A Wild Bootstrap approach for the selection of biomarkers in early diagnostic trials

Size: px
Start display at page:

Download "A Wild Bootstrap approach for the selection of biomarkers in early diagnostic trials"

Transcription

1 Zapf et al. BMC Medcal Research Methodology 25 5:43 DOI.86/s y RESEARCH ARTICLE Open Access A Wld Bootstrap approach for the selecton of bomarkers n early dagnostc trals Antona Zapf *, Edgar Brunner and Frank Konetschke 2 Abstract Background: In early dagnostc trals, partcularly n bomarker studes, the am often to select dagnostc tests among several methods. In case of metrc, dcrete, or even ordered categorcal data, the area under the recever operatng charactertc ROC curve denoted by AUC an approprate overall accuracy measure for the selecton, because the AUC ndependent of cut-off ponts. Methods: For selecton of bomarkers the ndvdual AUC s are compared wth a pre-defned threshold. To keep the overall coverage probablty or the multple type-i error rate, smultaneous confdence ntervals and multple contrast tests are consdered. We propose a purely nonparametrc approach for the estmaton of the AUC s wth the correspondng confdence ntervals and stattcal tests. Th approach uses the correlaton among the stattcs to account for multplcty. For small sample szes, a Wld-Bootstrap approach presented. It shown that the correspondng ntervals and tests are asymptotcally exact. Results: Extensve smulaton studes ndcate that the derved Wld-Bootstrap approach keeps and explots the nomnal type-i error at best, even for hgh accuraces and n case of small samples szes. The strength of the correlaton, the type of covarance structure, a skewed dtrbuton, and also a moderate mbalanced case-control rato do not have any mpact on the behavor of the approach. A real data set llustrates the applcaton of the proposed methods. Concluson: We recommend the new Wld Bootstrap approach for the selecton of bomarkers n early dagnostc trals, especally for hgh accuraces and small samples szes. Keywords: AUC, Dagnostc study, Resamplng, Smultaneous ntervals, Wld bootstrap Background The am of early dagnostc trals, partcularly of bomarker studes, often to select the most promng markers from a canddate set. For convenence, all dfferent knds of dagnostc tests, e.g., magng technques or bomarkers, wll be denoted by dagnostc tests throughout the paper. In these studes, response varables are often not bnary, but measured on a contnuous, dcrete or even ordnal scale and a cut-off value c has not yet been chosen. Therefore, the senstvty.e. true postve proporton and the specfcty true negatve proporton both beng computed based on c cannot be used as selecton crtera. In contrast, the Recever Operatng Charactertc ROC curve llustrates the overall dagnostc *Correspondence: Antona.Zapf@med.un-goettngen.de Department of Medcal Stattcs, Unversty Medcal Center Göttngen, Humboldtallee 32, 3773 Göttngen, Germany Full lt of author nformaton avalable at the end of the artcle performance because t ndependent of the chosen cutoff values see, e.g., DeLong, DeLong and Clark-Pearson []. Because the ROC curve of a dagnostc test nvarant wth respect to any monotone transformaton of the test measurement scale, t an adequate measure for comparng dagnostc tests beng measured even on dfferent scales. The Area Under the ROC-curve AUC represents an accuracy measure whch ndependent from the selected cut-off value c and whch nvarant under any monotone transformaton of the data. Therefore, t an approprate selecton crteron for promng dagnostc tests, and n partcular Xa et al. [2] p. 286 state n ther tutoral about translatonal bomarker dcovery n clncal metabolomcs that the AUC wdely used for performance comparon across dfferent bomarker models. As an example for the evaluaton of dfferent bomarkers we consder the ICM tral by Derchs et al. [3], whch ams 25 Zapf et al.; lcensee BoMed Central. Th an Open Access artcle dtrbuted under the terms of the Creatve Commons Attrbuton Lcense whch permts unrestrcted use, dtrbuton, and reproducton n any medum, provded the orgnal work properly credted. The Creatve Commons Publc Doman Dedcaton waver apples to the data made avalable n th artcle, unless otherwe stated.

2 Zapf et al. BMC Medcal Research Methodology 25 5:43 Page 2 of 3 to evaluate the dagnostc accuracy of ntestnal current measurement ICM wth regard to questonable cystc fbros CF. Th study was conducted wth the approval of the local ethcs commttee, MH Hannover, Germany and all patents and/or parents and healthy controls gave ther wrtten nformed consent. In th tral, a total of N = 67 chldren and adults were enrolled. The true dease state of the patents was defned by a composte gold standard, whch consts of typcal CF symptoms plus ether a postve sweat test and/or gene mutatons. By th defnton 26 patents were classfed nto CF referred to as cases and 4 nto CF unlkely referred to as controls. Furthermore, four bomarkers were consdered: I sc,carbachol, I sc,camp/forskoln,and I sc,htamne abbrevated by I carb, I camp,and I hta aswellasthesumof the three measured values, I sum. Boxplots of the data are dplayed n Fgure. In the ROC-curves n Fgure 2 the correspondng estmated AUC s are added. It can be readly seen that the dagnostc accuracy of I carb, I camp,and I hta qute good, and that I sum perfectly dfferentates the cases and the controls. Thus, the remanng queston whch bomarkers have suffcent dagnostc accuracy. There no consensus about the threshold for suffcent dagnostc accuracy. Xa et al. [2] characterze a bomarker wth an AUC <.7 as a qute weak bomarker. In ther study about a bloodbased bomarker panel for stratfyng current rk for colorectal cancer Marshall et al. [4] accept a canddate model wth an AUC >.75 as a predctve model. In contrast, Broadhurst and Kell [5] refer to an AUC >.9 as excellent and to an AUC >.8 as good. Dependng on prevous knowledge or expectatons a threshold for the AUC as ndcator for suffcent dagnostc accuracy should be chosen durng the plannng of the tral. Note that the am of such trals not to test multple hypotheses formulated n terms of AUC dfferences across the bomarkers, but to verfy suffcent dagnostc accuracy for all bomarkers ndvdually. Then comparng the lower lmt of the confdence nterval for the estmated AUC wth th threshold ndcates whether or not the dagnostc test has suffcent dagnostc accuracy. The Gudelne on the choce of the non-nferorty margn of the European Medcnes Agency [6] recommends to demonstrate non-nferorty by use of two-sded 95% or one-sded 97.5% confdence ntervals. Ifseveraldagnostctestsareevaluatednthesametral, t mportant to adjust the confdence ntervals for multplcty. Otherwe there a hgh rk that the accuracy of some dagnostc tests overestmated. Xa et al. [2] p.288 pont out that The probablty of fndng a random assocaton between a gven metabolte and the outcome ncreases wth the total number of comparons. Furthermore they note that the Bonferron correcton a smple but very conservatve method. If the dagnostc tests are repeatedly measured on the same subjects, hence, these Fgure Boxplots of the bomarkers. Boxplot of the four bomarkers n the example, separately for cases and controls.

3 Zapf et al. BMC Medcal Research Methodology 25 5:43 Page 3 of 3 Fgure 2 ROC-curves of the bomarkers. ROC-curves of the four bomarkers n the example and correspondng AUC s n the legend. measurements are correlated n general. Therefore t of hghly practcal mportance to take nto account these correlatons n the estmaton of the dagnostc accuracy. The multplcty expert group of the Stattcans n the Pharmaceutcal Industry [7] p.258 states that The partcpants dd, however, agree that for non-nferorty and equvalence trals, compatble smultaneous CIs for the prmary endponts should be presented n all cases.furthermore Strassburger and Bretz [8] recommend the use of sngle-step procedures f the am not to reject as many hypotheses as possble. Therefore we wll confne ourselves to smultaneous confdence ntervals from snglestep procedures whch are compatble wth the results obtaned by hypotheses tests. Among others, Hothorn et al. [9] proposed parametrc smultaneous confdence ntervals, whch correspond to multple contrast tests. However, snce these parametrc approaches are lmted to normally dtrbuted data, Konetschke et al. [] proposed nonparametrc multple contrast tests and compatble asymptotc smultaneous confdence ntervals for relatve treatment effects for ndependent samples based on some theoretcal results developed by Brunner et al. []. In the partcular case of two samples cases and controls the relatve treatment effect equvalent to the AUC see Bamber [2]. In th artcle we wll use th approach n the framework of dagnostc studes, but for pared samples n a multvarate layout. The challenge n early dagnostc trals often that smaller sample szes and hgher AUC s occur. For example nthesystematcrevewofstudesaboutthedagnostc accuracy of pleural flud NT-pro-BNP for pleural effusons of cardac orgn, performed by Janda and Swton [3], the medan total sample sze was 4 mean 2, and the pooled AUC was 98%. Wang et al. [4] reported n another systematc revew about cardac testng for coronary artery dease n potental kdney transplant recpents AUC s between.78 and.92. Kottas et al. [5] found that the Logt tranformaton based confdence nterval for a sngle AUC leads to slghtly conservatve results for small sample szes. Here we suggest Wld Bootstrap based smultaneous confdence ntervals to obtan robust methods for small sample szes and potentally qute large

