite of three tools to assign relevance vs reliability, align know vs show, and refine perception vs reality aspects of information; and four steps to foster effective community involvement and support. The framework and supporting process are generally applicable to meeting any objective.

Discordant increases in CD4+ T cells in human immunodeficiency virus-infected patients experiencing virologic treatment failure: role of changes in thymic output and T cell death.

Some patients infected with human immunodeficiency virus (HIV) who are experiencing antiretroviral treatment failure have persistent improvement in CD4+ T cell counts despite high plasma viremia. To explore the mechanisms responsible for this phenomenon, 2 parameters influencing the dynamics of CD4+ T cells were evaluated: death of mature CD4+ T cells and replenishment of the CD4+ T cell pool by the thymus. The improvement in CD4+ T cells observed in patients with treatment failure was not correlated with spontaneous, Fas ligand-induced, or activation-induced T cell death. In contrast, a significant correlation between the improvement in CD4+ T cell counts and thymic output, as assessed by measurement of T cell receptor excision circles, was observed. These observations suggest that increased thymic output contributes to the dissociation between CD4+ T cell counts and viremia in patients failing antiretroviral therapy and support a model in which drug-resistant HIV strains may have reduced replication rates and pathogenicity in the thymus.

Drug adherence in chronic kidney diseases and dialysis.

Poor long-term adherence and persistence to drug therapy is universally recognized as one of the major clinical issues in the management of chronic diseases, and patients with renal diseases are also concerned by this important phenomenon. Chronic kidney disease (CKD) patients belong to the group of subjects with one of the highest burdens of daily pill intake with up to >20 pills per day depending on the severity of their disease. The purpose of the present review is to discuss the difficulties encountered by nephrologists in diagnosing and managing poor adherence and persistence in CKD patients including in patients receiving maintenance dialysis. Our review will also attempt to provide some clues and new perspectives on how drug adherence could actually be addressed and possibly improved. Working on drug adherence may look like a long and tedious path, but physicians and healthcare providers should always be aware that drug adherence is in general much lower than what they may think and that there are many ways to improve and support drug adherence and persistence so that renal patients obtain the full benefits of their treatments.

Non-invasive detection of coronary endothelial response to sequential handgrip exercise in coronary artery disease patients and healthy adults.

OBJECTIVES: Our objective is to test the hypothesis that coronary endothelial function (CorEndoFx) does not change with repeated isometric handgrip (IHG) stress in CAD patients or healthy subjects. BACKGROUND: Coronary responses to endothelial-dependent stressors are important measures of vascular risk that can change in response to environmental stimuli or pharmacologic interventions. The evaluation of the effect of an acute intervention on endothelial response is only valid if the measurement does not change significantly in the short term under normal conditions. Using 3.0 Tesla (T) MRI, we non-invasively compared two coronary artery endothelial function measurements separated by a ten minute interval in healthy subjects and patients with coronary artery disease (CAD). METHODS: Twenty healthy adult subjects and 12 CAD patients were studied on a commercial 3.0 T whole-body MR imaging system. Coronary cross-sectional area (CSA), peak diastolic coronary flow velocity (PDFV) and blood-flow were quantified before and during continuous IHG stress, an endothelial-dependent stressor. The IHG exercise with imaging was repeated after a 10 minute recovery period. RESULTS: In healthy adults, coronary artery CSA changes and blood-flow increases did not differ between the first and second stresses (mean % change ±SEM, first vs. second stress CSA: 14.8%±3.3% vs. 17.8%±3.6%, p = 0.24; PDFV: 27.5%±4.9% vs. 24.2%±4.5%, p = 0.54; blood-flow: 44.3%±8.3 vs. 44.8%±8.1, p = 0.84). The coronary vasoreactive responses in the CAD patients also did not differ between the first and second stresses (mean % change ±SEM, first stress vs. second stress: CSA: -6.4%±2.0% vs. -5.0%±2.4%, p = 0.22; PDFV: -4.0%±4.6% vs. -4.2%±5.3%, p = 0.83; blood-flow: -9.7%±5.1% vs. -8.7%±6.3%, p = 0.38). CONCLUSION: MRI measures of CorEndoFx are unchanged during repeated isometric handgrip exercise tests in CAD patients and healthy adults. These findings demonstrate the repeatability of noninvasive 3T MRI assessment of CorEndoFx and support its use in future studies designed to determine the effects of acute interventions on coronary vasoreactivity.

