A simple permutation test for clusteredness

Hierarchical clustering is a popular method for finding structure in multivariate data,resulting in a binary tree constructed on the particular objects of the study, usually samplingunits. The user faces the decision where to cut the binary tree in order to determine the numberof clusters to interpret and there are various ad hoc rules for arriving at a decision. A simplepermutation test is presented that diagnoses whether non-random levels of clustering are presentin the set of objects and, if so, indicates the specific level at which the tree can be cut. The test isvalidated against random matrices to verify the type I error probability and a power study isperformed on data sets with known clusteredness to study the type II error.

A randomized concave programming method for choice network revenue management

Models incorporating more realistic models of customer behavior, as customers choosing froman offer set, have recently become popular in assortment optimization and revenue management.The dynamic program for these models is intractable and approximated by a deterministiclinear program called the CDLP which has an exponential number of columns. However, whenthe segment consideration sets overlap, the CDLP is difficult to solve. Column generationhas been proposed but finding an entering column has been shown to be NP-hard. In thispaper we propose a new approach called SDCP to solving CDLP based on segments and theirconsideration sets. SDCP is a relaxation of CDLP and hence forms a looser upper bound onthe dynamic program but coincides with CDLP for the case of non-overlapping segments. Ifthe number of elements in a consideration set for a segment is not very large (SDCP) can beapplied to any discrete-choice model of consumer behavior. We tighten the SDCP bound by(i) simulations, called the randomized concave programming (RCP) method, and (ii) by addingcuts to a recent compact formulation of the problem for a latent multinomial-choice model ofdemand (SBLP+). This latter approach turns out to be very effective, essentially obtainingCDLP value, and excellent revenue performance in simulations, even for overlapping segments.By formulating the problem as a separation problem, we give insight into why CDLP is easyfor the MNL with non-overlapping considerations sets and why generalizations of MNL posedifficulties. We perform numerical simulations to determine the revenue performance of all themethods on reference data sets in the literature.

The contributions of rare objects in correspondence analysis

Correspondence analysis, when used to visualize relationships in a table of counts(for example, abundance data in ecology), has been frequently criticized as being too sensitiveto objects (for example, species) that occur with very low frequency or in very few samples. Inthis statistical report we show that this criticism is generally unfounded. We demonstrate this inseveral data sets by calculating the actual contributions of rare objects to the results ofcorrespondence analysis and canonical correspondence analysis, both to the determination ofthe principal axes and to the chi-square distance. It is a fact that rare objects are oftenpositioned as outliers in correspondence analysis maps, which gives the impression that theyare highly influential, but their low weight offsets their distant positions and reduces their effecton the results. An alternative scaling of the correspondence analysis solution, the contributionbiplot, is proposed as a way of mapping the results in order to avoid the problem of outlying andlow contributing rare objects.

Factor analysis of the North American Spine Society outcome assessment instrument: a study based on a spine registry of patients treated with lumbar and cervical disc arthroplasty.

BACKGROUND CONTEXT: Studies involving factor analysis (FA) of the items in the North American Spine Society (NASS) outcome assessment instrument have revealed inconsistent factor structures for the individual items. PURPOSE: This study examined whether the factor structure of the NASS varied in relation to the severity of the back/neck problem and differed from that originally recommended by the developers of the questionnaire, by analyzing data before and after surgery in a large series of patients undergoing lumbar or cervical disc arthroplasty. STUDY DESIGN/SETTING: Prospective multicenter observational case series. PATIENT SAMPLE: Three hundred ninety-one patients with low back pain and 553 patients with neck pain completed questionnaires preoperatively and again at 3 to 6 and 12 months follow-ups (FUs), in connection with the SWISSspine disc arthroplasty registry. OUTCOME MEASURES: North American Spine Society outcome assessment instrument. METHODS: First, an exploratory FA without a priori assumptions and subsequently a confirmatory FA were performed on the 17 items of the NASS-lumbar and 19 items of the NASS-cervical collected at each assessment time point. The item-loading invariance was tested in the German version of the questionnaire for baseline and FU. RESULTS: Both NASS-lumbar and NASS-cervical factor structures differed between baseline and postoperative data sets. The confirmatory analysis and item-loading invariance showed better fit for a three-factor (3F) structure for NASS-lumbar, containing items on "disability," "back pain," and "radiating pain, numbness, and weakness (leg/foot)" and for a 5F structure for NASS-cervical including disability, "neck pain," "radiating pain and numbness (arm/hand)," "weakness (arm/hand)," and "motor deficit (legs)." CONCLUSIONS: The best-fitting factor structure at both baseline and FU was selected for both the lumbar- and cervical-NASS questionnaires. It differed from that proposed by the originators of the NASS instruments. Although the NASS questionnaire represents a valid outcome measure for degenerative spine diseases, it is able to distinguish among all major symptom domains (factors) in patients undergoing lumbar and cervical disc arthroplasty; overall, the item structure could be improved. Any potential revision of the NASS should consider its factorial structure; factorial invariance over time should be aimed for, to allow for more precise interpretations of treatment success.