4 Zapf et al. BMC Medcal Research Methodology 25 5:43 Page 4 of 3 AUC s. Hereby, we generalze the method proposed by Arlot et al. [6] for multvarte hgh-dmensonal normal data. In th artcle nonparametrc smultaneous confdence ntervals for multple AUC s n dagnostc studes are presented. Asymptotc ntervals wll be derved as well as ntervals usng the Wld Bootstrap approach. The propertes of these smultaneous ntervals are nvestgated n a smulaton study regardng the type-i error rate and the stattcal power. Furthermore, the results of all ntervals are gven for the example data set presented before n th secton. In the next secton we present the methods, ncludng the stattcal model wth the correspondng hypotheses, and the pont estmators wth ther asymptotc dtrbuton. Furthermore multple contrast tests and correspondng smultaneous confdence ntervals wth or wthout Logt transformaton are derved, and the Wld Bootstrap approach presented n partcular for small sample szes. The results of a smulaton study ncludng robustness evaluatons, and the applcaton of the methods to the example presented above are gven n the Results secton. Fnally, all results are summarzed and dcussed, and a recommendaton gven. Methods Stattcal model and hypotheses We consder a wthn-subject mult-modalty dagnostc tral gven by ndependent and dentcally dtrbuted random vectors X = X,..., X d F, =, control, case; subject s =,..., n, wth margnal dtrbutons X l F l, l =,..., d, 2 where d denotes the number of dagnostc tests. The partton of the data n cases = or controls = based on the gold or reference standard, whch assumed to represent the true dease status of the subjects. In order to allow for contnuous, dcrete or even ordered categorcal data n a unfed way, we use the normalzed verson of the margnal dtrbuton func- tons,.e., F l x = 2 F +,l x + F,l x, where F +,l x = P X l x denotes the rght-contnuous and F,l x = P X l < x denotes the left-contnuous verson of the dtrbuton functon respectvely. In the context of nonparametrc models, the normalzed verson of the dtrbuton functon was frst mentoned by Kruskal [7] and generally dates back to Lévy [8]. Later on, t was used by Ruymgaart [9], Munzel [2], Brunner and Pur [2], Kaufmann et al. [22], among others, to derve asymptotc results for rank stattcs ncludng the case of tes. We note that F l x may be arbtrary dtrbuton functons, wth the excepton of the trval case that both dtrbutons are one-pont dtrbutons see Lange and Brunner [23]. The wthn-subject desgn gven n, whch means that all dagnostc tests are performed n each ndvdual, recommended n the EMA gudelne about dagnostc agents [24] and refers to Desgn n Brunner and Zapf [25]. Foreachofthed dagnostc tests the true AUC gven by AUC l = PX l < Xl +.5 PXl = Xl = F l 3 dfl, l =,..., d. For a convenent dervaton of asymptotc results, the AUC sarecollectednthevectorauc = AUC,..., AUC d. In order to select the most promng dagnostc tests from the canddate set of the d dfferent methods, t our am to test the non-nferorty null hypotheses H : H : d l= d l= { H l : AUC l AUC } { H l : AUC l > AUC } versus wth strong control of the famlywe error rate FWER α smultaneously. The non-nferorty margn AUC assumed to have been fxed durng the plannng phase of the tral. Thus, the set of promng dagnostc tests consts of all markers, whose correspondng AUC l have been declared to be larger than AUC by an adequate multple testng procedure. Pont estmators and asymptotc dtrbuton Unbased and L 2 -constent pont estmators for the AUC s defned n 3 are derved by replacng the unknown dtrbuton functons F l and F l by ther emprcal counterparts F l x = n cx X l, =, ; l =,..., d, n s= where cx denotes the normalzed verson of the count functon,.e. cx {, 2,} correspondng to {x <, x =, x > }, respectvely. The pont estmator ÂUC l = F l l d F = R l. R l. + 5 N 2 can easly be computed usng the means R l. = n of the md- ranks R l, =,. Here, R l n s= Rl 4 denotes the rank of X l among all N = n +

5 Zapf et al. BMC Medcal Research Methodology 25 5:43 Page 5 of 3 n observatons X l,..., Xl n, X l l =,..., d. Further let R =,..., Xl n per marker,..., Rd R denote the vectors of the mdranks and let ÂUC = ÂUC,..., ÂUC d denote the vector of the pont estmators. Brunner et al. [] have shown that the vector NÂUC AUC follows, asymptotcally, as N, a multvarate normal dtrbuton wth expectaton and covarance matrx V N = Cov NB, 6 where B = B,..., B d denotes a random vector the components of whch are sums of ndependent random varables B l = n F l n Xl s n s= n s= F l Xl s + 2 AUCl. The covarance matrx V N wth elements v l,m,however, unknown and has to be estmated. Let R l denote the so-called nternal rank of X l among all n observatons X l,..., Xl n for the dagnostc test l n dease status group,andletr = R,..., R d denote the vectors of these nternal ranks. Furthermore, let Z = R R 8 N n denote the vectors of the normed placements F l Xl s = R l s R l s and n F l Xl s = R l s R l s, n respectvely. Then a constent estmator of the covarance matrx gven by V N = N V N, /n + V N, /n,where V N, = n n 7 Z Z. Z Z., =,. 9 s= Here, Z = n n s= Z denotes the vector of means of the normed placements. For more detals we refer to Brunner et al. [] and Kaufmann et al. [22]. Test stattcs and confdence ntervals In order to test the null hypotheses formulated n 4, we frst need to derve an unvarate test stattc for testng the ndvdual null hypothes H l : AUC l AUC. It follows from the asymptotc multvarate normalty of the vector NÂUC AUC that NÂUC l AUC l has, asymptotcally as N, a unvarate normal dtrbuton wth mean and varance v l,l,.e.n, v l,l. Here, v l,l denotes the l-th dagonal element of V N n 6. Hence, by Slutzky s theorem, t follows that T l = ÂUC l N AUC l v l,l D N,, asn, where v l,l denotes the dagonal elements of V N, defned n 9. In partcular, each stattc studentzed wth an ndvdual constent varance estmator and{ thus, the set of hypotheses } and test stattcs = H l, Tl, l =,..., d consttutes a jonttestng famly n the sense of Gabrel [26]. Attenton should be pad to the fact that the estmated varance v l,l equaltozerofâuc l = or.thus,theteststattc T l can not be computed. One possblty to solve th problem to modfy the data slghtly see the analys of the example n the Results secton. A qute conservatve selecton approach can be derved by applyng the Bonferron method denoted as Bonf,.e., the ndvdual null hypothes H l : AUC l AUC wll be rejected at multple level α, ft l z α/d,, where z α/d, denotes the one-sded α/d-quantle of the standard normal dtrbuton. Asymptotc onesded smultaneous confdence ntervals for the treatment effects AUC l are then gven by CI l Bonf = [ÂUC l z α/d, v l,l N ; ]. The global null hypothes H : AUC AUC as defned n 4 wll be rejected, f max{t,..., T d } > z α/d, or, equvalently, f the maxmum of the lower lmts of the confdence ntervals max{ci Bonf,l,..., CId Bonf,l } > AUC.Here =,..., denotes a d-dmensonal vector of s. The Bonferron method, however, a qute conservatve selecton approach see Results secton for more detals. The reason for th that the apparent correlatons among the dfferent pvotal quantttes T,..., T d are not taken nto account by th method. Multple contrast tests and smultaneous confdence ntervals In order to use the correlaton n the selecton approach, t our dea to apply the multple contrast test prncple denoted by MCP, whch uses the correlaton among dfferent test stattcs. The key pont of these procedures to use the jont dtrbuton of a set of stattcs to adjust for multplcty. Thus, the asymptotc multvarate dtrbuton of the vector T = T,..., T d requred. The detals are stated n the next theorem. Theorem. Under the assumpton that N such that N/n N <, =,, the vector T follows, asymptotcally, a multvarate normal dtrbuton

6 Zapf et al. BMC Medcal Research Methodology 25 5:43 Page 6 of 3 wth expectaton and correlaton matrx R, wherer = [ r l,m ] l,m=...,d,andr l,m v = l,m. v l,l v m,m The jont dtrbuton of T can be used for the dervaton of a smultaneous test procedure. Let z α, R denote the one-sded α equcoordnate quantle of the multvarate normal dtrbuton wth expectaton and correlaton matrx R,.e., N, R,that d P l= { T l z α, R} = α. For detals see Bretz et al. [27]. Then, the ndvdual null hypothes H l AUCl AUC wll be rejected at multple level α,f T l z α, R. 2 Asymptotc one-sded smultaneous confdence ntervals for AUC l are gven by ] CI l MCP [ÂUC = l v z α, R l,l N ;. 3 The global null hypothes wll be rejected f max{t,..., T d } > z α, R or f max{ci MCP,l,..., CI d MCP,l } > AUC. The correlaton matrx R, however, unknown and must be replaced by a constent estmator R. We propose to replace R by R n the consderatons above, where R =[ r l,m ] l,m=,...,d and r l,m = respectvely. v vl,m l,l v, m,m Smulaton studes ndcate, however, that the speed of convergence of T to a multvarate normal dtrbuton qute slow, partcularly when smaller sample szes and larger numbers of dagnostc tests are consdered. In a varety of applcatons, see e.g. Zou and Yue [28] or Konetschkeetal.[],tturnsoutthattheuseofadequate transformatons e.g., the Logt-transformaton tend to ncrease the speed of convergence. Therefore, smultaneous confdence ntervals wth Logt transformaton wll be derved n the next secton. Multple contrast tests and smultaneous confdence ntervals wth Logt transformaton To derve smultaneous Logt-transformed confdence ntervals let gauc = gauc,..., gauc d :, d R d, denote the vector of Logt-transformed AUC s, where gauc l = log AUC l AUC l Furthermore, let = dag AUC AUC,..., AUC d AUC d. denote the dagonal Jacoban matrx of gauc.underthe addtonal assumpton that N such that N/n f, t follows from Cramer s multvarate δ-theorem see, e.g., Ferguson [29], Theorem 7.4 that N g ÂUC g AUC D N, SN 4 where S N = V N and V N gvenn6.toestmate the asymptotc covarance matrx S N,let = dag ÂUC ÂUC,..., ÂUC d ÂUC d denote the estmated Jacoban matrx of gauc and note that the estmator Ŝ N = V N a constent estmator of S N. Agan there a problem f ÂUC l = or.here, and n turn Ŝ N cannot be calculated. Th problem addressed n the analys of the example n the Results secton. To test the ndvdual hypothes H l : AUC l AUC defne the pvotal quanttes T l = gâuc l N gauc l D N,, ŝ l,l N, l =,..., d, 5 where ŝ l,l denotes the l-th dagonal element of Ŝ N.The jont dtrbuton of the vector T = T,..., T d gvennthenexttheorem. Theorem 2. If N such that N/n f <, then the vector T= T,..., T d follows, asymptotcally, a multvarate normal dtrbuton wth expectaton and correlaton matrx R,where R gvenntheorem. It follows from Theorem 2 that both the vectors T and T have, asymptotcally, as N, the same jont dtrbuton. Both the correlaton matrces of T and T asymptotcally concde due to the dagonal structure of. Now, a smultaneous test procedure, whch takes the correlaton nto account can be derved. The ndvdual null hypothes H l : AUC l AUC wll be rejected at multple level α,f T l z α, R, 6 where z α, R denotes the one-sded equcoordnate quantle of the correspondng multvarate normal dtrbuton where the correlaton matrx R replaced wth the constent estmator R. One-sded smultaneous confdence ntervals for AUC l are then gven by CI l Logt = expt g ÂUC l z α, R ŝ l,l N,, 7

