Please activate JavaScript!
Please install Adobe Flash Player, click here for download

ISCB2014_abstract_book

66 ISCB 2014 Vienna, Austria • Abstracts - Oral PresentationsWednesday, 27th August 2014 • 9:00-10:48 Monday25thAugustTuesday26thAugustThursday28thAugustAuthorIndexPostersWednesday27thAugustSunday24thAugust to other three cases. We studied numerically the efficiency of the worst case maximin sample sizes instead of using others. Finally, an expression is derived that enables calculating optimal and maximin sample sizes that yield sufficient power to test the cost-effectiveness of two treatments.   C35.5 Student Conference Award Correcting for bias in the detection and validation of informative diagnostic tests DS Robertson1 , J Bowden1 , AT Prevost2 1 MRC Biostatistics Unit, Cambridge, United Kingdom, 2 King’s College London, London, United Kingdom   When developing a new diagnostic test for a disease, there are often mul- tiple candidate classifiers to choose from, and it is unclear if any will offer an improvement in performance compared to current technology. A two- stage design can be used to select a promising classifier (if one exists) in stage one for definitive validation in stage two. However, estimating the true properties of the chosen classifier is compli- cated by the first stage selection rules. In particular, the usual maximum likelihood estimator (MLE) that combines data from both stages will be bi- ased high. Consequently, confidence intervals and p-values flowing from the MLE will also be incorrect. Building on the results of Pepe et al. (SIM 28:762-779) and others, we de- rive the most efficient conditionally unbiased estimator and exact confi- dence intervals for a classifier’s sensitivity in a two-stage design with ar- bitrary selection rules; the condition being that the trial proceeds to the validation stage. We apply our estimation strategy to data from a recent family history screening tool validation study by Walter et al. (BJGP 63: 393-400), and are able to identify and successfully adjust for bias in the tool’s estimated sensitivity to detect those at high risk of breast cancer.   C35.6 Modelling and choice of cutoff in meta-analysis of diagnostic studies with varying cut-off value D Böhning1 1 Southampton Statistical Sciences Research Institute, University of Southampton, Southampton, United Kingdom   Meta-analysis of diagnostic studies is often done on the basis of one pair of sensititivity and specificity per study. For this kind of situation the sum- mary receiver operating charactistic has been suggested as a sutiable ap- proach sicne it accommodates a varying cut-off value across studies. This approach becomes less favourable if the cut-off value itself is the param- eter of primary interest. We suggest a range of generalized linear models as natural extension of Youden index, diagnostic odds ratio or the likelihood ratio. Some case studies will illustrate the suggested methodology. C36 Issues in multiple testing C36.1 An informative modification of the fallback procedure S Schmidt1 , W Brannath1 1 University of Bremen, Bremen, Germany   The fallback procedure is an extension of the hierarchical test allowing for a more flexible alpha allocation. It can be applied for example in dose finding studies. If interest is in extending the fallback procedure to simul- taneous confidence intervals, one may use the construction proposed by Strassburger and Bretz (Stat. Med. 2008; 27: 4914--4927). However, these intervals are not optimal in the sense that non-informative rejections may arise. This means that the confidence interval of a rejected null hypothesis may contain all parameters of the alternative and thus gives no useful information about the true value of the effect parameter. Guilbaud (Biometrical Journal 2009; 51: 721--735) exploited the fact that the fallback procedure is not alpha-exhaustive in order to improve this deficiency. However, a positive probability for non-informative rejections remains. We will present a modification of the fallback procedure with correspond- ing simultaneous confidence intervals which is informative in every case where a hypothesis is rejected. Our method is a straightforward extension of a former approach with respect to the hierarchical test. The main idea consists of a continuous parameter dependent level splitting after rejec- tion of a null hypothesis to test a nested family of informative hypotheses. We will explain our idea, illustrate it by a graphical description and com- pare it to the approach of Guilbaud by simulations in the context of a clini- cal trial.   C36.2 Confirmatory testing for a beneficial treatment effect in dose-response studies using MCP-Mod F König1 , B Bornkamp2 , F Bretz2 , E Glimm2 1 Medical University of Vienna, CeMSIIS, Vienna, Austria, 2 Novartis Pharma AG, Basel, Switzerland   The MCP-Mod approach from Bretz et al. (2005) has attracted attention in the recent years due to its potential to increase the efficiency of selecting the“right”dose. The testing part of MCP-Mod was originally developed to significant dose response signal conduct proof-of-concept (PoC) tests, i.e., to demonstrate that the dose response relationship of the test drug is not flat. But it is not appropriate to make a claim that the drug has a positive effect at some specific dose. In this presentation we extend the MCP-Mod approach by using the closed testing procedure from Marcus et. (1976) to obtain confirmatory p-values for dose response signal detection as well as for the pairwise comparisons of individual doses against placebo. The proposed test uses two-sided op- timal contrasts tests based on a-priori information about plausible dose response shapes available at the planning stage of a clinical trial. However, by using two-sided contrast tests only weak Type I error rate control can be achieved when testing superiority for individual doses. We show suit- able restrictions for the contrasts are needed to achieve strongType I error rate control. The operating characteristics of the proposed method will be evaluated for certain dose-response profiles.  

Pages Overview