How accurate is the current picture of human genetic variation?

Our understanding of the distribution of worldwide human genomic diversity has greatly increased over recent years thanks to the availability of large data sets derived from short tandem repeats (STRs), insertion deletion polymorphisms (indels) and single nucleotide polymorphisms (SNPs). A concern, however, is that the current picture of worldwide human genomic diversity may be inaccurate because of biases in the selection process of genetic markers (so-called 'ascertainment bias'). To evaluate this problem, we first compared the distribution of genomic diversity between these three types of genetic markers in the populations from the HGDP-CEPH panel for evidence of bias or incongruities. In a second step, using a very relaxed set of criteria to prevent the intrusion of bias, we developed a new set of unbiased STR markers and compared the results against those from available panels. Contrarily to recent claims, our results show that the STR markers suffer from no discernible bias, and can thus be used as a baseline reference for human genetic diversity and population differentiation. The bias on SNPs is moderate compared to that on the set of indels analysed, which we recommend should be avoided for work describing the distribution of human genetic diversity or making inference on human settlement history.

Gene expression profiling of human glioblastomas for the identification of predictive factors and characterisation of molecular mechanism implicated in response to therapy and outcome

ABSTRACT Poor outcome for glioblastoma patients is largely due to resistance to chemoradiation therapy. While epigenetic inactivation of MGMT mediated DNA repair is highly predictive for benefit from the alkylating agent therapy Temozolomide, additional mechanisms for resistance associated with molecular alterations exist. Furthermore, new concepts in cancer suggest that resistance to treatment may be linked to cancer stem cells that escape therapy and act as source for tumour recurrence. We determined gene expression signatures associated with outcome in glioblastoma patients enrolled in a phase II and phase III clinical trial establishing the new combination therapy of radiation plus concomitant and adjuvant Temozolomide. Correlating stable gene clusters emerging from unsupervised analysis with survival of 42 treated patients identified a number of biological processes associated with outcome. Most prominent, a gene cluster dominated by HOX genes and comprising PROM1, was associated with resistance. PROM1 encodes CD133, a marker for a subpopulation of tumour cells enriched for glioblastoma stem- like cells. The core of this correlated HOX cluster was comprised in the top genes of a "self-renewal signature" defined in a mouse model for MLL-AF9 initiated leukaemia. The association of the HOX gene cluster with tumour resistance was confirmed in two external data sets of 146 malignant glioma As additional resistance factors we identified over-expression of the epidermal growth factor receptor gene, EGFR, while increased gene expression related to biological features of tumour host interaction, including markers for tumour vascular and cell adhesion, and innate immune response, were associated with better outcome. The "self-renewal" signature associated with resistance to the new combination chemoradiation therapy provides first clinical evidence that glioma stem like cells may implicated in resistance in a uniformly treated cohort of glioblastoma patients. This study underlines the need to target the tumour stem cell compartment, and provides some testable hypothesis for biological mechanisms relevant for malignant behaviour of glioblastoma that may be targeted in new treatment approaches. Résumé Le glioblastome, tumeur cérébrale primaire maligne la plus fréquente, est connue pour son mauvais pronostique. Des avancées chimiothérapeutiques récentes avec des agents alkylants comme le témozolomide (TMZ), ont permis une amélioration notable dans la survie de certains patients. Les bénéficiaires ont la caractéristique commune de présenter une particularité génétique, la methylation du MGMT (methylguanine methyltransferase). Néanmoins, d'autres mécanismes de résistance en fonction des aberrations moléculaires existent. Nous avons établi les profils d'expressions génétiques des patients traités par irradiation et TMZ dans des études cliniques de phase II et III. En combinant des méthodes non-supervisées et supervisées, de l'étude de la cohorte des patients traités nous avons découvert des groupes de gènes associés à la survie. Un ensemble de gènes contenant les gènes Hox semble lié au mécanisme de résistance au traitement. Récemment, les gènes Hox ont été décrits comme faisant partie d"une signature d'autorenouvellement (self-renewal) des cellules souches cancéreuses de la leucémie. L'autorenouvellement est un processus grâce auquel les cellules souches se maintiennent tout au long de la vie. Cette association à la résistance est confirmée dans deux autres études indépendantes. Un autre facteur de résistance au traitement est la surexpression du gène EGFR. D'autre part, deux groupes de gènes associés à la relation entre hôte-tumeur tels que les marqueurs des vaisseaux tumoraux et de la réponse immunitaire innée s'avèrent avoir un effet positif sur la survie des patients traités. La découverte de la signature d'autorenouvellement comme facteur de résistance à la nouvelle chimio-radiothérapie offre une preuve clinique que les cellules souches cancéreuses sont impliquées dans la résistance au traitement. If est donc logique de penser que le traitement ciblé contre des cellules souches cancéreuses va dans l'avenir permettre des thérapies anticancéreuses plus performantes.