7 Zapf et al. BMC Medcal Research Methodology 25 5:43 Page 7 of 3 where expty = expy +expy denotes the nverse Logttransformaton. The global null hypothes H : AUC AUC wll be rejected, f max { T,..., T d} z α, R, orfmax{ci Logt,l,..., CId Logt,l } > AUC.Snce the Logt-functon monotone, the procedure asymptotcally controls the famlywe error rate n the strong sense [26]. Small sample approxmatons wth Wld Bootstrap In the prevous secton approaches for the selecton of dagnostc tests based on the AUC s have been derved. The procedures are based on the asymptotc jont dtrbuton of the vectors T or T, respectvely. The proposed approaches for selecton of dagnostc tests are vald for large sample szes. In order to nvestgate the accuraces of the procedures n terms of controllng the pre-assgned type-i error level under the null hypothes, mantanng the nomnal coverage probablty of the correspondng smultaneous confdence ntervals, and ther powers to detect certan alternatves, extensve smulaton studes were conducted. These smulaton studes ndcate, however, that both the stattcs T n 2 and T n 5 tend to result n lberal or conservatve decons n case of smaller sample szes N and larger AUC AUC.8. The results are n concordance wth the smulaton results proposed for unvarate stattcs by Kottas et al. [5] or Qn and Hotlovac [3]. Therefore, we propose a Wld Bootstrap approach to approxmate ther samplng dtrbutons for small sample szes. Resamplng procedures are wdely known to be qute robust methods, even for small sample szes. However, permutaton methods cannot be used n th setup, snce the dtrbutons of the test stattcs and the resamplng stattcs do not concde, not even asymptotcally Pauly M, Asendorf T, Konetschke F: Permutaton tests and confdence ntervals for the area under the ROC curve, submtted. Smulaton studes ndcate that the use of the conventonal Bootstrap from Efron [3] results n lberal conclusons, partcularly when confronted wth an AUC.7 see Table. Therefore, we dd not further nvestgate the conventonal Bootstrap. In contrast, the Wld Bootstrap approach ensures that the resamplng dtrbuton of the stattcs mmcs the dtrbuton of T Table Emprcal type-i error theoretcal 2.5%of the normal Bootstrap for d = 5 and N = 5 wth varyng case-control-rato and varyng AUC ccr AUC :.68% 2.28% 3.% 5.% 7.8% : 4.9% 2.96% 4.7% 6.4% 2.% and T, asymptotcally. The Wld Bootstrap technque motvated by the resdual bootstrap commonly appled n regresson analys [32-35], and n tme-seres testng problems [36-38]. It also proposed n the context of survval analys [39-42], and wll be explaned n the followng. Let W,..., W n, W,..., W n 8 denote ndependent and dentcally dtrbuted random weghts wth EW = andvarw =, whch are ndependent of the data. We wll nvestgate three dfferent knds of random weghts W n our extensve smulaton study: Rademacher weghts: PW = = PW = = 2. Standard normal weghts: W,..., W n N,. Unform weghts: W,..., W n U Let Z = W Z Z = W Z Z =,, s =,..., n,,..., W Z d [ 2 2, 2 2 Z d, ]. 9 denote N resamplng vectors, where Z gven n 8. Furthermore, let Z = n n k= Z = Z,..., Z d denote ther means and let v l,l = n n s= Z l Z l 2 denote the emprcal varance of Z l,..., Z l n, l =,..., d. In the next theorem t wll be shown that the condtonal resamplng dtrbuton of the vector T = T l = N T,..., T d,where Z l Z l, v l,l /n + v l,l /n 2 mmcs the dtrbuton of both the vectors T and T, asymptotcally. Theorem 3. If N such that N n converges to some fnte constant f, then the condtonal dtrbuton of T gven the data X converges n probablty to the multvarate normal dtrbuton wth expectaton and correlaton matrx R.

8 Zapf et al. BMC Medcal Research Methodology 25 5:43 Page 8 of 3 For proof see Addtonal fle. Note that Theorem 3 vald under the null as well as under the alternatve,.e., the resamplng dtrbuton mmcs the dtrbutons of T and T for arbtrary values of AUC = AUC,..., AUC d. Next we wll explan the computaton of the smultaneous confdence ntervals:. Gven the data X, compute the pont estmators ÂUC and V N as gven n 5 and 9, respectvely. 2. Generate N = n + n random weghts W,..., W n as descrbed n 8 3. Compute A j := max{t,..., T d } as gven n Repeat the steps nboot tmes e.g. nboot =, and obtan the values A,..., A nboot. 5a. Compare each A j wth max { T }. Then the ndvdual p-value for H l : AUC l AUC obtaned from nboot nboot j= I{ T l A j },wherei{ } denotes the ndcator functon. 5b. Estmate the quantle z α, R by the one-sded α-quantle z α, of A,..., A nboot to obtan the one-sded α smultaneous confdence ntervals gven by CI l WB = expt g ÂUC l z α, ŝ l,l N,. 2 Results Smulaton results We performed a smulaton study to nvestgate the propertes of the dfferent approaches. All smulatons were conducted wth R envronment, verson R Development Core Team, 2, each wth 5, smulaton runs and 5, bootstrap repettons. The nomnal type- I error was set to 2.5% one-sded and the global null hypothes accordng to 4 was rejected, f at least one of the one-sded p-values was smaller than α = 2.5%. Th means, the famly we error rate n the strong sense FWER controlled, and the one-sded emprcal type- I error should be closed to 2.5%. It also possble to use the correspondng confdence ntervals for decon. Then the global null hypothes rejected f the lower lmt of at least one confdence nterval was above AUC. We generated multvarate normally dtrbuted random vectors wth compound symmetrc correlaton structure and defned the followng scenaro as standard scenaro: a total sample sze N = wth a case-control rato ccr of :, d = 5 dagnostc tests and a correlaton of ρ =.9 between the tests motvated by [2,3,24]; and the example data set. The dfferent parameters and condtons were vared afterwards as follows: The true AUC.5,...,.9 The number of dagnostc tests d 5,, 2 The total sample sze N 5,, 2 The case-control rato ccr :, :2, :4, :9 The true correlaton between the dagnostc tests ρ.3,.6,.9 The covarance structure n the data compound symmetry, unstructured, and dagonal matrx wth heterogeneous varances and postve or negatve parng The dtrbuton of the data normal, skewed = log-normal, ordnal The dfferent parameter constellatons and all smulaton results can be seen n the Addtonal fle 2. Due to computatonal complexty, and ts weak behavor n standard stuatons, we dd not further nvestgate the conventonal Bootstrap n our smulaton study. In a frst step, th standard scenaro was used for the comparon of the three random weghts for the Wld Bootstrap: Rademacher WB-Rade, standard normal WB-Normal and unform WB-Unf weghts. The results are dplayed n the Addtonal fle 3. For an AUC of.5 the three weghts lead to nearly the same emprcal type-i error and are qute conservatve emprcal α.5. For larger AUC s the results are less conservatve and for AUC s above.8 the emprcal type-i error around 2.5%. The Wld Bootstrap approach wth unform weghts, however, more conservatve, whle the standard normal and the Rademacher weghts lead nearly to the same results. Therefore, and to present the smulaton results more clearly, we only consder the standard normal weghts n the followng. The smulaton results for the other weghts are provded n the Addtonal fle 2. In practce often unadjusted wth the local type-i error α equal to the global type-i error α or Bonferron adjusted confdence ntervals for the sngle AUC s are used see for example Shotan et al. [43]. Therefore, n a second step, we compared these approaches agan for the standard scenaro usng the multple contrast test MCP, the smultaneous Logt Logt and the Wld Bootstrap WB-Normal approach. In Fgure 3 t becomes apparent that unadjusted ntervals Unadj lead to hghly lberal conclusons emprcal type-i error 8 9%, whle the Bonferron correcton Bonf too conservatve..5%. Therefore we wll not consder these approaches n the sequel. The MCP approach keeps the type-i error for an AUC of.5, but becomes more and more lberal for larger AUC s up to 4% for AUC =.9. The emprcal type I error of the Logt and the WB- Normal approach comparable and between.5% and 2.9%. In the followng we wll nvestgate the nfluence of

9 Zapf et al. BMC Medcal Research Methodology 25 5:43 Page 9 of 3 Fgure 3 Emprcal type-i error for varyng AUC s. Emprcal type-i error of the dfferent approaches for the standard scenaro see text wth varyng AUC s. the dfferent parameter settngs on the type-i error of the Logt and the WB-Normal approach, and also of the MCP approach as the bas of both the approaches despte of ts lberal behavor. The strength of the correlaton, the type of the covarance structure and a skewed dtrbuton do not have any mpact on the behavor of the test see fgures and tables n the Addtonal fles 2, 4, 5, and 6. The mpact of the sample sze N and the number of dagnostc tests d shown n Fgure 4. As expected, for a larger sample sze and a small number of dagnostc tests the type-i error better exploted. As already seen n Fgure 3 the Logt and the WB-Normal approach are comparable f AUC.8 ndependent of N and d. For larger AUC s, the WB-Normal approach leads to a larger emprcal type-i error. On the one hand, th means that α better exploted, on the other hand, th means that the results are lberal. The emprcal type-i error of the Logt approach for AUC =.9 ranges from.3% to 2.%, and of the WB-Normal approach from 2.2% to 2.9%. If the case-control rato ccr not balanced, the emprcal type-i error ncreases wth ncreasng mbalance see Fgure 5. For an AUC of.8 or smaller both approaches are robust to an mbalance up to : 4. For AUC =.9 the lberalty of the WB-Normal approach a dadvantage here, the emprcal type-i error above 2.5%. For a casecontrol rato of : 9, both approaches are far too lberal. Ordnal data was generated usng dcreted normal dtrbutons wth a gven AUC. For th data, representng a 5-pont gradng scale, the emprcal type-i error decreases wth ncreasng AUC AUC =.5: Logt = 2.3%, WB-Normal = 2.2% to AUC =.9: Logt =.7%, WB-Normal =.6%. For detals see Addtonal fle 2. The power was calculated for one example scenaro N = 2, d = 5, ccr = :,ρ =.9, AUC =.7, where the emprcal type-i error of the Logt and of the WB-Normal approach was nearly the same. The true AUC ncreasng from.7 whch equal to AUC to.85, accordng AUC =,...,.5. The power of the two approaches bascally the same. For an AUC of..e. AUC =.8 vs. AUC =.7 the power greater than 8% see Addtonal fle 2. Results for the analys of the example The pont estmators for the AUC s are presented n the Background secton n Fgure 2. The number of 26 cases and 4 controls correspond to a case-control rato of :.6. The Spearman correlaton coeffcents between the bomarkers range from.64 to.95. For I sum the result was AUC=. Because logt =, we modfed the

10 Zapf et al. BMC Medcal Research Methodology 25 5:43 Page of 3 Fgure 4 Emprcal type-i error for varyng N and d. Emprcal type-i error of the MCP, the Logt and the WB-Normal approach for varyng sample sze and number of dagnostc tests. Fgure 5 Emprcal type-i error for varyng ccr. Emprcal type-i error of the MCP, the Logt and the WB-Normal approach for varyng case-control ratos. data for I sum such that we replaced the largest measurement of the controls wth the smallest measurement of the cases. Th mnmal change leads to a pont estmator for the AUC of.9999, and enables us to calculate the confdence ntervals. Th replacement strategy conservatve, snce the effect decreased, and the varance ncreased. The one-sded 97.5% confdence ntervals for all bomarkers usng the MCP, the Logt, and the Wld Bootstrap approach are dplayed n Fgure 6. The results of the Wld Bootstrap wth the three dfferent weghts dffered just n the thrd decmal place. For constency we dplayed the WB-Normal approach here. The pattern of the results the same for all four bomarkers. Accordng to the smulaton results, the MCP ntervals are the shortest, the Logt ntervals are the broadest, and the WB ntervals are n between. In the artcle of Derchs et al. [3] no threshold defned. In Fgure 6 four possble thresholds.8,.85,.9,.95 are marked by sold horzontal lnes. In Table 2 for each of these thresholds the numbers of selected bomarkers, dependng on the ndvdual approach, are lted. Apparently, the Logt approach a more conservatve selecton crteron than the Wld Bootstrap approach. Although the MCP ntervals are clearly shorter than the Wld Bootstrap ntervals, the number of selected bomarkers the same

