Hierarchical conditional random fields for parts based models matching

A parts based model is a parametrization of an object class using a collection of landmarks following the object structure. The matching of parts based models is one of the problems where pairwise Conditional Random Fields have been successfully applied. The main reason of their effectiveness is tractable inference and learning due to the simplicity of involved graphs, usually trees. However, these models do not consider possible patterns of statistics among sets of landmarks, and thus they sufffer from using too myopic information. To overcome this limitation, we propoese a novel structure based on a hierarchical Conditional Random Fields, which we explain in the first part of this memory. We build a hierarchy of combinations of landmarks, where matching is performed taking into account the whole hierarchy. To preserve tractable inference we effectively sample the label set. We test our method on facial feature selection and human pose estimation on two challenging datasets: Buffy and MultiPIE. In the second part of this memory, we present a novel approach to multiple kernel combination that relies on stacked classification. This method can be used to evaluate the landmarks of the parts-based model approach. Our method is based on combining responses of a set of independent classifiers for each individual kernel. Unlike earlier approaches that linearly combine kernel responses, our approach uses them as inputs to another set of classifiers. We will show that we outperform state-of-the-art methods on most of the standard benchmark datasets.

Improving outcomes for people with skin tumours including melanoma (update)

The management of low-risk basal cell carcinomas in the community.In February 2006, the National Institute for Health and Clinical Excellence (NICE) published service guidance on skin cancer, ‘Improving outcomes for people with skin tumours including melanoma’ (NICE guidance on cancer services). Many of the recommendations in this guidance were converted into peer review measures published in the ‘Manual for cancer services 2008: skin measures..

Policy Diffusion: An Agent-Based Approach

Methods like Event History Analysis can show the existence of diffusion and part of its nature, but do not study the process itself. Nowadays, thanks to the increasing performance of computers, processes can be studied using computational modeling. This thesis presents an agent-based model of policy diffusion mainly inspired from the model developed by Braun and Gilardi (2006). I first start by developing a theoretical framework of policy diffusion that presents the main internal drivers of policy diffusion - such as the preference for the policy, the effectiveness of the policy, the institutional constraints, and the ideology - and its main mechanisms, namely learning, competition, emulation, and coercion. Therefore diffusion, expressed by these interdependencies, is a complex process that needs to be studied with computational agent-based modeling. In a second step, computational agent-based modeling is defined along with its most significant concepts: complexity and emergence. Using computational agent-based modeling implies the development of an algorithm and its programming. When this latter has been developed, we let the different agents interact. Consequently, a phenomenon of diffusion, derived from learning, emerges, meaning that the choice made by an agent is conditional to that made by its neighbors. As a result, learning follows an inverted S-curve, which leads to partial convergence - global divergence and local convergence - that triggers the emergence of political clusters; i.e. the creation of regions with the same policy. Furthermore, the average effectiveness in this computational world tends to follow a J-shaped curve, meaning that not only time is needed for a policy to deploy its effects, but that it also takes time for a country to find the best-suited policy. To conclude, diffusion is an emergent phenomenon from complex interactions and its outcomes as ensued from my model are in line with the theoretical expectations and the empirical evidence.Les méthodes d'analyse de biographie (event history analysis) permettent de mettre en évidence l'existence de phénomènes de diffusion et de les décrire, mais ne permettent pas d'en étudier le processus. Les simulations informatiques, grâce aux performances croissantes des ordinateurs, rendent possible l'étude des processus en tant que tels. Cette thèse, basée sur le modèle théorique développé par Braun et Gilardi (2006), présente une simulation centrée sur les agents des phénomènes de diffusion des politiques. Le point de départ de ce travail met en lumière, au niveau théorique, les principaux facteurs de changement internes à un pays : la préférence pour une politique donnée, l'efficacité de cette dernière, les contraintes institutionnelles, l'idéologie, et les principaux mécanismes de diffusion que sont l'apprentissage, la compétition, l'émulation et la coercition. La diffusion, définie par l'interdépendance des différents acteurs, est un système complexe dont l'étude est rendue possible par les simulations centrées sur les agents. Au niveau méthodologique, nous présenterons également les principaux concepts sous-jacents aux simulations, notamment la complexité et l'émergence. De plus, l'utilisation de simulations informatiques implique le développement d'un algorithme et sa programmation. Cette dernière réalisée, les agents peuvent interagir, avec comme résultat l'émergence d'un phénomène de diffusion, dérivé de l'apprentissage, où le choix d'un agent dépend en grande partie de ceux faits par ses voisins. De plus, ce phénomène suit une courbe en S caractéristique, poussant à la création de régions politiquement identiques, mais divergentes au niveau globale. Enfin, l'efficacité moyenne, dans ce monde simulé, suit une courbe en J, ce qui signifie qu'il faut du temps, non seulement pour que la politique montre ses effets, mais également pour qu'un pays introduise la politique la plus efficace. En conclusion, la diffusion est un phénomène émergent résultant d'interactions complexes dont les résultats du processus tel que développé dans ce modèle correspondent tant aux attentes théoriques qu'aux résultats pratiques.