The effect of unemployment spells on subsequent wages in Spain

In our analysis we try and recover the wage loss from unemploymentin Spain and see how it is affected by previous unemploymentexperience, unemployment duration, eligibility for unemploymentbenefits, and previous wages. We also study its variations acrossgroups. Our main conclusion is that while there is some evidencethat labour market rigidities tend to lower it, the wage loss ofdisplaced workers is remarkably high: more than 30%, that is,twice the equivalent figure for the US and France. Wages in Spainsuffer from a serious mismeasurement problems that we do our best tocontrol, so that our results are less robust than the ones thatwould be obtained with better data sets. However, they indicate a large level of wage flexibility in Spain.

Dynamic perceptual mapping

Perceptual maps have been used for decades by market researchers to illuminatethem about the similarity between brands in terms of a set of attributes, to position consumersrelative to brands in terms of their preferences, or to study how demographic and psychometricvariables relate to consumer choice. Invariably these maps are two-dimensional and static. Aswe enter the era of electronic publishing, the possibilities for dynamic graphics are opening up.We demonstrate the usefulness of introducing motion into perceptual maps through fourexamples. The first example shows how a perceptual map can be viewed in three dimensions,and the second one moves between two analyses of the data that were collected according todifferent protocols. In a third example we move from the best view of the data at the individuallevel to one which focuses on between-group differences in aggregated data. A final exampleconsiders the case when several demographic variables or market segments are available foreach respondent, showing an animation with increasingly detailed demographic comparisons.These examples of dynamic maps use several data sets from marketing and social scienceresearch.

Social security and democracy

Many political economic theories use and emphasize the process of votingin their explanation of the growth of Social Security, governmentspending, and other public policies. But is there an empirical connectionbetween democracy and Social Security program size or design? Using somenew international data sets to produce both country-panel econometricestimates as well as case studies of South American and southern Europeancountries, we find that Social Security policy varies according toeconomic and demographic factors, but that very different politicalhistories can result in the same Social Security policy. We find littlepartial effect of democracy on the size of Social Security budgets, onhow those budgets are allocated, or how economic and demographic factorsaffect Social Security. If there is any observed difference, democraciesspend a little less of their GDP on Social Security, grow their budgetsa bit more slowly, and cap their payroll tax more often, than doeconomically and demographically similar nondemocracies. Democracies andnondemocracies are equally likely to have benefit formulas inducingretirement and, conditional on GDP per capita, equally likely to induceretirement with a retirement test vs. an earnings test.

Disability, capacity for work and the business cycle: An international perspective

An important policy issue in recent years concerns the number of people claimingdisability benefits for reasons of incapacity for work. We distinguish between workdisability , which may have its roots in economic and social circumstances, and healthdisability which arises from clear diagnosed medical conditions. Although there is a linkbetween work and health disability, economic conditions, and in particular the businesscycle and variations in the risk of unemployment over time and across localities, mayplay an important part in explaining both the stock of disability benefit claimants andinflows to and outflow from that stock. We employ a variety of cross?country andcountry?specific household panel data sets, as well as administrative data, to testwhether disability benefit claims rise when unemployment is higher, and also toinvestigate the impact of unemployment rates on flows on and off the benefit rolls. Wefind strong evidence that local variations in unemployment have an importantexplanatory role for disability benefit receipt, with higher total enrolments, loweroutflows from rolls and, often, higher inflows into disability rolls in regions and periodsof above?average unemployment. Although general subjective measures of selfreporteddisability and longstanding illness are also positively associated withunemployment rates, inclusion of self?reported health measures does not eliminate thestatistical relationship between unemployment rates and disability benefit receipt;indeed including general measures of health often strengthens that underlyingrelationship. Intriguingly, we also find some evidence from the United Kingdom and theUnited States that the prevalence of self?reported objective specific indicators ofdisability are often pro?cyclical that is, the incidence of specific forms of disability arepro?cyclical whereas claims for disability benefits given specific health conditions arecounter?cyclical. Overall, the analysis suggests that, for a range of countries and datasets, levels of claims for disability benefits are not simply related to changes in theincidence of health disability in the population and are strongly influenced by prevailingeconomic conditions. We discuss the policy implications of these various findings.