11 Zapf et al. BMC Medcal Research Methodology 25 5:43 Page of 3 Fgure 6 Confdence ntervals for the bomarkers. One-sded 97.5% confdence ntervals for the four bomarkers usng the MCP, the Logt, and the WB-Normal approach. forthemcpandthewbapproachforthreethresholds. Only for the threshold of.85 the MCP approach would select one bomarker more. Consderng the smulaton results of th secton we would recommend to use the WB-Normal approach. Dcusson It wdely dcussed n the lterature, whether the type- I error should be adjusted for multplcty and whether the Bonferron correcton an approprate approach. Among many others, Wttes [44] states that lack of adjustment can lead to a mnterpretaton of the study results as well as Bonferron adjustment can do. Furthermore Table 2 Number of selected bomarkers of the MCP, the Logt, and the WB-Normal approach for dfferent thresholds based on one-sded 97.5% confdence ntervals Threshold MCP Logt WB-Normal Perneger [45] states that In summary, Bonferron adjustments have, at best, lmted applcatons n bomedcal research, and should not be used when assessng evdence about specfc hypotheses. Nevertheless, n practce often Bonferron adjusted or even unadjusted confdence ntervals for the sngle AUC s are used see for example [43]. Konetschke et al. [] proposed nonparametrc multple contrast tests and smultaneous confdence ntervals for adequate correcton of the type-i error, whch take the dependences wthn the data nto account. Furthermore the authors recommended the transformaton method for example the Logt-transformaton to get less lberal results. However, Qn and Hotlovac [3] notced that the Logt-transformed ntervals are conservatve for hgh accuraces. The reason that the estmator logtâuc qute unstable f ÂUC close to or because of a possbly larger varance. Obuchowsk and Leber [46] compared dfferent confdence ntervals for the AUC and concluded that for small sample szes none of them provdes adequate coverage for hgh accuraces. Concluson InthartclewedervedaWldBootstrapapproach, whch explots the type-i error much better than the Logt-approach, even for hgh accuraces and small

12 Zapf et al. BMC Medcal Research Methodology 25 5:43 Page 2 of 3 samples. Nether the strength of correlaton, nor the structure of the covarance matrx, nor a skewed dtrbuton, nor a moderate mbalanced case-control rato has any mpact on th desrable property of the Wld Bootstrap approach. Correspondng to these results we recommend to use the Wld Bootstrap approach wth standard normally dtrbuted weghts for the selecton of bomarkers n early dagnostc trals wth the AUC as selecton crteron. Addtonal fles Addtonal fle : Proof of Theorem 3. Addtonal fle 2: Tables of smulaton results. Addtonal fle 3: Fgure S. Emprcal type-i error of the Wld Bootstrap approach wth the three dfferent weghts for the standard scenaro see artcle, Secton Smulaton results wth varyng AUC s. Addtonal fle 4: Fgure S2. Emprcal type-i error of the MCP, the Logt and the WB-Normal approach for varyng strength of correlaton. Addtonal fle 5: Fgure S3. Emprcal type-i error of the MCP, the Logt and the WB-Normal approach for dfferent covarance structures CS: compound symmetry, UN: unstructured, PP/NP: dagonal matrx wth heterogeneous varances and postve/negatve parng. Addtonal fle 6: Fgure S4. Emprcal type-i error of the MCP, the Logt and the WB-Normal approach for normal and log-normal dtrbuted data. Competng nterests The authors declare that they have no competng nterests. Authors contrbutons AZ, FK, and EB derved the Wld Bootstrap approach. AZ and FK performed the smulaton study and wrote the artcle. EB reved the manuscrpt. All authors read and approved the fnal manuscrpt. Acknowledgements Th work was supported by the Federal Mntry of Educaton and Research [5MMGB]. The authors thank Prof. Ballmann from the DRK-Knderklnk Segen for supportng th work by provdng study data. Author detals Department of Medcal Stattcs, Unversty Medcal Center Göttngen, Humboldtallee 32, 3773 Göttngen, Germany. 2 Department of Mathematcal Scences, The Unversty of Texas at Dallas, 8 W Campbell Road, 758 Rchardson, TX, USA. Receved: 5 October 24 Accepted: 25 March 25 References. DeLong E, DeLong D, Clark-Pearson D. Comparng the areas under two or more correlated recever operatng charactertc curves: a nonparametrc approach. Bometrcs. 988;44: Xa J, Broadhurst D, Wlson M, Whart D. Translatonal bomarker dcovery n clncal metabolomcs: an ntroductory tutoral. Metabolomcs. 23;9: Derchs N, Sanz J, Von Kanel T, Stolpe C, Zapf A, Tümmler B, et al. Intestnal current measurement for dagnostc classfcaton of patents wth quastonable cystc fbros: valdaton and reference data. Thorax. 2;65: Marshall K, Mohr S, Khettab F, Nossova N, Chao S, Bao W, et al. A blood-based bomarker panel for stratfyng current rk for colorectal cancer. Int J Cancer. 2;26: Broadhurst D, Kell D. Stattcal strateges for avodng false dcoveres n metabolomcs and related experments. Metabolomcs. 26;2: EMA. Gudelne on the choce of the non-nferorty margn. Doc. Ref. EMEA/CPMP/EWP/258/ ncludes/document/open_document.jsp?webcontentid=wc53636 date of last access 3/4/5. 7. Phllps A, Fletcher C, Atknson G, Channon E, Dour A, Jak T, et al. Multplcty: dcusson ponts from the stattcans n the pharmaceutcal ndustry multplcty expert group. Pharm Stat. 23;2: Strassburger K, Bretz F. Compatble smultaneous confdence bounds for the Holm procedure and other Bonferron-based closed tests. Stat Med. 28;27: Hothorn T, Bretz F, Westfall P. Smultaneous nference n general parametrc models. Bometrcal J. 28;5: Konetschke F, Hothorn L, Brunner E. Rank-based multple test procedures and smultaneous confdence ntervals. Electron J Stat. 22;6: Brunner E, Munzel U, Pur M. The multvarate nonparametrc Behrens-Fher problem. J Stat Plannng Inference. 22;8: Bamber D. The area above the ordnal domnance graph and the area below recever operatng charactertc graph. J Math Psychol. 975;2: Janda S, Swton J. Dagnostc accuracy of pleural flud NT-pro-BNP for pleural effusons of cardac orgn: a systematc revew and meta-analys. BMC Pulmonary Med. 2;: Wang L, Fahm M, Hayen A, Mtchell R, Banes L, Lord S. Cardac testng for coronary artery dease n potental kdney transplant recpents. Cochrane Database Syst Rev. 2;2:. DOI:.2/ CD869.pub2. 5. Kottas M, Kuss O, Zapf A. A modfed Wald nterval for the area under the ROC curve AUC n dagnostc case-control studes. BMC Med Res Methodology. 24;4: Arlot S, Blanchard G, Roquan E. Some nonasymptotc results on resamplng n hgh dmenson, I: confdence regons. Ann Stat. 2;38: Kruskal W. A nonparametrc test for the several sample problem. Ann Math Stat. 952;23: Lévy P. Calcul des Probabltées. Par: Gauthers-Vllars, Édteurs; Ruymgaart F. A unfed approach to the asymptotc dtrbuton theory of certan mdrank stattcs In: Raoult JP, edtor. Stattque Non Parametrque Asymptotque vol. Lecture Notes on Mathematcs, No. 82. Sprnger, Berln Hedelberg; 98. p Munzel U. Lnear rank score stattcs when tes are present. Stat Probablty Lett. 999;4: Brunner E, Pur M. Nonparametrc methods n factoral desgns. Stat Pap. 2;42: Kaufmann J, Werner C, Brunner E. Nonparametrc methods for analysng the accuracy of dagnostc tests wth multple readers. Stat Methods Med Res. 25;4: Lange K, Brunner E. Senstvty, specfcty and ROC-curves n multple reader dagnostc trals - a unfed, nonparametrc approach. Stat Methodology. 22;9: EMA. Gudelne on clncal evaluaton of dagnostc agents. Doc. Ref. CPMP/EWP/9/98/Rev ncludes/document/open_document.jsp?webcontentid=wc5358 date of last access 3/4/ Brunner E, Zapf A. Nonparametrc ROC analys for dagnostc trals In: Balkrhnan N, edtor. Methods and Applcatons of Stattcs n Clncal Trals vol. Volume 2: Plannng, Analys, and Inferental Methods. Hoboken, New Jersey: John Wley & Sons; 24. p Gabrel K. Smultaneous test procedures - some theory of multple comparons. Ann Math Stat. 969;4: Bretz F, Landgrebe J, Brunner E. Multplcty sues n mcroarray experments. Methods Inf Med. 25;44: Zou G, Yue L. Usng confdence ntervals to compare several correlated areas under the recever operatng charactertc curves. Stat Med. 22;32: Ferguson T. A Course n Large Sample Theory. London: Chapman & Hall; Qn G, Hotlovac L. Comparon of non-parametrc confdence nterval for the area under the ROC curve of a contnuous-scale dagnostc test. Stat Methods Med Res. 28;7: Efron B. Bootstrap methods: Another look at the Jackknfe. Ann Stat. 979;7: 26.