Second cranio-facial malignancies in hereditary retinoblastoma survivors previously treated with radiation therapy: clinic and radiologic characteristics and survival outcomes.

INTRODUCTION: Hereditary retinoblastoma survivors have an increased risk for cranio-facial second primary tumours (SPT), especially after treatment with external beam radiotherapy (EBRT). This multicentre study evaluates the clinical and imaging characteristics and outcomes of cranio-facial SPTs in irradiated retinoblastoma survivors. PATIENTS AND METHODS: Clinical and radiological data of 42 hereditary retinoblastoma patients with 44 second and third malignancies were reviewed. Radiological data included anatomic location and computed tomography (CT) and magnetic resonance (MR) characteristics. Cox regression and likelihood ratio chi-square test were used to evaluate differences in patients' survival rates. RESULTS: Cranio-facial SPTs were diagnosed at a median age of 13 years. Histological types included osteosarcomas (43%), rhabdomyosarcomas (20%) (57% embryonal, 43% alveolar) and a variety of other types of SPT (37%). Predilection sites were: temporal fossa (39%), ethmoid sinus (23%), orbit (18%), maxillary sinus (16%) and intracranial dura mater (4%). Most of the osteosarcomas (78%) and rhabdomyosarcomas (80%) occurred in patients treated with EBRT in the first year-of-life. Treatment of SPTs with a microscopically complete surgical resection led to a significantly better 5-year overall survival (OS) (P=0.017) and event-free survival (EFS) (P=0.012) compared to patients treated without surgery or incomplete resection (OS: 83% versus 52%; EFS: 80% versus 47%). CONCLUSIONS: Osteosarcomas and rhabdomyosarcomas are the most common cranio-facial SPTs in irradiated hereditary retinoblastoma survivors, which develop in specific locations and occur predominantly in patients irradiated in their first year-of-life. Microscopically complete surgical resection of SPTs is a major prognostic factor, suggesting the potential benefit of early detection by imaging.

Improving Outcomes: A Strategy for Cancer: Third Annual Report

The third annual Improving Outcomes: A Strategy for Cancer report, in partnership with NHS England and Public Health England, reports on: significant developments in cancer screening - particularly on the first phase of introducing Bowel Scope Screening (BSS) activity to promote earlier diagnosis of symptomatic cancers through the Be Clear on Cancer campaigns and the associated work with primary and secondary care progress in ensuring better access for all to the best possible treatment significant developments in the collection and reporting of new datasets and the analysis of information to drive improvements and inform patients

Therapeutic drug monitoring of imatinib: Bayesian and alternative methods to predict trough levels