The size and burden of mental disorders and other disorders of the brain in Europe 2010.

Aims: To provide 12-month prevalence and disability burden estimates of a broad range of mental and neurological disorders in the European Union (EU) and to compare these findings to previous estimates. Referring to our previous 2005 review, improved up-to-date data for the enlarged EU on a broader range of disorders than previously covered are needed for basic, clinical and public health research and policy decisions and to inform about the estimated number of persons affected in the EU. Method: Stepwise multi-method approach, consisting of systematic literature reviews, reanalyses of existing data sets, national surveys and expert consultations. Studies and data from all member states of the European Union (EU-27) plus Switzerland, Iceland and Norway were included. Supplementary information about neurological disorders is provided, although methodological constraints prohibited the derivation of overall prevalence estimates for mental and neurological disorders. Disease burden was measured by disability adjusted life years (DALY). Results: Prevalence: It is estimated that each year 38.2% of the EU population suffers from a mental disorder. Adjusted for age and comorbidity, this corresponds to 164.8 million persons affected. Compared to 2005 (27.4%) this higher estimate is entirely due to the inclusion of 14 new disorders also covering childhood/adolescence as well as the elderly. The estimated higher number of persons affected (2011: 165 m vs. 2005: 82 m) is due to coverage of childhood and old age populations, new disorders and of new EU membership states. The most frequent disorders are anxiety disorders (14.0%), insomnia (7.0%), major depression (6.9%), somatoform (6.3%), alcohol and drug dependence (>4%), ADHD (5%) in the young, and dementia (1-30%, depending on age). Except for substance use disorders and mental retardation, there were no substantial cultural or country variations. Although many sources, including national health insurance programs, reveal increases in sick leave, early retirement and treatment rates due to mental disorders, rates in the community have not increased with a few exceptions (i.e. dementia). There were also no consistent indications of improvements with regard to low treatment rates, delayed treatment provision and grossly inadequate treatment. Disability: Disorders of the brain and mental disorders in particular, contribute 26.6% of the total all cause burden, thus a greater proportion as compared to other regions of the world. The rank order of the most disabling diseases differs markedly by gender and age group; overall, the four most disabling single conditions were: depression, dementias, alcohol use disorders and stroke. Conclusion: In every year over a third of the total EU population suffers from mental disorders. The true size of "disorders of the brain" including neurological disorders is even considerably larger. Disorders of the brain are the largest contributor to the all cause morbidity burden as measured by DALY in the EU. No indications for increasing overall rates of mental disorders were found nor of improved care and treatment since 2005; less than one third of all cases receive any treatment, suggesting a considerable level of unmet needs. We conclude that the true size and burden of disorders of the brain in the EU was significantly underestimated in the past.Concerted priority action is needed at all levels, including substantially increased funding for basic, clinical and public health research in order to identify better strategies for improved prevention and treatment for isorders of the brain as the core health challenge of the 21st century. (C) 2011 Published by Elsevier B.V.

Definition of clinically distinct molecular subtypes in estrogen receptor-positive breast carcinomas through genomic grade.