13 Zapf et al. BMC Medcal Research Methodology 25 5:43 Page 3 of Wu C. Jackknfe, Bootstrap and other resamplng methods n regresson analys. Ann Stat. 986;4: Mammen E. When does Bootstrap work? Asymptotc results and smulatons. New York: Sprnger; Beran R. Dagnosng Bootstrap success. Ann Inst Stat Mathematcs. 997;49: Janssen A. Nonparametrc symmetry tests for stattcal functonals. Math Methods Stat. 999;8: Kres J, Paparodt E. Bootstrap for dependent data: a revew, wth dcusson, and a rejonder. J Korean Stat Soc. 2;4: Kres J, Paparodt E. Bootstrappng locally statonary processes. J R Stat Soc - Ser B. 24;77: Konetschke F, Pauly M. Bootstrappng and permutng pared t-test type stattcs. Stat Comput. 24;24: Ln D. Non-parametrc nference for cumulatve ncdence functons n competng rks studes. Stat Med. 997;6:9. 4. Beyersmann J, d Termn S, Pauly M. Weak convergence of the Wld Bootstrap for the Aalen-Johansen estmator of the cumulatve ncdence functon of a competng rk. Scand J Stat. 24;4: Pauly M. Weghted resamplng of martngale dfference arrays wth applcatons. Electron J Stat. 2;5: Dobler D, Pauly M. How to Bootstrap Aalen-Johansen processes for competng rks? Handcaps, solutons, lmtatons. Electron J Stat. 24;8: Shotan A, Murao T, Kmura Y, Matsumoto H, Kamada T, Kusunok H, et al. Identfcaton of serum mrnas as novel non-nvasve bomarkers for detecton of hgh rk for early gastrc cancer. Br J Cancer. 23;9: Wttes J. Clncal trals must cope better wth multplcty. Nat Med. 22;8: Perneger T. What s wrong wth Bonferron adjustments. Br Med J. 998;36: Obuchowsk N, Leber M. Confdence ntervals for the recever operatng charactertc area n studes wth small samples. Academc Radology. 998;5:56 7. Submt your next manuscrpt to BoMed Central and take full advantage of: Convenent onlne submson Thorough peer revew No space constrants or color fgure charges Immedate publcaton on acceptance Incluson n PubMed, CAS, Scopus and Google Scholar Research whch freely avalable for redtrbuton Submt your manuscrpt at

Parameter Estimates of a Random Regression Test Day Model for First Three Lactation Somatic Cell Scores

Parameter Estimates of a Random Regression Test Day Model for First Three Lactation Somatic Cell Scores Parameter Estmates of a Random Regresson Test Day Model for Frst Three actaton Somatc Cell Scores Z. u, F. Renhardt and R. Reents Unted Datasystems for Anmal Producton (VIT), Hedeweg 1, D-27280 Verden,

More information

Joint Modelling Approaches in diabetes research. Francisco Gude Clinical Epidemiology Unit, Hospital Clínico Universitario de Santiago

Joint Modelling Approaches in diabetes research. Francisco Gude Clinical Epidemiology Unit, Hospital Clínico Universitario de Santiago Jont Modellng Approaches n dabetes research Clncal Epdemology Unt, Hosptal Clínco Unverstaro de Santago Outlne 1 Dabetes 2 Our research 3 Some applcatons Dabetes melltus Is a serous lfe-long health condton

More information

Using the Perpendicular Distance to the Nearest Fracture as a Proxy for Conventional Fracture Spacing Measures

Using the Perpendicular Distance to the Nearest Fracture as a Proxy for Conventional Fracture Spacing Measures Usng the Perpendcular Dstance to the Nearest Fracture as a Proxy for Conventonal Fracture Spacng Measures Erc B. Nven and Clayton V. Deutsch Dscrete fracture network smulaton ams to reproduce dstrbutons

More information

Resampling Methods for the Area Under the ROC Curve

Resampling Methods for the Area Under the ROC Curve Resamplng ethods for the Area Under the ROC Curve Andry I. Bandos AB6@PITT.EDU Howard E. Rockette HERBST@PITT.EDU Department of Bostatstcs, Graduate School of Publc Health, Unversty of Pttsburgh, Pttsburgh,

More information

Copy Number Variation Methods and Data

Copy Number Variation Methods and Data Copy Number Varaton Methods and Data Copy number varaton (CNV) Reference Sequence ACCTGCAATGAT TAAGCCCGGG TTGCAACGTTAGGCA Populaton ACCTGCAATGAT TAAGCCCGGG TTGCAACGTTAGGCA ACCTGCAATGAT TTGCAACGTTAGGCA

More information

Modeling Multi Layer Feed-forward Neural. Network Model on the Influence of Hypertension. and Diabetes Mellitus on Family History of

Modeling Multi Layer Feed-forward Neural. Network Model on the Influence of Hypertension. and Diabetes Mellitus on Family History of Appled Mathematcal Scences, Vol. 7, 2013, no. 41, 2047-2053 HIKARI Ltd, www.m-hkar.com Modelng Mult Layer Feed-forward Neural Network Model on the Influence of Hypertenson and Dabetes Melltus on Famly

More information

Modeling the Survival of Retrospective Clinical Data from Prostate Cancer Patients in Komfo Anokye Teaching Hospital, Ghana

Modeling the Survival of Retrospective Clinical Data from Prostate Cancer Patients in Komfo Anokye Teaching Hospital, Ghana Internatonal Journal of Appled Scence and Technology Vol. 5, No. 6; December 2015 Modelng the Survval of Retrospectve Clncal Data from Prostate Cancer Patents n Komfo Anokye Teachng Hosptal, Ghana Asedu-Addo,

More information

International Journal of Emerging Technologies in Computational and Applied Sciences (IJETCAS)

International Journal of Emerging Technologies in Computational and Applied Sciences (IJETCAS) Internatonal Assocaton of Scentfc Innovaton and Research (IASIR (An Assocaton Unfyng the Scences, Engneerng, and Appled Research Internatonal Journal of Emergng Technologes n Computatonal and Appled Scences

More information

Estimation of Relative Survival Based on Cancer Registry Data

Estimation of Relative Survival Based on Cancer Registry Data Revew of Bonformatcs and Bometrcs (RBB) Volume 2 Issue 4, December 203 www.sepub.org/rbb Estmaton of Relatve Based on Cancer Regstry Data Olaf Schoffer *, Ante Nedostate 2, Stefane J. Klug,2 Cancer Epdemology,

More information

Appendix for. Institutions and Behavior: Experimental Evidence on the Effects of Democracy

Appendix for. Institutions and Behavior: Experimental Evidence on the Effects of Democracy Appendx for Insttutons and Behavor: Expermental Evdence on the Effects of Democrac 1. Instructons 1.1 Orgnal sessons Welcome You are about to partcpate n a stud on decson-makng, and ou wll be pad for our

More information

ALMALAUREA WORKING PAPERS no. 9

ALMALAUREA WORKING PAPERS no. 9 Snce 1994 Inter-Unversty Consortum Connectng Unverstes, the Labour Market and Professonals AlmaLaurea Workng Papers ISSN 2239-9453 ALMALAUREA WORKING PAPERS no. 9 September 211 Propensty Score Methods

More information

Optimal Planning of Charging Station for Phased Electric Vehicle *

Optimal Planning of Charging Station for Phased Electric Vehicle * Energy and Power Engneerng, 2013, 5, 1393-1397 do:10.4236/epe.2013.54b264 Publshed Onlne July 2013 (http://www.scrp.org/ournal/epe) Optmal Plannng of Chargng Staton for Phased Electrc Vehcle * Yang Gao,

More information

Physical Model for the Evolution of the Genetic Code

Physical Model for the Evolution of the Genetic Code Physcal Model for the Evoluton of the Genetc Code Tatsuro Yamashta Osamu Narkyo Department of Physcs, Kyushu Unversty, Fukuoka 8-856, Japan Abstract We propose a physcal model to descrbe the mechansms

More information

THE NATURAL HISTORY AND THE EFFECT OF PIVMECILLINAM IN LOWER URINARY TRACT INFECTION.

THE NATURAL HISTORY AND THE EFFECT OF PIVMECILLINAM IN LOWER URINARY TRACT INFECTION. MET9401 SE 10May 2000 Page 13 of 154 2 SYNOPSS MET9401 SE THE NATURAL HSTORY AND THE EFFECT OF PVMECLLNAM N LOWER URNARY TRACT NFECTON. L A study of the natural hstory and the treatment effect wth pvmecllnam

More information

The Limits of Individual Identification from Sample Allele Frequencies: Theory and Statistical Analysis

The Limits of Individual Identification from Sample Allele Frequencies: Theory and Statistical Analysis The Lmts of Indvdual Identfcaton from Sample Allele Frequences: Theory and Statstcal Analyss Peter M. Vsscher 1 *, Wllam G. Hll 2 1 Queensland Insttute of Medcal Research, Brsbane, Australa, 2 Insttute

More information

Study and Comparison of Various Techniques of Image Edge Detection

Study and Comparison of Various Techniques of Image Edge Detection Gureet Sngh et al Int. Journal of Engneerng Research Applcatons RESEARCH ARTICLE OPEN ACCESS Study Comparson of Varous Technques of Image Edge Detecton Gureet Sngh*, Er. Harnder sngh** *(Department of

More information

Using Past Queries for Resource Selection in Distributed Information Retrieval

Using Past Queries for Resource Selection in Distributed Information Retrieval Purdue Unversty Purdue e-pubs Department of Computer Scence Techncal Reports Department of Computer Scence 2011 Usng Past Queres for Resource Selecton n Dstrbuted Informaton Retreval Sulleyman Cetntas

More information

Prediction of Total Pressure Drop in Stenotic Coronary Arteries with Their Geometric Parameters

Prediction of Total Pressure Drop in Stenotic Coronary Arteries with Their Geometric Parameters Tenth Internatonal Conference on Computatonal Flud Dynamcs (ICCFD10), Barcelona, Span, July 9-13, 2018 ICCFD10-227 Predcton of Total Pressure Drop n Stenotc Coronary Arteres wth Ther Geometrc Parameters

More information

310 Int'l Conf. Par. and Dist. Proc. Tech. and Appl. PDPTA'16

310 Int'l Conf. Par. and Dist. Proc. Tech. and Appl. PDPTA'16 310 Int'l Conf. Par. and Dst. Proc. Tech. and Appl. PDPTA'16 Akra Sasatan and Hrosh Ish Graduate School of Informaton and Telecommuncaton Engneerng, Toka Unversty, Mnato, Tokyo, Japan Abstract The end-to-end

More information

NUMERICAL COMPARISONS OF BIOASSAY METHODS IN ESTIMATING LC50 TIANHONG ZHOU

NUMERICAL COMPARISONS OF BIOASSAY METHODS IN ESTIMATING LC50 TIANHONG ZHOU NUMERICAL COMPARISONS OF BIOASSAY METHODS IN ESTIMATING LC50 by TIANHONG ZHOU B.S., Chna Agrcultural Unversty, 2003 M.S., Chna Agrcultural Unversty, 2006 A THESIS submtted n partal fulfllment of the requrements

More information

A MIXTURE OF EXPERTS FOR CATARACT DIAGNOSIS IN HOSPITAL SCREENING DATA

A MIXTURE OF EXPERTS FOR CATARACT DIAGNOSIS IN HOSPITAL SCREENING DATA Journal of Theoretcal and Appled Informaton Technology 2005 ongong JATIT & LLS ISSN: 1992-8645 www.jatt.org E-ISSN: 1817-3195 A MIXTURE OF EXPERTS FOR CATARACT DIAGNOSIS IN HOSPITAL SCREENING DATA 1 SUNGMIN

More information

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and Ths artcle appeared n a journal publshed by Elsever. The attached copy s furnshed to the author for nternal non-commercal research and educaton use, ncludng for nstructon at the authors nsttuton and sharng

More information

Richard Williams Notre Dame Sociology Meetings of the European Survey Research Association Ljubljana,

Richard Williams Notre Dame Sociology   Meetings of the European Survey Research Association Ljubljana, Rchard Wllams Notre Dame Socology rwllam@nd.edu http://www.nd.edu/~rwllam Meetngs of the European Survey Research Assocaton Ljubljana, Slovena July 19, 2013 Comparng Logt and Probt Coeffcents across groups

More information

A comparison of statistical methods in interrupted time series analysis to estimate an intervention effect