Mult-class differentiation of cannabis seedlings in a forensic context

This article presents an experimental study about the classification ability of several classifiers for multi-classclassification of cannabis seedlings. As the cultivation of drug type cannabis is forbidden in Switzerland lawenforcement authorities regularly ask forensic laboratories to determinate the chemotype of a seized cannabisplant and then to conclude if the plantation is legal or not. This classification is mainly performed when theplant is mature as required by the EU official protocol and then the classification of cannabis seedlings is a timeconsuming and costly procedure. A previous study made by the authors has investigated this problematic [1]and showed that it is possible to differentiate between drug type (illegal) and fibre type (legal) cannabis at anearly stage of growth using gas chromatography interfaced with mass spectrometry (GC-MS) based on therelative proportions of eight major leaf compounds. The aims of the present work are on one hand to continueformer work and to optimize the methodology for the discrimination of drug- and fibre type cannabisdeveloped in the previous study and on the other hand to investigate the possibility to predict illegal cannabisvarieties. Seven classifiers for differentiating between cannabis seedlings are evaluated in this paper, namelyLinear Discriminant Analysis (LDA), Partial Least Squares Discriminant Analysis (PLS-DA), Nearest NeighbourClassification (NNC), Learning Vector Quantization (LVQ), Radial Basis Function Support Vector Machines(RBF SVMs), Random Forest (RF) and Artificial Neural Networks (ANN). The performance of each method wasassessed using the same analytical dataset that consists of 861 samples split into drug- and fibre type cannabiswith drug type cannabis being made up of 12 varieties (i.e. 12 classes). The results show that linear classifiersare not able to manage the distribution of classes in which some overlap areas exist for both classificationproblems. Unlike linear classifiers, NNC and RBF SVMs best differentiate cannabis samples both for 2-class and12-class classifications with average classification results up to 99% and 98%, respectively. Furthermore, RBFSVMs correctly classified into drug type cannabis the independent validation set, which consists of cannabisplants coming from police seizures. In forensic case work this study shows that the discrimination betweencannabis samples at an early stage of growth is possible with fairly high classification performance fordiscriminating between cannabis chemotypes or between drug type cannabis varieties.

Major environmental change and bonebed genesis prior to the Triassic-Jurassic mass extinction

We present new geochemical and sedimentological data from marginal marine strata of Penarth Bay, south Wales (UK) to elucidate the origin of widespread but enigmatic concentrations of vertebrate hard parts (bonebeds) in marine successions of Rhaetian age (late Triassic). Sedimentological evidence shows that the phosphatic constituents of the bonebeds were subjected to intense phosphatization in shallow current-dominated settings and subsequently reworked and transported basinward by storms. Interbedded organic-rich strata deposited under quiescent and poorly oxygenated conditions record enhanced phosphorus regeneration from sedimentary organic matter into the water column and probably provided the main source of phosphate required for heavy bonebed clast phosphatization. The stratigraphically limited interval showing evidence for oxygen depletion and accelerated P-cycling coincides with a negative 4% organic carbon isotope excursion, which possibly reflects supra-regional changes in carbon cycling and clearly predates the 'initial isotope excursion' characterizing many Triassic-Jurassic boundary strata. our data indicate that Rhaetian bonebeds are the lithological signature of profound, climatically driven changes in carbon cycling and redox conditions and support the idea of a multi-pulsed environmental crisis at the end of the Triassic, possibly linked to successive episodes of igneous activity in the central Atlantic Magmatic Province.

Oral and intravenous methadone use: some clinical and pharmacokinetic aspects.

A sample of 15 patients participating in an injectable methadone trial and of 15 patients in an oral methadone maintenance treatment, who admitted injecting part or all of their methadone take-home doses, were compared to 20 patients in maintenance treatment who use methadone exclusively by mouth. The present study confirms the poorer general health, the higher levels of emotional, psychological or psychiatric problems, the higher use of illicit drugs, and the higher number of problems related to employment and support associated with the use of the intravenous mode of administration of methadone. As expected, due to the shunt of metabolism in the gut wall and of the liver first-pass effect, higher concentration to dose ratios of (R)-methadone, which is the active enantiomer, were measured in the intravenous group (23% increase). This difference reached an almost statistically significant value (P = 0.054). This raises the question whether the effect of a higher methadone dose could be unconsciously sought by some of the intravenous methadone users.

Qualitative analysis of barriers and facilitators encountered by HIV patients in an ART adherence programme.