PURPOSE: A number of microarray studies have reported distinct molecular profiles of breast cancers (BC), such as basal-like, ErbB2-like, and two to three luminal-like subtypes. These were associated with different clinical outcomes. However, although the basal and the ErbB2 subtypes are repeatedly recognized, identification of estrogen receptor (ER) -positive subtypes has been inconsistent. Therefore, refinement of their molecular definition is needed. MATERIALS AND METHODS: We have previously reported a gene expression grade index (GGI), which defines histologic grade based on gene expression profiles. Using this algorithm, we assigned ER-positive BC to either high-or low-genomic grade subgroups and compared these with previously reported ER-positive molecular classifications. As further validation, we classified 666 ER-positive samples into subtypes and assessed their clinical outcome. RESULTS: Two ER-positive molecular subgroups (high and low genomic grade) could be defined using the GGI. Despite tracking a single biologic pathway, these were highly comparable to the previously described luminal A and B classification and significantly correlated to the risk groups produced using the 21-gene recurrence score. The two subtypes were associated with statistically distinct clinical outcome in both systemically untreated and tamoxifen-treated populations. CONCLUSION: The use of genomic grade can identify two clinically distinct ER-positive molecular subtypes in a simple and highly reproducible manner across multiple data sets. This study emphasizes the important role of proliferation-related genes in predicting prognosis in ER-positive BC.

Risk management under a two-factor model of the term structure of interest rates

This paper presents several applications to interest rate risk managementbased on a two-factor continuous-time model of the term structure of interestrates previously presented in Moreno (1996). This model assumes that defaultfree discount bond prices are determined by the time to maturity and twofactors, the long-term interest rate and the spread (difference between thelong-term rate and the short-term (instantaneous) riskless rate). Several newmeasures of ``generalized duration" are presented and applied in differentsituations in order to manage market risk and yield curve risk. By means ofthese measures, we are able to compute the hedging ratios that allows us toimmunize a bond portfolio by means of options on bonds. Focusing on thehedging problem, it is shown that these new measures allow us to immunize abond portfolio against changes (parallel and/or in the slope) in the yieldcurve. Finally, a proposal of solution of the limitations of conventionalduration by means of these new measures is presented and illustratednumerically.

How well do individuals predict the selling prices of their homes?

Self-reported home values are widely used as a measure of housing wealth by researchers employing a variety of data sets and studying a number of different individual and household level decisions. The accuracy of this measure is an open empirical question, and requires some type of market assessment of the values reported. In this research, we study the predictive power of self-reported housing wealth when estimating sales prices utilizing the Health and Retirement Study. We find that homeowners, on average, overestimate the value of their properties by between 5% and 10%. More importantly, we are the first to document a strong correlation between accuracy and the economic conditions at the time of the purchase of the property (measured by the prevalent interest rate, the growth of household income, and the growth of median housing prices). While most individuals overestimate the value of their properties, those who bought during more difficult economic times tend to be more accurate, and in some cases even underestimate the value of their house. These results establish a surprisingly strong, likely permanent, and in many cases long-lived, effect of the initial conditions surrounding the purchases of properties, on how individuals value them. This cyclicality of the overestimation of house prices can provide some explanations for the difficulties currently faced by many homeowners, who were expecting large appreciations in home value to rescue them in case of increases in interest rates which could jeopardize their ability to live up to their financial commitments.

Automatic classification of MR scans in Alzheimer's disease.

To be diagnostically useful, structural MRI must reliably distinguish Alzheimer's disease (AD) from normal aging in individual scans. Recent advances in statistical learning theory have led to the application of support vector machines to MRI for detection of a variety of disease states. The aims of this study were to assess how successfully support vector machines assigned individual diagnoses and to determine whether data-sets combined from multiple scanners and different centres could be used to obtain effective classification of scans. We used linear support vector machines to classify the grey matter segment of T1-weighted MR scans from pathologically proven AD patients and cognitively normal elderly individuals obtained from two centres with different scanning equipment. Because the clinical diagnosis of mild AD is difficult we also tested the ability of support vector machines to differentiate control scans from patients without post-mortem confirmation. Finally we sought to use these methods to differentiate scans between patients suffering from AD from those with frontotemporal lobar degeneration. Up to 96% of pathologically verified AD patients were correctly classified using whole brain images. Data from different centres were successfully combined achieving comparable results from the separate analyses. Importantly, data from one centre could be used to train a support vector machine to accurately differentiate AD and normal ageing scans obtained from another centre with different subjects and different scanner equipment. Patients with mild, clinically probable AD and age/sex matched controls were correctly separated in 89% of cases which is compatible with published diagnosis rates in the best clinical centres. This method correctly assigned 89% of patients with post-mortem confirmed diagnosis of either AD or frontotemporal lobar degeneration to their respective group. Our study leads to three conclusions: Firstly, support vector machines successfully separate patients with AD from healthy aging subjects. Secondly, they perform well in the differential diagnosis of two different forms of dementia. Thirdly, the method is robust and can be generalized across different centres. This suggests an important role for computer based diagnostic image analysis for clinical practice.