Background: The imatinib trough plasma concentration (C(min)) correlates with clinical response in cancer patients. Therapeutic drug monitoring (TDM) of plasma C(min) is therefore suggested. In practice, however, blood sampling for TDM is often not performed at trough. The corresponding measurement is thus only remotely informative about C(min) exposure. Objectives: The objectives of this study were to improve the interpretation of randomly measured concentrations by using a Bayesian approach for the prediction of C(min), incorporating correlation between pharmacokinetic parameters, and to compare the predictive performance of this method with alternative approaches, by comparing predictions with actual measured trough levels, and with predictions obtained by a reference method, respectively. Methods: A Bayesian maximum a posteriori (MAP) estimation method accounting for correlation (MAP-ρ) between pharmacokinetic parameters was developed on the basis of a population pharmacokinetic model, which was validated on external data. Thirty-one paired random and trough levels, observed in gastrointestinal stromal tumour patients, were then used for the evaluation of the Bayesian MAP-ρ method: individual C(min) predictions, derived from single random observations, were compared with actual measured trough levels for assessment of predictive performance (accuracy and precision). The method was also compared with alternative approaches: classical Bayesian MAP estimation assuming uncorrelated pharmacokinetic parameters, linear extrapolation along the typical elimination constant of imatinib, and non-linear mixed-effects modelling (NONMEM) first-order conditional estimation (FOCE) with interaction. Predictions of all methods were finally compared with 'best-possible' predictions obtained by a reference method (NONMEM FOCE, using both random and trough observations for individual C(min) prediction). Results: The developed Bayesian MAP-ρ method accounting for correlation between pharmacokinetic parameters allowed non-biased prediction of imatinib C(min) with a precision of ±30.7%. This predictive performance was similar for the alternative methods that were applied. The range of relative prediction errors was, however, smallest for the Bayesian MAP-ρ method and largest for the linear extrapolation method. When compared with the reference method, predictive performance was comparable for all methods. The time interval between random and trough sampling did not influence the precision of Bayesian MAP-ρ predictions. Conclusion: Clinical interpretation of randomly measured imatinib plasma concentrations can be assisted by Bayesian TDM. Classical Bayesian MAP estimation can be applied even without consideration of the correlation between pharmacokinetic parameters. Individual C(min) predictions are expected to vary less through Bayesian TDM than linear extrapolation. Bayesian TDM could be developed in the future for other targeted anticancer drugs and for the prediction of other pharmacokinetic parameters that have been correlated with clinical outcomes.

Conference 2012: Reaching Children's Outcomes: An All-Island Approach

Presentations from the inaugural conference of the Children's Research Network for Ireland and Northern Ireland, 26th September 2012.

Blood glucose level on postoperative day 1 is predictive of adverse outcomes after cardiovascular surgery.

AIM: Hyperglycaemia is now a recognized predictive factor of morbidity and mortality after coronary artery bypass grafting (CABG). For this reason, we aimed to evaluate the postoperative management of glucose control in patients undergoing cardiovascular surgery, and to assess the impact of glucose levels on in-hospital mortality and morbidity. METHODS: This was a retrospective study investigating the association between postoperative blood glucose and outcomes, including death, post-surgical complications, and length of stay in the intensive care unit (ICU) and in hospital. RESULTS: A total of 642 consecutive patients were enrolled into the study after cardiovascular surgery (CABG, carotid endarterectomy and bypass in the lower limbs). Patients' mean age was 68+/-10 years, and 74% were male. In-hospital mortality was 5% in diabetic patients vs 2% in non-diabetic patients (OR: 1.66, P=0.076). Having blood glucose levels in the upper quartile range (> or =8.8 mmol/L) on postoperative day 1 was independently associated with death (OR: 10.16, P=0.0002), infectious complications (OR: 1.76, P=0.04) and prolonged ICU stay (OR: 3.10, P<0.0001). Patients presenting with three or more hypoglycaemic episodes (<4.1 mmol/L) had increased rates of mortality (OR: 9.08, P<0.0001) and complications (OR: 8.57, P<0.0001). CONCLUSION: Glucose levels greater than 8.8 mmol/L on postoperative day 1 and having three or more hypoglycaemic episodes in the postoperative period were predictive of mortality and morbidity among patients undergoing cardiovascular surgery. This suggests that a multidisciplinary approach may be able to achieve better postoperative blood glucose control.

Surgical outcomes following repeat non-penetrating glaucoma filtration surgery at a tertiary referral centre

Purpose:To determine the surgical outcomes of patients undergoing repeat deep sclerectomies with collagen implant (DSCI) at a tertiary referral centre. Methods:The medical notes of 208 patients undergoing multiple DSCI were reviewed. Those undergoing repeat DSCI were identified and post operative data for each DSCI were analysed. Group A: the first DSCI; group B: second DSCI; group C: third DSCI. Results:Mean age was 66.8 ±13.0 years. At 12 months, percentage of mean IOP reduction in groups were 18%, 24% and 17% respectively. Mean interval to starting glaucoma medications, re-operation, mitomycin injection and goniopuncture all decreased as the number of operations increased. There was a significant reduction in complete success rates between groups A and B and groups B and C. Few minor complications were observed in all 3 groups. Conclusions:Despite the possibility of bleb independent outflow pathways in patients undergoing non-penetrating surgery, there are significantly reduced success rates in eyes undergoing repeat DSCI. This has important implications for the choice of subsequent operations in patients who have failed non-penetrating filtration surgery.

