Data-driven clustering : new methods and applications

Abstract : This work is concerned with the development and application of novel unsupervised learning methods, having in mind two target applications: the analysis of forensic case data and the classification of remote sensing images. First, a method based on a symbolic optimization of the inter-sample distance measure is proposed to improve the flexibility of spectral clustering algorithms, and applied to the problem of forensic case data. This distance is optimized using a loss function related to the preservation of neighborhood structure between the input space and the space of principal components, and solutions are found using genetic programming. Results are compared to a variety of state-of--the-art clustering algorithms. Subsequently, a new large-scale clustering method based on a joint optimization of feature extraction and classification is proposed and applied to various databases, including two hyperspectral remote sensing images. The algorithm makes uses of a functional model (e.g., a neural network) for clustering which is trained by stochastic gradient descent. Results indicate that such a technique can easily scale to huge databases, can avoid the so-called out-of-sample problem, and can compete with or even outperform existing clustering algorithms on both artificial data and real remote sensing images. This is verified on small databases as well as very large problems. Résumé : Ce travail de recherche porte sur le développement et l'application de méthodes d'apprentissage dites non supervisées. Les applications visées par ces méthodes sont l'analyse de données forensiques et la classification d'images hyperspectrales en télédétection. Dans un premier temps, une méthodologie de classification non supervisée fondée sur l'optimisation symbolique d'une mesure de distance inter-échantillons est proposée. Cette mesure est obtenue en optimisant une fonction de coût reliée à la préservation de la structure de voisinage d'un point entre l'espace des variables initiales et l'espace des composantes principales. Cette méthode est appliquée à l'analyse de données forensiques et comparée à un éventail de méthodes déjà existantes. En second lieu, une méthode fondée sur une optimisation conjointe des tâches de sélection de variables et de classification est implémentée dans un réseau de neurones et appliquée à diverses bases de données, dont deux images hyperspectrales. Le réseau de neurones est entraîné à l'aide d'un algorithme de gradient stochastique, ce qui rend cette technique applicable à des images de très haute résolution. Les résultats de l'application de cette dernière montrent que l'utilisation d'une telle technique permet de classifier de très grandes bases de données sans difficulté et donne des résultats avantageusement comparables aux méthodes existantes.

Multiplex PCR of sonication fluid accurately differentiates between prosthetic joint infection and aseptic failure.

OBJECTIVE: Cultures have limited sensitivity in the diagnosis of prosthetic joint infection (PJI), especially in low-grade infections. We assessed the value of multiplex PCR in differentiating PJI from aseptic failure (AF). METHODS: Included were patients in whom the joint prosthesis was removed and submitted for sonication. The resulting sonication fluid was cultured and investigated by multiplex PCR, and compared with periprosthetic tissue culture. RESULTS: Among 86 explanted prostheses (56 knee, 25 hip, 3 elbow and 2 shoulder prostheses), AF was diagnosed in 62 cases (72%) and PJI in 24 cases (28%). PJI was more common detected by multiplex PCR (n=23, 96%) than by periprosthetic tissue (n=17, 71%, p=0.031) or sonication fluid culture (n=16, 67%, p=0.016). Among 12 patients with PJI who previously received antibiotics, periprosthetic tissue cultures were positive in 8 cases (67%), sonication fluid cultures in 6 cases (50%) and multiplex PCR in 11 cases (92%). In AF cases, periprosthetic tissue grew organisms in 11% and sonication fluid in 10%, whereas multiplex PCR detected no organisms. CONCLUSIONS: Multiplex PCR of sonication fluid demonstrated high sensitivity (96%) and specificity (100%) for diagnosing PJI, providing good discriminative power towards AF, especially in patients previously receiving antibiotics.

Exposition des enfants à la fumée passive: le point sur des campagnes de sensibilisation

