University of London

About: University of London is a education organization based out in London, United Kingdom. It is known for research contribution in the topics: Population & Public health. The organization has 44838 authors who have published 88086 publications receiving 4002499 citations. The organization is also known as: London University & Lond..

Structural risk minimization over data-dependent hierarchies

Abstract: The paper introduces some generalizations of Vapnik's (1982) method of structural risk minimization (SRM). As well as making explicit some of the details on SRM, it provides a result that allows one to trade off errors on the training sample against improved generalization performance. It then considers the more general case when the hierarchy of classes is chosen in response to the data. A result is presented on the generalization performance of classifiers with a "large margin". This theoretically explains the impressive generalization performance of the maximal margin hyperplane algorithm of Vapnik and co-workers (which is the basis for their support vector machines). The paper concludes with a more general result in terms of "luckiness" functions, which provides a quite general way for exploiting serendipitous simplicity in observed data to obtain better prediction accuracy from small training sets. Four examples are given of such functions, including the Vapnik-Chervonenkis (1971) dimension measured on the sample.

Extra-pleural pneumonectomy versus no extra-pleural pneumonectomy for patients with malignant pleural mesothelioma: clinical outcomes of the Mesothelioma and Radical Surgery (MARS) randomised feasibility study

Abstract: Background The eff ects of extra-pleural pneumonectomy (EPP) on survival and quality of life in patients with malignant pleural mesothelioma have, to our knowledge, not been assessed in a randomised trial. We aimed to assess the clinical outcomes of patients who were randomly assigned to EPP or no EPP in the context of trimodal therapy in the Mesothelioma and Radical S urgery (MARS) feasibility study. Methods MARS was a multicentre randomised controlled trial in 12 UK hospitals. Patients aged 18 years or older who had pathologically confi rmed mesothelioma and were deemed fi t enough to undergo trimodal therapy were included. In a prerandomisation registration phase, all patients underwent induction platinum-based chemotherapy followed by clinical review. Af ter further consent, patients were randomly assigned (1:1) to EPP followed by postoperative hemithorax irradiation or to no EPP. Randomisation was done centrally with computer-generated permuted blocks stratifi ed by surgical centre. The main endpoints were feasibility of randomly assigning 50 patients in 1 year (results detailed in another report), proportion randomised who received treatment, proportion eligible (registered) who proceeded to randomisation, perioperative mortality, and quality of life. Patients and investigators were not masked to treatment allocation. This is the principal report of the MARS study; all patients have been recruited. Analyses were by intention to treat. This trial is registered, number ISRCTN95583524. Findings Between Oct 1, 2005, and Nov 3, 2008, 112 patients were registered and 50 were subsequently randomly assigned: 24 to EPP and 26 to no EPP. The main reasons for not proceeding to randomisation were disease progression (33 patients), inoperability (fi ve patients), and patient choice (19 patients). EPP was completed satisfactorily in 16 of 24 patients assigned to EPP; in fi ve patients EPP was not started and in three patients it was abandoned. Two patients in the EPP group died within 30 days and a further patient died without leaving hospital. One patient in the no EPP group died perioperatively af ter receiving EPP off trial in a non-MARS centre. The hazard ratio [HR] for overall survival between the EPP and no EPP groups was 1·90 (95% CI 0·92–3·93; exact p=0·082), and after adjustment for sex, histological subtype, stage, and age at randomisation the HR was 2·75 (1·21–6·26; p=0·016). Median survival was 14·4 months (5·3–18·7) for the EPP group and 19·5 months (13·4 to time not yet reached) for the no EPP group. Of the 49 randomly assigned patients who consented to quality of life assessment (EPP n=23; no EPP n=26), 12 patients in the EPP group and 19 in the no EPP group completed the quality of life questionnaires. Although median quality of life scores were lower in the EPP group than the no EPP group, no signifi cant diff erences between groups were reported in the quality of life analyses. There were ten serious adverse events reported in the EPP group and two in the no EPP group. Interpretation In view of the high morbidity associated with EPP in this trial and in other non-randomised studies a larger study is not feasible. These data, although limited, suggest that radical surgery in the form of EPP within trimodal therapy off ers no benefi t and possibly harms patients.

Sequence variants in IL10, ARPC2 and multiple other loci contribute to ulcerative colitis susceptibility

Abstract: Inflammatory bowel disease (IBD) typically manifests as either ulcerative colitis (UC) or Crohn's disease (CD). Systematic identification of susceptibility genes for IBD has thus far focused mainly on CD, and little is known about the genetic architecture of UC. Here we report a genome-wide association study with 440,794 SNPs genotyped in 1,167 individuals with UC and 777 healthy controls. Twenty of the most significantly associated SNPs were tested for replication in three independent European case-control panels comprising a total of 1,855 individuals with UC and 3,091 controls. Among the four consistently replicated markers, SNP rs3024505 immediately flanking the IL10 (interleukin 10) gene on chromosome 1q32.1 showed the most significant association in the combined verification samples (P = 1.35 x 10(-12); OR = 1.46 (1.31-1.62)). The other markers were located in ARPC2 and in the HLA-BTNL2 region. Association between rs3024505 and CD (1,848 cases, 1,804 controls) was weak (P = 0.013; OR = 1.17 (1.01-1.34)). IL10 is an immunosuppressive cytokine that has long been proposed to influence IBD pathophysiology. Our findings strongly suggest that defective IL10 function is central to the pathogenesis of the UC subtype of IBD.