A comparison of statistical methods in interrupted time series analysis to estimate an intervention effect Peer revew stream A comparson of statstcal methods n nterrupted tme seres analyss to estmate an nterventon effect a,b, J.J.J., Walter c, S., Grzebeta a, R. & Olver b, J. a Transport and Road Safety, Unversty

More information

Appendix F: The Grant Impact for SBIR Mills

Appendix F: The Grant Impact for SBIR Mills Appendx F: The Grant Impact for SBIR Mlls Asmallsubsetofthefrmsnmydataapplymorethanonce.Ofthe7,436applcant frms, 71% appled only once, and a further 14% appled twce. Wthn my data, seven companes each submtted

More information

Project title: Mathematical Models of Fish Populations in Marine Reserves

Project title: Mathematical Models of Fish Populations in Marine Reserves Applcaton for Fundng (Malaspna Research Fund) Date: November 0, 2005 Project ttle: Mathematcal Models of Fsh Populatons n Marne Reserves Dr. Lev V. Idels Unversty College Professor Mathematcs Department

More information

4.2 Scheduling to Minimize Maximum Lateness

4.2 Scheduling to Minimize Maximum Lateness 4. Schedulng to Mnmze Maxmum Lateness Schedulng to Mnmzng Maxmum Lateness Mnmzng lateness problem. Sngle resource processes one ob at a tme. Job requres t unts of processng tme and s due at tme d. If starts

More information

An Introduction to Modern Measurement Theory

An Introduction to Modern Measurement Theory An Introducton to Modern Measurement Theory Ths tutoral was wrtten as an ntroducton to the bascs of tem response theory (IRT) modelng and ts applcatons to health outcomes measurement for the Natonal Cancer

More information

Price linkages in value chains: methodology

Price linkages in value chains: methodology Prce lnkages n value chans: methodology Prof. Trond Bjorndal, CEMARE. Unversty of Portsmouth, UK. and Prof. José Fernández-Polanco Unversty of Cantabra, Span. FAO INFOSAMAK Tangers, Morocco 14 March 2012

More information

Gene Selection Based on Mutual Information for the Classification of Multi-class Cancer

Gene Selection Based on Mutual Information for the Classification of Multi-class Cancer Gene Selecton Based on Mutual Informaton for the Classfcaton of Mult-class Cancer Sheng-Bo Guo,, Mchael R. Lyu 3, and Tat-Mng Lok 4 Department of Automaton, Unversty of Scence and Technology of Chna, Hefe,

More information

ARTICLE IN PRESS Neuropsychologia xxx (2010) xxx xxx

ARTICLE IN PRESS Neuropsychologia xxx (2010) xxx xxx Neuropsychologa xxx (200) xxx xxx Contents lsts avalable at ScenceDrect Neuropsychologa journal homepage: www.elsever.com/locate/neuropsychologa Storage and bndng of object features n vsual workng memory

More information

Integration of sensory information within touch and across modalities

Integration of sensory information within touch and across modalities Integraton of sensory nformaton wthn touch and across modaltes Marc O. Ernst, Jean-Perre Brescan, Knut Drewng & Henrch H. Bülthoff Max Planck Insttute for Bologcal Cybernetcs 72076 Tübngen, Germany marc.ernst@tuebngen.mpg.de

More information

Estimating the distribution of the window period for recent HIV infections: A comparison of statistical methods

Estimating the distribution of the window period for recent HIV infections: A comparison of statistical methods Research Artcle Receved 30 September 2009, Accepted 15 March 2010 Publshed onlne n Wley Onlne Lbrary (wleyonlnelbrary.com) DOI: 10.1002/sm.3941 Estmatng the dstrbuton of the wndow perod for recent HIV

More information

Alma Mater Studiorum Università di Bologna DOTTORATO DI RICERCA IN METODOLOGIA STATISTICA PER LA RICERCA SCIENTIFICA

Alma Mater Studiorum Università di Bologna DOTTORATO DI RICERCA IN METODOLOGIA STATISTICA PER LA RICERCA SCIENTIFICA Alma Mater Studorum Unverstà d Bologna DOTTORATO DI RICERCA IN METODOLOGIA STATISTICA PER LA RICERCA SCIENTIFICA Cclo XXVII Settore Concorsuale d afferenza: 13/D1 Settore Scentfco dscplnare: SECS-S/02

More information

The effect of salvage therapy on survival in a longitudinal study with treatment by indication

The effect of salvage therapy on survival in a longitudinal study with treatment by indication Research Artcle Receved 28 October 2009, Accepted 8 June 2010 Publshed onlne 30 August 2010 n Wley Onlne Lbrary (wleyonlnelbrary.com) DOI: 10.1002/sm.4017 The effect of salvage therapy on survval n a longtudnal

More information

Impact of Imputation of Missing Data on Estimation of Survival Rates: An Example in Breast Cancer

Impact of Imputation of Missing Data on Estimation of Survival Rates: An Example in Breast Cancer Orgnal Artcle Impact of Imputaton of Mssng Data on Estmaton of Survval Rates: An Example n Breast Cancer Banesh MR 1, Tale AR 2 Abstract Background: Multfactoral regresson models are frequently used n

More information

Introduction ORIGINAL RESEARCH

Introduction ORIGINAL RESEARCH ORIGINAL RESEARCH Assessng the Statstcal Sgnfcance of the Acheved Classfcaton Error of Classfers Constructed usng Serum Peptde Profles, and a Prescrpton for Random Samplng Repeated Studes for Massve Hgh-Throughput

More information

Evaluation of the generalized gamma as a tool for treatment planning optimization

Evaluation of the generalized gamma as a tool for treatment planning optimization Internatonal Journal of Cancer Therapy and Oncology www.jcto.org Evaluaton of the generalzed gamma as a tool for treatment plannng optmzaton Emmanoul I Petrou 1,, Ganesh Narayanasamy 3, Eleftheros Lavdas

More information

IMPROVING THE EFFICIENCY OF BIOMARKER IDENTIFICATION USING BIOLOGICAL KNOWLEDGE

IMPROVING THE EFFICIENCY OF BIOMARKER IDENTIFICATION USING BIOLOGICAL KNOWLEDGE IMPROVING THE EFFICIENCY OF BIOMARKER IDENTIFICATION USING BIOLOGICAL KNOWLEDGE JOHN H. PHAN The Wallace H. Coulter Department of Bomedcal Engneerng, Georga Insttute of Technology, 313 Ferst Drve Atlanta,

More information

Saeed Ghanbari, Seyyed Mohammad Taghi Ayatollahi*, Najaf Zare

Saeed Ghanbari, Seyyed Mohammad Taghi Ayatollahi*, Najaf Zare DOI:http://dx.do.org/10.7314/APJCP.2015.16.14.5655 and Anthracyclne- Breast Cancer Treatment and Survval n the Eastern Medterranean and Asa: a Meta-analyss RESEARCH ARTICLE Comparng Role of Two Chemotherapy

More information

Economic crisis and follow-up of the conditions that define metabolic syndrome in a cohort of Catalonia,

Economic crisis and follow-up of the conditions that define metabolic syndrome in a cohort of Catalonia, Economc crss and follow-up of the condtons that defne metabolc syndrome n a cohort of Catalona, 2005-2012 Laa Maynou 1,2,3, Joan Gl 4, Gabrel Coll-de-Tuero 5,2, Ton Mora 6, Carme Saurna 1,2, Anton Scras

More information

Incorrect Beliefs. Overconfidence. Types of Overconfidence. Outline. Overprecision 4/22/2015. Econ 1820: Behavioral Economics Mark Dean Spring 2015

Incorrect Beliefs. Overconfidence. Types of Overconfidence. Outline. Overprecision 4/22/2015. Econ 1820: Behavioral Economics Mark Dean Spring 2015 Incorrect Belefs Overconfdence Econ 1820: Behavoral Economcs Mark Dean Sprng 2015 In objectve EU we assumed that everyone agreed on what the probabltes of dfferent events were In subjectve expected utlty

More information

HERMAN AGUINIS University of Colorado at Denver. SCOTT A. PETERSEN U.S. Military Academy at West Point. CHARLES A. PIERCE Montana State University

HERMAN AGUINIS University of Colorado at Denver. SCOTT A. PETERSEN U.S. Military Academy at West Point. CHARLES A. PIERCE Montana State University ORGANIZATIONAL Aguns et al. / MODERATING RESEARCH EFFECTS METHODS Apprasal of the Homogenety of Error Varance Assumpton and Alternatves to Multple Regresson for Estmatng Moderatng Effects of Categorcal

More information

Optimal probability weights for estimating causal effects of time-varying treatments with marginal structural Cox models

Optimal probability weights for estimating causal effects of time-varying treatments with marginal structural Cox models Optmal probablty weghts for estmatng causal effects of tme-varyng treatments wth margnal structural Cox models Mchele Santacatterna, Cela García-Pareja Rno Bellocco, Anders Sönnerborg, Anna Ma Ekström

More information

(From the Gastroenterology Division, Cornell University Medical College, New York 10021)

(From the Gastroenterology Division, Cornell University Medical College, New York 10021) ROLE OF HEPATIC ANION-BINDING PROTEIN IN BROMSULPHTHALEIN CONJUGATION* BY N. KAPLOWITZ, I. W. PERC -ROBB,~ ANn N. B. JAVITT (From the Gastroenterology Dvson, Cornell Unversty Medcal College, New York 10021)

More information

INITIAL ANALYSIS OF AWS-OBSERVED TEMPERATURE

INITIAL ANALYSIS OF AWS-OBSERVED TEMPERATURE INITIAL ANALYSIS OF AWS-OBSERVED TEMPERATURE Wang Yng, Lu Xaonng, Ren Zhhua, Natonal Meteorologcal Informaton Center, Bejng, Chna Tel.:+86 684755, E-mal:cdcsjk@cma.gov.cn Abstract From, n Chna meteorologcal

More information

J. H. Rohrer, S. H. Baron, E. L. Hoffman, D. V. Swander

J. H. Rohrer, S. H. Baron, E. L. Hoffman, D. V. Swander 2?Hr a! A Report of Research on o ^^ -^~" r" THE STABILITY OF AUTOKINETIC JUDGMENTS J. H. Rohrer, S. H. Baron, E. L. Hoffman, D. V. Swander A techncal report made under ONR Contract Nonr-475(01) between

More information

TOPICS IN HEALTH ECONOMETRICS

TOPICS IN HEALTH ECONOMETRICS TOPICS IN HEALTH ECONOMETRICS By VIDHURA SENANI BANDARA WIJAYAWARDHANA TENNEKOON A dssertaton submtted n partal fulfllment of the requrements for the degree of DOCTOR OF PHILOSOPHY WASHINGTON STATE UNIVERSITY

More information

Insights in Genetics and Genomics

Insights in Genetics and Genomics Insghts n Genetcs and Genomcs Research Artcle Open Access New Score Tests for Equalty of Varances n the Applcaton of DNA Methylaton Data Analyss [Verson ] Welang Qu Xuan L Jarrett Morrow Dawn L DeMeo Scott