Background Medication adherence is a complex, dynamic and changing behaviour that is affected by a variety of factors, including the patient's beliefs and life circumstances. Studies have highlighted barriers to medication adherence (e.g., unmanaged side effects or a lack of social support), as well as facilitators of medication adherence (e.g., technical simplicity of treatment and psychological acceptance of the disease). Since August 2004, in Lausanne (Switzerland), physicians have referred patients who are either experiencing or are at risk of experiencing problems with their HIV antiretroviral treatment (ART) to a routine interdisciplinary ART adherence programme. This programme consists of multifactorial intervention including electronic drug monitoring (MEMS(TM)). Objective This study's objective was to identify the barriers and facilitators encountered by HIV patients with suboptimal medication adherence (≤90 % adherence over the study period). Setting The community pharmacy of the Department of Ambulatory Care and Community Medicine in Lausanne (Switzerland). Method The study consisted of a retrospective, qualitative, thematic content analysis of pharmacists' notes that were taken during semi-structured interviews with patients and conducted as part of the ART adherence programme between August 2004 and May 2008. Main outcome measure Barriers and facilitators encountered by HIV patients. Results Barriers to and facilitators of adherence were identified for the 17 included patients. These factors fell into three main categories: (1) cognitive, emotional and motivational; (2) environmental, organisational and social; and (3) treatment and disease. Conclusion The pharmacists' notes revealed that diverse barriers and facilitators were discussed during medication adherence interviews. Indeed, the results showed that the 17 non-adherent patients encountered barriers and benefited from facilitators. Therefore, pharmacists should inquire about all factors, regardless of whether they have a negative or a positive impact on medication adherence, and should consider all dimensions of patient adherence. The simultaneous strengthening of facilitators and better management of barriers may allow healthcare providers to tailor care to a patient's specific needs and support each individual patient in improving his medication-related behaviour.

Analysis of an iterative deconvolution approach for estimating the source wavelet during waveform inversion of crosshole ground-penetrating radar data

A major issue in the application of waveform inversion methods to crosshole georadar data is the accurate estimation of the source wavelet. Here, we explore the viability and robustness of incorporating this step into a time-domain waveform inversion procedure through an iterative deconvolution approach. Our results indicate that, at least in non-dispersive electrical environments, such an approach provides remarkably accurate and robust estimates of the source wavelet even in the presence of strong heterogeneity in both the dielectric permittivity and electrical conductivity. Our results also indicate that the proposed source wavelet estimation approach is relatively insensitive to ambient noise and to the phase characteristics of the starting wavelet. Finally, there appears to be little-to-no trade-off between the wavelet estimation and the tomographic imaging procedures.

Collaboration entre médecin de famille et psychiatre pour les problèmes de santé mentale [Family physicians and psychiatrists' collaborative care for mental health problems].

The burden of disease linked to mental disorders represents more than one-fifth of years lived with disability in the world. Less than half of people suffering from mental disorders are adequately treated. Three quarter of those who receive treatment are followed by primary care. Collaborative care aims to increase the efficiency of direct general practitioner's treatment. Main components are sustainable and individualized consultation-liaison relationship (1/2 day of psychiatrist by 15 days for 10-15 general practitioners), and support of a clinical case manager for complex situations. Collaboration is bidirectional: early or crisis access to specialist care and long-term followup by general practitioner. This model is a challenge for the doctor-patient dual relationship and requires incentives in a public health perspective.

Cancer du sein et obésité, une liaison dangereuse [Breast cancer and obesity, a dangerous relation].

Obesity is associated with different cancers including breast cancer, whose incidence is increased in postmenopausal women. It has an adverse impact on the prognosis of the patients, regardless of their menopausal status. The fact of receiving a systemic adjuvant therapy does not neutralize the prognostic role of obesity. Moderate weight loss after cancer diagnosis could improve the outcome of the patients, while a weight gain during treatment seems without significant effect. Currently available data are still too incomplete to justify systematic programs to lose weight with an oncologic therapeutic aim. However, it is worth to encourage and support our patients to have an optimal diet, physical activity, and to lose weight as promotion of general health.

Learning wind fields with multiple kernels

This paper presents multiple kernel learning (MKL) regression as an exploratory spatial data analysis and modelling tool. The MKL approach is introduced as an extension of support vector regression, where MKL uses dedicated kernels to divide a given task into sub-problems and to treat them separately in an effective way. It provides better interpretability to non-linear robust kernel regression at the cost of a more complex numerical optimization. In particular, we investigate the use of MKL as a tool that allows us to avoid using ad-hoc topographic indices as covariables in statistical models in complex terrains. Instead, MKL learns these relationships from the data in a non-parametric fashion. A study on data simulated from real terrain features confirms the ability of MKL to enhance the interpretability of data-driven models and to aid feature selection without degrading predictive performances. Here we examine the stability of the MKL algorithm with respect to the number of training data samples and to the presence of noise. The results of a real case study are also presented, where MKL is able to exploit a large set of terrain features computed at multiple spatial scales, when predicting mean wind speed in an Alpine region.