Spinal cord ischemia : clinical and imaging patterns, pathogenesis, and outcomes in 27 patients

RÉSUMÉ Introduction: l'histoire naturelle et la physiopathologie des infarctus de la moelle épinière restent largement inconnues. En effet, la plupart des études cliniques portent sur des patients qui ont souffert d'infarctus médullaire secondaire à des chirurgies aortiques ou des hypotensions prolongées. Méthode: ce travail analyse les données cliniques, le laboratoire, l'imagerie (IRM) et l'évolution de 27 patients souffrant d'infarctus de la moelle épinière admis dans le service de Neurologie du CHUV. Parmi ces patients, il y avait 11 hommes et 16 femmes (âge moyen de 56 ans, tranche d'âge de 19 à 80 ans). Résultats: dix patients (37%) souffraient d'infarctus de l'artère spinale antérieure, 4 (15%) d'infarctus unilatéraux antérieurs, 4 (15%) unilatéraux postérieurs, 3 (11%) d'infarctus centraux, 2 (7%) d'infarctus des artères spinales postérieures, 2 (7%) d'infarctus transverse tandis que 2 patients présentaient des tableaux cliniques inclassables. Vingt patients (74%) n'avaient pas d'étiologie identifiable. Les patients avec infarctus centraux ou transverses présentaient fréquemment (40%) des artériopathies périphériques et tous les infarctus transverses survenaient à la suite d'hypotensions artérielles prolongées. Le début de tous les autres types d'infarctus était associé à des facteurs mécaniques (p=0.02} et ces patients avaient fréquemment des pathologies du rachis (p=0.003) au niveau de la lésion médullaire. Dans ces cas, les données cliniques suggèrent une lésion d'une racine nerveuse au niveau de l'infarctus médullaire compromettant mécaniquement le flux de son artère radiculaire. L'évolution clinique était généralement favorable, seuls 13 patients (48%) présentaient une atteinte significative de la marche à la sortie de l'hôpital. Conclusion: ce travail montre qu'il existe 2 types principaux d'infarctus de la moelle épinière : d'une part les infarctus dans le territoire d'une artère radiculaire (infarctus de l'artère spinale antérieure, des artères spinales postérieures et infarctus unilatéraux) et d'autre part les hypoperfusions régionales globales de la moelle épinière (infarctus centraux et transverses). Chacune de ces 2 catégories d'infarctus ont des caractéristiques cliniques, radiologiques, physiopathologiques et pronostiques distinctes.

A randomized controlled trial of the effectiveness of a Computer-Assisted Cognitive Remediation (CACR) program in adolescents with psychosis or at high risk of psychosis: Short term and long term outcomes

The present study investigates the short- and long-term outcomes of a computer-assisted cognitive remediation (CACR) program in adolescents with psychosis or at high risk. 32 adolescents participated in a blinded 8-week randomized controlled trial of CACR treatment compared to computer games (CG). Clinical and neuropsychological evaluations were undertaken at baseline, at the end of the program and at 6-month. At the end of the program (n = 28), results indicated that visuospatial abilities (Repeatable Battery for the Assessment of Neuropsychological Status, RBANS; P = .005) improved signifi cantly more in the CACR group compared to the CG group. Furthermore, other cognitive functions (RBANS), psychotic symptoms (Positive and Negative Symptom Scale) and psychosocial functioning (Social and Occupational Functioning Assessment Scale) improved signifi cantly, but at similar rates, in the two groups. At long term (n = 22), cognitive abilities did not demonstrated any amelioration in the control group while, in the CACR group, signifi cant long-term improvements in inhibition (Stroop; P = .040) and reasoning (Block Design Test; P = .005) were observed. In addition, symptom severity (Clinical Global Improvement) decreased signifi cantly in the control group (P = .046) and marginally in the CACR group (P = .088). To sum up, CACR can be successfully administered in this population. CACR proved to be effective over and above CG for the most intensively trained cognitive ability. Finally, on the long-term, enhanced reasoning and inhibition abilities, which are necessary to execute higher-order goals or to adapt behavior to the ever-changing environment, were observed in adolescents benefi ting from a CACR.