More information

EVALUATION OF BULK MODULUS AND RING DIAMETER OF SOME TELLURITE GLASS SYSTEMS

EVALUATION OF BULK MODULUS AND RING DIAMETER OF SOME TELLURITE GLASS SYSTEMS Chalcogende Letters Vol. 12, No. 2, February 2015, p. 67-74 EVALUATION OF BULK MODULUS AND RING DIAMETER OF SOME TELLURITE GLASS SYSTEMS R. EL-MALLAWANY a*, M.S. GAAFAR b, N. VEERAIAH c a Physcs Dept.,

More information

The Reliability of Subjective Well-Being Measures

The Reliability of Subjective Well-Being Measures The Relablty of Subjectve Well-Beng Measures Alan B. Krueger Prnceton Unversty Davd A. Schkade Unversty of Calforna, San Dego Draft: August 2006 PRELIMINARY RESULTS: DO NOT CITE WITHOUT PERMISSION The

More information

Bimodal Bidding in Experimental All-Pay Auctions

Bimodal Bidding in Experimental All-Pay Auctions Bmodal Bddng n Expermental All-Pay Auctons Chrstane Ernst and Chrstan Thön August 2009 Dscusson Paper no. 2009-25 Department of Economcs Unversty of St. Gallen Edtor: Publsher: Electronc Publcaton: Martna

More information

NHS Outcomes Framework

NHS Outcomes Framework NHS Outcomes Framework Doman 1 Preventng people from dyng prematurely Indcator Specfcatons Verson: 1.21 Date: May 2018 Author: Clncal Indcators Team NHS Outcomes Framework: Doman 1 Preventng people from

More information

What Determines Attitude Improvements? Does Religiosity Help?

What Determines Attitude Improvements? Does Religiosity Help? Internatonal Journal of Busness and Socal Scence Vol. 4 No. 9; August 2013 What Determnes Atttude Improvements? Does Relgosty Help? Madhu S. Mohanty Calforna State Unversty-Los Angeles Los Angeles, 5151

More information

FAST DETECTION OF MASSES IN MAMMOGRAMS WITH DIFFICULT CASE EXCLUSION

FAST DETECTION OF MASSES IN MAMMOGRAMS WITH DIFFICULT CASE EXCLUSION computng@tanet.edu.te.ua www.tanet.edu.te.ua/computng ISSN 727-6209 Internatonal Scentfc Journal of Computng FAST DETECTION OF MASSES IN MAMMOGRAMS WITH DIFFICULT CASE EXCLUSION Gábor Takács ), Béla Patak

More information

HIV/AIDS-related Expectations and Risky Sexual Behavior in Malawi

HIV/AIDS-related Expectations and Risky Sexual Behavior in Malawi Unversty of Pennsylvana ScholarlyCommons PSC Workng Paper Seres 7-29-20 HIV/AIDS-related Expectatons and Rsky Sexual Behavor n Malaw Adelne Delavande RAND Corporaton, Nova School of Busness and Economcs

More information

A Meta-Analysis of the Effect of Education on Social Capital

A Meta-Analysis of the Effect of Education on Social Capital A Meta-Analyss of the Effect of Educaton on Socal Captal Huang Jan ** "Scholar" Research Center for Educaton and Labor Market Department of Economcs, Unversty of Amsterdam and Tnbergen Insttute by Henrëtte

More information

Non-linear Multiple-Cue Judgment Tasks

Non-linear Multiple-Cue Judgment Tasks Non-lnear Multple-Cue Tasks Anna-Carn Olsson (anna-carn.olsson@psy.umu.se) Department of Psychology, Umeå Unversty SE-09 87, Umeå, Sweden Tommy Enqvst (tommy.enqvst@psyk.uu.se) Department of Psychology,

More information

Non-parametric Survival Analysis for Breast Cancer Using nonmedical

Non-parametric Survival Analysis for Breast Cancer Using nonmedical IOSR Journal Of Humantes And Socal Scence (IOSR-JHSS) Volume 1, Issue 5, Ver. 1 (May. 16) PP -34 e-issn: 79-837, p-issn: 79-845. www.osrjournals.org Non-parametrc Survval Analyss for Breast Cancer Usng

More information

Unobserved Heterogeneity and the Statistical Analysis of Highway Accident Data

Unobserved Heterogeneity and the Statistical Analysis of Highway Accident Data Unobserved Heterogenety and the Statstcal Analyss of Hghway Accdent Data Fred L. Mannerng Professor of Cvl and Envronmental Engneerng Courtesy Department of Economcs Unversty of South Florda 4202 E. Fowler

More information

THIS IS AN OFFICIAL NH DHHS HEALTH ALERT

THIS IS AN OFFICIAL NH DHHS HEALTH ALERT THIS IS AN OFFICIAL NH DHHS HEALTH ALERT Dstrbuted by the NH Health Alert Network Health.Alert@dhhs.nh.gov August 26, 2016 1430 EDT (2:30 PM EDT) NH-HAN 20160826 Recommendatons for Accurate Dagnoss of

More information

WHO S ASSESSMENT OF HEALTH CARE INDUSTRY PERFORMANCE: RATING THE RANKINGS

WHO S ASSESSMENT OF HEALTH CARE INDUSTRY PERFORMANCE: RATING THE RANKINGS WHO S ASSESSMENT OF HEALTH CARE INDUSTRY PERFORMANCE: RATING THE RANKINGS ELLIOTT PARKER and JEANNE WENDEL * Department of Economcs, Unversty of Nevada, Reno, NV, USA SUMMARY Ths paper examnes the econometrc

More information

Length of Hospital Stay After Acute Myocardial Infarction in the Myocardial Infarction Triage and Intervention (MITI) Project Registry

Length of Hospital Stay After Acute Myocardial Infarction in the Myocardial Infarction Triage and Intervention (MITI) Project Registry JACC Vol. 28, No. 2 287 CLINICAL STUDIES MYOCARDIAL INFARCTION Length of Hosptal Stay After Acute Myocardal Infarcton n the Myocardal Infarcton Trage and Interventon (MITI) Project Regstry NATHAN R. EVERY,

More information

Economists are increasingly analyzing data on subjective well-being. Since 2000, 157

Economists are increasingly analyzing data on subjective well-being. Since 2000, 157 The Relablty of Subjectve Well-Beng Measures by Alan B. Krueger, Prnceton Unversty Davd A. Schkade, Unversty of Calforna, San Dego CEPS Workng Paper No. 138 January 007 The authors thank our colleagues

More information

Strategies for the Early Diagnosis of Acute Myocardial Infarction Using Biochemical Markers

Strategies for the Early Diagnosis of Acute Myocardial Infarction Using Biochemical Markers Clncal Chemstry / EARLY DIAGNOSIS OF ACUTE MYOCARDIAL INFARCTION USING IOCHEMICAL MARKERS Strateges for the Early Dagnoss of Acute Myocardal Infarcton Usng ochemcal Markers Martna Zannotto, Leopoldo Celegon,

More information

Chapter 20. Aggregation and calibration. Betina Dimaranan, Thomas Hertel, Robert McDougall

Chapter 20. Aggregation and calibration. Betina Dimaranan, Thomas Hertel, Robert McDougall Chapter 20 Aggregaton and calbraton Betna Dmaranan, Thomas Hertel, Robert McDougall In the prevous chapter we dscussed how the fnal verson 3 GTAP data base was assembled. Ths data base s extremely large.

More information

National Polyp Study data: evidence for regression of adenomas

National Polyp Study data: evidence for regression of adenomas 5 Natonal Polyp Study data: evdence for regresson of adenomas 78 Chapter 5 Abstract Objectves The data of the Natonal Polyp Study, a large longtudnal study on survellance of adenoma patents, s used for

More information

CONSTRUCTION OF STOCHASTIC MODEL FOR TIME TO DENGUE VIRUS TRANSMISSION WITH EXPONENTIAL DISTRIBUTION

CONSTRUCTION OF STOCHASTIC MODEL FOR TIME TO DENGUE VIRUS TRANSMISSION WITH EXPONENTIAL DISTRIBUTION Internatonal Journal of Pure and Appled Mathematcal Scences. ISSN 97-988 Volume, Number (7), pp. 3- Research Inda Publcatons http://www.rpublcaton.com ONSTRUTION OF STOHASTI MODEL FOR TIME TO DENGUE VIRUS

More information

Latent Class Analysis for Marketing Scales Development

Latent Class Analysis for Marketing Scales Development Workng Paper Seres, N.16, 2009 Latent Class Analyss for Marketng Scales Development Francesca Bass Department of Statstcal Scences Unversty of Padua Italy Abstract: Measurement scales are a crucal nstrument

More information

Importance of Atrial Compliance in Cardiac Performance

Importance of Atrial Compliance in Cardiac Performance Importance of Atral Complance n Cardac Performance By Hroyuk Suga ABSTRACT Effects of changes n atral complance on cardac performance were analyzed usng a crculatory analog model. The atrum was assumed

More information

Drug Prescription Behavior and Decision Support Systems

Drug Prescription Behavior and Decision Support Systems Drug Prescrpton Behavor and Decson Support Systems ABSTRACT Adverse drug events plague the outcomes of health care servces. In ths research, we propose a clncal learnng model that ncorporates the use of

More information

Investigation of zinc oxide thin film by spectroscopic ellipsometry

Investigation of zinc oxide thin film by spectroscopic ellipsometry VNU Journal of Scence, Mathematcs - Physcs 24 (2008) 16-23 Investgaton of znc oxde thn flm by spectroscopc ellpsometry Nguyen Nang Dnh 1, Tran Quang Trung 2, Le Khac Bnh 2, Nguyen Dang Khoa 2, Vo Th Ma

More information

Combined Temporal and Spatial Filter Structures for CDMA Systems

Combined Temporal and Spatial Filter Structures for CDMA Systems Combned Temporal and Spatal Flter Structures for CDMA Systems Ayln Yener WINLAB, Rutgers Unversty yener@wnlab.rutgers.edu Roy D. Yates WINLAB, Rutgers Unversty ryates@wnlab.rutgers.edu Sennur Ulukus AT&T

More information

Rainbow trout survival and capture probabilities in the upper Rangitikei River, New Zealand

Rainbow trout survival and capture probabilities in the upper Rangitikei River, New Zealand Ranbow trout survval and capture probabltes n the upper Rangtke Rver, New Zealand Rchard J Barker Department of Mathematcs and Statstcs Unversty of Otago P.O. Box 56 Dunedn, New Zealand Peter H Taylor

More information

BIOSTATISTICS. Lecture 1 Data Presentation and Descriptive Statistics. dr. Petr Nazarov