Enfants de moins de 10 ans fumant passivement 14 cigarettes ! D'avril 2010 à avril 2011, l'exposition de 148 enfants (81 garçons et 67 filles) a été testée: 10 enfants de moins d'un an, 25 de 1 à 5 ans, 19 de 5 à 10 ans, 30 de 10 à 15 ans et 64 de 15 à 18 ans. 10 d'entre eux sont des fumeurs et la plus jeune de 14 ans fume 10 cigarettes par jour. Leurs parents, ou parfois des jeunes eux-mêmes, ont commandé de manière volontaire, via les sites Internet des CIPRET Valais, Vaud et Genève, un badge MoNIC gratuit. Les résultats quant à l'exposition de ces enfants interpellent et méritent l'attention.Pour l'ensemble des enfants, la concentration moyenne de nicotine dans leur environnement intérieur mesurée via les dispositifs MoNIC est de 0,5 mg/m3, avec des maximums pouvant aller jusqu'à 21 mg/m3. Pour le collectif d'enfants âgés de moins de 10 ans (26 garçons et 28 filles; tous non-fumeurs), la concentration de nicotine n'est pas négligeable (moyenne 0,069 mg/m3, min 0, max 0,583 mg/m3). En convertissant ce résultat en équivalent de cigarettes inhalées passivement, nous obtenons des chiffres allant de 0 à 14 cigarettes par jour* avec une moyenne se situant à 1.6 cig/j. Encore plus surprenant, les enfants de moins d'un an (4 garçons et 6 filles) inhalent passivement, dans le cadre familial, en moyenne 1 cigarette (min 0, max 2.2). Pour les deux autres collectifs: 10-15 ans et 15-18 ans, les valeurs maximales avoisinent les 22 cigarettes. Notons cependant que ce résultat est influencé, ce qui n'est pas le cas des enfants plus jeunes, par le fait que ces jeunes sont également parfois des fumeurs actifs.* Quand la durée d'exposition dépassait 1 jour (8 heures), le nombre d'heures a toujours été divisé par 8 heures. Le résultat obtenu donne l'équivalent de cigarettes fumées passivement en huit heures. Il s'agit de ce fait d'une moyenne, ce qui veut dire que durant cette période les enfants ont pu être exposés irrégulièrement à des valeurs supérieures ou inférieures à cette moyenne. [Auteurs]

Official Positions for FRAX® clinical regarding smoking from Joint Official Positions Development Conference of the International Society for Clinical Densitometry and International Osteoporosis Foundation on FRAX®.

The worldwide prevalence of smoking has been estimated at about 50% in men, and 10% in women, with larger variations among different populations studied. Smoking has been shown to affect many organ systems resulting in severe morbidity and increased mortality. In addition, smoking has been identified as a predictor of ten-year fracture risk in men and women, largely independent of an individual's bone mineral density. This finding has eventually lead to incorporation of this risk factor into FRAX®, an algorithm that has been developed to calculate an individual's ten-year fracture risk. However, only little, or conflicting data is available on a possible association between smoking dose, duration, length of time after cessation, type of tobacco and fracture risk, limiting this risk factor's applicability in the context of FRAX®.

Bootstrapping pairs in Distance-Based Regression

La regressió basada en distàncies és un mètode de predicció que consisteix en dos passos: a partir de les distàncies entre observacions obtenim les variables latents, les quals passen a ser els regressors en un model lineal de mínims quadrats ordinaris. Les distàncies les calculem a partir dels predictors originals fent us d'una funció de dissimilaritats adequada. Donat que, en general, els regressors estan relacionats de manera no lineal amb la resposta, la seva selecció amb el test F usual no és possible. En aquest treball proposem una solució a aquest problema de selecció de predictors definint tests estadístics generalitzats i adaptant un mètode de bootstrap no paramètric per a l'estimació dels p-valors. Incluim un exemple numèric amb dades de l'assegurança d'automòbils.

Corynebacterium bovis shoulder prosthetic joint infection: the first reported case.

We report the first case of Corynebacterium bovis shoulder prosthetic joint infection. The organism was isolated from intraoperative tissue culture and from the removed prosthesis using sonication. A 2-stage exchange and 3 months of antibiotic therapy were performed. C. bovis may cause implant-associated infections, which can manifest as low-grade infection.

Assessment of therapeutic benefit of antiviral therapy in chronic hepatitis C: is hepatic venous pressure gradient a better end point?

Chronic hepatitis C is a major healthcare problem. The response to antiviral therapy for patients with chronic hepatitis C has previously been defined biochemically and by PCR. However, changes in the hepatic venous pressure gradient (HVPG) may be considered as an adjunctive end point for the therapeutic evaluation of antiviral therapy in chronic hepatitis C. It is a validated technique which is safe, well tolerated, well established, and reproducible. Serial HVPG measurements may be the best way to evaluate response to therapy in chronic hepatitis C.

Sepsis liées aux accès vasculaires : mise au point, développements récents et prévention [Catheter-related infections : owerview and lipdate]

Les infections liées aux accès vasculaires sont l'une des causes principales des infections nosocomiales. Elles englobent leur colonisation par des micro-organismes, les infections du site d'insertion et les bactériémies et fongémies qui leur sont attribuées. Une bactériémie complique en moyenne 3 à 5 voies veineuses sur 100, ou représente de 2 à 14 épisodes pour 1000 jour-cathéters. Cette proportion n'est que la partie visible de l'iceberg puisque la plupart des épisodes de sepsis clinique sans foyer infectieux apparent associé sont actuellement considérés comme secondaires aux accès vasculaires. Les principes thérapeutiques sont présentés après une brève revue de leur physiopathologie. Plusieurs approches préventives sont ensuite discutées, y compris des éléments récents sur l'utilisation de cathéters imprégnés de désinfectants ou d'antibiotiques.