Long-term outcomes of school-based treatment for control of urinary schistosomiasis: a review of experience in Coast Province, Kenya

Urinary schistosomiasis remains a significant burden for Africa and the Middle East. The success of population-based control programs will depend on their impact, over many years, on Schistosoma haematobium reinfection and associated disease. In a multi-year (1984-1992) control program in Kenya, we examined risk for S. haematobium reinfection and late disease during and after annual school-based treatment. In this setting, long-term risk of new infection was independently associated with location, age, hematuria, and incomplete treatment, but not with sex or frequency of water contact. Thus, very local environmental features and age-related factors played an important role in S. haematobium transmission, such that population-based control programs should optimally tailor their efforts to local conditions on a village-by-village basis. In 2001-2002, the late benefits of earlier participation in school-based antischistosomal therapy were estimated in a cohort of formerly-treated adult residents compared to never-treated adults from the same villages. Among age-matched subjects, current infection prevalence was lower among those who had received remote therapy. In addition, prevalence of bladder abnormality was lower in the treated group, who were free of severe bladder disease. Treatment of affected adults resulted in rapid resolution of infection and any detectable bladder abnormalities. We conclude that continued treatment into adulthood, as well as efforts at long-term prevention of infection (transmission control) are necessary to achieve optimal morbidity control in affected communities.

Referral patterns and outcomes in noncritically ill patients with hospital-acquired acute kidney injury.

Background and objectives Despite modern treatment, the case fatality rate of hospital-acquired acute kidney injury (HA-AKI) is still high. We retrospectively described the prevalence and the outcome of HA-AKI without nephrology referral (nrHA-AKI) and late referred HA-AKI patients to nephrologists (lrHA-AKI) compared with early referral patients (erHA-AKI) with respect to renal function recovery, renal replacement therapy (RRT) requirement, and in-hospital mortality of HA-AKI. Design, setting, participants, & measurements Noncritically ill patients admitted to the tertiary care academic center of Lausanne, Switzerland, between 2004 and 2008 in the medical and surgical services were included. Acute kidney injury was defined using the Acute Kidney Injury Network (AKIN) classification. Results During 5 years, 4296 patients (4.12% of admissions) experienced 4727 episodes of HA-AKI during their hospital stay. The mean ± SD age of the patients was 61 ± 15 years with a 55% male predominance. There were 958 patients with nrHA-AKI (22.3%) and 2504 patients with lrHA-AKI (58.3%). RRT was required in 31% of the patients with lrHA-AKI compared with 24% of the patients with erHA-AKI. In the multiple risk factor analysis, compared with erHA-AKI, nrHA-AKI and lrHA-AKI were significantly associated with worse renal outcome and higher in-hospital mortality. Conclusions These data suggest that HA-AKI is frequent and the patients with nrHA-AKI or lrHA-AKI are at increased risk for in-hospital morbidity and mortality.

Good clinical and cost outcomes using dexrazoxane to treat accidental epirubicin extravasation

A 75-year-old man diagnosed with lower esophageal adenocarcinoma suffered from epirubicin extravasation during the second cycle of neoadjuvant chemotherapy with epirubicin and oxaliplatin. A full recovery was achieved after treatment with dexrazoxane (Cardioxane® ). This is the first time in our hospital that extravasation of an anthracycline has been treated with dexrazoxane. We used Cardioxane® , approved for the prevention of anthracycline-induced cardiotoxicity, while Savene® is indicated for the treatment of anthracycline extravasation. The treatment was effective, and the selection of Cardioxane® (seven-fold cheaper than Savene® ) yielded a cost saving. Consequently, Cardioxane® has been included in our guidelines for anthracycline extravasation.

Conditional compositional biplots: theory and application

The biplot has proved to be a powerful descriptive and analytical tool in many areasof applications of statistics. For compositional data the necessary theoreticaladaptation has been provided, with illustrative applications, by Aitchison (1990) andAitchison and Greenacre (2002). These papers were restricted to the interpretation ofsimple compositional data sets. In many situations the problem has to be described insome form of conditional modelling. For example, in a clinical trial where interest isin how patients’ steroid metabolite compositions may change as a result of differenttreatment regimes, interest is in relating the compositions after treatment to thecompositions before treatment and the nature of the treatments applied. To study thisthrough a biplot technique requires the development of some form of conditionalcompositional biplot. This is the purpose of this paper. We choose as a motivatingapplication an analysis of the 1992 US President ial Election, where interest may be inhow the three-part composition, the percentage division among the three candidates -Bush, Clinton and Perot - of the presidential vote in each state, depends on the ethniccomposition and on the urban-rural composition of the state. The methodology ofconditional compositional biplots is first developed and a detailed interpretation of the1992 US Presidential Election provided. We use a second application involving theconditional variability of tektite mineral compositions with respect to major oxidecompositions to demonstrate some hazards of simplistic interpretation of biplots.Finally we conjecture on further possible applications of conditional compositionalbiplots