A single point mutation in the pore region of the epithelial Na+ channel changes ion selectivity by modifying molecular sieving.

The epithelial Na+ channel (ENaC) belongs to a new class of channel proteins called the ENaC/DEG superfamily involved in epithelial Na+ transport, mechanotransduction, and neurotransmission. The role of ENaC in Na+ homeostasis and in the control of blood pressure has been demonstrated recently by the identification of mutations in ENaC beta and gamma subunits causing hypertension. The function of ENaC in Na+ reabsorption depends critically on its ability to discriminate between Na+ and other ions like K+ or Ca2+. ENaC is virtually impermeant to K+ ions, and the molecular basis for its high ionic selectivity is largely unknown. We have identified a conserved Ser residue in the second transmembrane domain of the ENaC alpha subunit (alphaS589), which when mutated allows larger ions such as K+, Rb+, Cs+, and divalent cations to pass through the channel. The relative ion permeability of each of the alphaS589 mutants is related inversely to the ionic radius of the permeant ion, indicating that alphaS589 mutations increase the molecular cutoff of the channel by modifying the pore geometry at the selectivity filter. Proper geometry of the pore is required to tightly accommodate Na+ and Li+ ions and to exclude larger cations. We provide evidence that ENaC discriminates between cations mainly on the basis of their size and the energy of dehydration.

Formation and distribution of point defects on a disclination line near a free nematic interface

Point defects of opposite signs can alternately nucleate on the -1/2 disclination line that forms near the free surface of a confined nematic liquid crystal. We show the existence of metastable configurations consisting of periodic repetitions of such defects. These configurations are characterized by a minimal interdefect spacing that is seen to depend on sample thickness and on an applied electric field. The time evolution of the defect distribution suggests that the defects attract at small distances and repel at large distances.

Joint Enrollment Report, 2011

The Iowa Department of Education collects information on joint enrollment in Iowa’s 15 community colleges. Jointly enrolled students are high school students enrolled in community college credit coursework. Most jointly enrolled students enroll through Senior Year Plus programs such as PSEO (Postsecondary Enrollment Options) and concurrent enrollment. Others enroll independently by paying tuition or enrolling in courses delivered through contractual agreements that do not meet the definition of concurrent enrollment. For more information about Senior Year Plus programs, please refer to the department’s website. This report consists of fiscal year and trend data on joint enrollment including headcount enrollment, credit hours, student demographics, and enrollment by program type and offering arrangement. All data included in this report is taken from the Community College Management Information System (MIS) and confirmed by each college, unless otherwise noted.

Analysis of joint laxity after total ankle arthroplasty: cadaver study.

BACKGROUND: Clinical results of total ankle arthroplasty with early designs were disappointing. Recently-developed ankle prostheses have good mid-term results; however, limited information is available regarding effects of total ankle arthroplasty on ankle laxity. METHODS: Eight cadaveric lower extremities were tested with a custom device which enabled measurement of multi-axial forces, moments, and displacement during applied axial, shear, and rotational loading. Tests consisted of anterior-posterior and medial-lateral translation and internal-external rotation of the talus relative to the tibia during axial loads on the tibia simulating body weight (700 N) and an unloaded condition (5 N). Tests were performed in neutral, dorsiflexion, and plantarflexion. Laxity was determined for the intact ankle, and following insertion of an unconstrained total ankle implant, comparing load-displacement curve. FINDINGS: Laxity after total ankle arthroplasty did not approximate the normal ankle in most conditions tested. Displacement was significantly greater for total ankle arthroplasty in both posterior and lateral translation, and internal rotation, with 5 N axial loading, and anterior-posterior, medial-lateral translation, and internal-external rotation for 700 N axial loading. For the 700 N axial load condition, in the neutral ankle position, total anterior-posterior translation averaged 0.4 mm (SD 0.2 mm), but 6.0 mm (SD 1.5 mm) after total ankle arthroplasty (P<0.01). This study demonstrated more laxity in the replaced ankle than normal ankle for both unloaded and 700 N axially loaded conditions. INTERPRETATION: These data indicate the increased responsibility of the ligaments for ankle laxity after total ankle arthroplasty and suggest the importance of meticulous ligament reconstruction with total ankle arthroplasty operations.