BIOSTATISTICS. Lecture 1 Data Presentation and Descriptive Statistics. dr. Petr Nazarov Genomcs Research Unt BIOSTATISTICS Lecture 1 Data Presentaton and Descrptve Statstcs dr. Petr Nazarov 3-03-2017 petr.nazarov@lh.lu COURSE OVERVIEW Organzaton: 60h = 12 days Theoretcal course (30h) Theory

More information

Evaluation of Literature-based Discovery Systems

Evaluation of Literature-based Discovery Systems Evaluaton of Lterature-based Dscovery Systems Melha Yetsgen-Yldz 1 and Wanda Pratt 1,2 1 The Informaton School, Unversty of Washngton, Seattle, USA. 2 Bomedcal and Health Informatcs, School of Medcne,

More information

Encoding processes, in memory scanning tasks

Encoding processes, in memory scanning tasks vlemory & Cognton 1976,4 (5), 501 506 Encodng processes, n memory scannng tasks JEFFREY O. MILLER and ROBERT G. PACHELLA Unversty of Mchgan, Ann Arbor, Mchgan 48101, Three experments are presented that

More information

HIV/AIDS-related Expectations and Risky Sexual Behavior in Malawi

HIV/AIDS-related Expectations and Risky Sexual Behavior in Malawi HIV/AIDS-related Expectatons and Rsky Sexual Behavor n Malaw Adelne Delavande Unversty of Essex and RAND Corporaton Hans-Peter Kohler Unversty of Pennsylvanna January 202 Abstract We use probablstc expectatons

More information

A Novel artifact for evaluating accuracies of gear profile and pitch measurements of gear measuring instruments

A Novel artifact for evaluating accuracies of gear profile and pitch measurements of gear measuring instruments A Novel artfact for evaluatng accuraces of gear profle and ptch measurements of gear measurng nstruments Sonko Osawa, Osamu Sato, Yohan Kondo, Toshyuk Takatsuj (NMIJ/AIST) Masaharu Komor (Kyoto Unversty)

More information

Incorporating prior biological knowledge for network-based differential gene expression analysis using differentially weighted graphical LASSO

Incorporating prior biological knowledge for network-based differential gene expression analysis using differentially weighted graphical LASSO Zuo et al. BMC Bonformatcs (2017) 18:99 DOI 10.1186/s12859-017-1515-1 METHODOLOGY ARTICLE Open Access Incorporatng pror bologcal knowledge for network-based dfferental gene expresson analyss usng dfferentally

More information

Prototypes in the Mist: The Early Epochs of Category Learning

Prototypes in the Mist: The Early Epochs of Category Learning Journal of Expermental Psychology: Learnng, Memory, and Cognton 1998, Vol. 24, No. 6, 1411-1436 Copyrght 1998 by the Amercan Psychologcal Assocaton, Inc. 0278-7393/98/S3.00 Prototypes n the Mst: The Early

More information

Journal of Economic Behavior & Organization

Journal of Economic Behavior & Organization Journal of Economc Behavor & Organzaton 133 (2017) 52 73 Contents lsts avalable at ScenceDrect Journal of Economc Behavor & Organzaton j ourna l ho me pa g e: www.elsever.com/locate/jebo Perceptons, ntentons,

More information

UNIVERISTY OF KWAZULU-NATAL, PIETERMARITZBURG SCHOOL OF MATHEMATICS, STATISTICS AND COMPUTER SCIENCE

UNIVERISTY OF KWAZULU-NATAL, PIETERMARITZBURG SCHOOL OF MATHEMATICS, STATISTICS AND COMPUTER SCIENCE UNIVERISTY OF KWAZULU-NATAL, PIETERMARITZBURG SCHOOL OF MATHEMATICS, STATISTICS AND COMPUTER SCIENCE A COMPLEX SURVEY DATA ANALYSIS OF TB AND HIV MORTALITY IN SOUTH AFRICA By JOIE LEA MURORUNKWERE STUDENT

More information

Heart Rate Variability Analysis Diagnosing Atrial Fibrillation

Heart Rate Variability Analysis Diagnosing Atrial Fibrillation X-ray PIV Measurements of Velocty Feld of Blood Flows Volume 5, umber 2: 46-52, October 2007 Internatonal Journal of Vascular Bomedcal Engneerng Heart Rate Varablty Analyss Dagnosng Atral Fbrllaton Jnho

More information

Does reporting heterogeneity bias the measurement of health disparities?

Does reporting heterogeneity bias the measurement of health disparities? HEDG Workng Paper 06/03 Does reportng heterogenety bas the measurement of health dspartes? Teresa Bago d Uva Eddy Van Doorslaer Maarten Lndeboom Owen O Donnell Somnath Chatterj March 2006 ISSN 1751-1976

More information

Subject-Adaptive Real-Time Sleep Stage Classification Based on Conditional Random Field

Subject-Adaptive Real-Time Sleep Stage Classification Based on Conditional Random Field Subject-Adaptve Real-Tme Sleep Stage Classfcaton Based on Condtonal Random Feld Gang Luo, PhD, Wanl Mn, PhD IBM TJ Watson Research Center, Hawthorne, NY {luog, wanlmn}@usbmcom Abstract Sleep stagng s the

More information

INTEGRATIVE NETWORK ANALYSIS TO IDENTIFY ABERRANT PATHWAY NETWORKS IN OVARIAN CANCER

INTEGRATIVE NETWORK ANALYSIS TO IDENTIFY ABERRANT PATHWAY NETWORKS IN OVARIAN CANCER INTEGRATIVE NETWORK ANALYSIS TO IDENTIFY ABERRANT PATHWAY NETWORKS IN OVARIAN CANCER LI CHEN 1,2, JIANHUA XUAN 1,*, JINGHUA GU 1, YUE WANG 1, ZHEN ZHANG 2, TIAN LI WANG 2, IE MING SHIH 2 1The Bradley Department

More information

Fast Algorithm for Vectorcardiogram and Interbeat Intervals Analysis: Application for Premature Ventricular Contractions Classification

Fast Algorithm for Vectorcardiogram and Interbeat Intervals Analysis: Application for Premature Ventricular Contractions Classification Fast Algorthm for Vectorcardogram and Interbeat Intervals Analyss: Applcaton for Premature Ventrcular Contractons Classfcaton Irena Jekova, Vessela Krasteva Centre of Bomedcal Engneerng Prof. Ivan Daskalov

More information

Does Context Matter More for Hypothetical Than for Actual Contributions?

Does Context Matter More for Hypothetical Than for Actual Contributions? Dscusson Paper Seres March 2008 EfD DP 08-02 Does Context Matter More for Hypothetcal Than for Actual Contrbutons? Evdence from a Natural Feld Experment Francsco Alpzar, Fredrk Carlsson, and Olof Johansson-Stenman

More information

Causal inference in nonexperimental studies typically

Causal inference in nonexperimental studies typically Orgnal Artcle Regresson Dscontnuty Desgns n Epdemology Causal Inference Wthout Randomzed Trals Jacob Bor, a,b,c Ellen Moscoe, c Porta Mutevedz, b Mare-Louse Newell, b,d and Tll Bärnghausen b,c Abstract:

More information

Debunking mathematically the logical fallacy that cancer risk is just bad luck

Debunking mathematically the logical fallacy that cancer risk is just bad luck Sornette and Favre EPJ Nonlnear Bomedcal Physcs (2015) 3:10 DOI 10.1140/epjnbp/s40366-015-0026-0 LETTER Open Access Debunkng mathematcally the logcal fallacy that cancer rsk s just bad luck D. Sornette

More information

Single-Case Designs and Clinical Biofeedback Experimentation

Single-Case Designs and Clinical Biofeedback Experimentation Bofeedback and Self-Regulaton, VoL 2, No. 3, 1977 Sngle-Case Desgns and Clncal Bofeedback Expermentaton Davd H. Barow: Brown Unversty and Butler Hosptal Edward B. Blanchard Unversty of Tennessee Medcal

More information

HIV/AIDS AND POVERTY IN SOUTH AFRICA: A BAYESIAN ESTIMATION OF SELECTION MODELS WITH CORRELATED FIXED-EFFECTS

HIV/AIDS AND POVERTY IN SOUTH AFRICA: A BAYESIAN ESTIMATION OF SELECTION MODELS WITH CORRELATED FIXED-EFFECTS HIV/AIDS AND POVERTY IN SOUTH AFRICA: A BAYESIAN ESTIMATION OF SELECTION MODELS WITH CORRELATED FIXED-EFFECTS FABRICE MURTIN* AND FEDERICA MARZO Abstract In ths paper, we estmate the causal mpact of human

More information

A New Diagnosis Loseless Compression Method for Digital Mammography Based on Multiple Arbitrary Shape ROIs Coding Framework

A New Diagnosis Loseless Compression Method for Digital Mammography Based on Multiple Arbitrary Shape ROIs Coding Framework I.J.Modern Educaton and Computer Scence, 2011, 5, 33-39 Publshed Onlne August 2011 n MECS (http://www.mecs-press.org/) A New Dagnoss Loseless Compresson Method for Dgtal Mammography Based on Multple Arbtrary

More information

Estimation for Pavement Performance Curve based on Kyoto Model : A Case Study for Highway in the State of Sao Paulo

Estimation for Pavement Performance Curve based on Kyoto Model : A Case Study for Highway in the State of Sao Paulo Estmaton for Pavement Performance Curve based on Kyoto Model : A Case Study for Kazuya AOKI, PASCO CORPORATION, Yokohama, JAPAN, Emal : kakzo603@pasco.co.jp Octávo de Souza Campos, Publc Servces Regulatory

More information

A Linear Regression Model to Detect User Emotion for Touch Input Interactive Systems

A Linear Regression Model to Detect User Emotion for Touch Input Interactive Systems 2015 Internatonal Conference on Affectve Computng and Intellgent Interacton (ACII) A Lnear Regresson Model to Detect User Emoton for Touch Input Interactve Systems Samt Bhattacharya Dept of Computer Scence

More information

A GEOGRAPHICAL AND STATISTICAL ANALYSIS OF LEUKEMIA DEATHS RELATING TO NUCLEAR POWER PLANTS. Whitney Thompson, Sarah McGinnis, Darius McDaniel,

A GEOGRAPHICAL AND STATISTICAL ANALYSIS OF LEUKEMIA DEATHS RELATING TO NUCLEAR POWER PLANTS. Whitney Thompson, Sarah McGinnis, Darius McDaniel, A GEOGRAPHICAL AD STATISTICAL AALYSIS OF LEUKEMIA DEATHS RELATIG TO UCLEAR POWER PLATS Whtney Thompson, Sarah McGnns, Darus McDanel, Jean Sexton, Rebecca Pettt, Sarah Anderson, Monca Jackson ABSTRACT:

More information

ENRICHING PROCESS OF ICE-CREAM RECOMMENDATION USING COMBINATORIAL RANKING OF AHP AND MONTE CARLO AHP

ENRICHING PROCESS OF ICE-CREAM RECOMMENDATION USING COMBINATORIAL RANKING OF AHP AND MONTE CARLO AHP ENRICHING PROCESS OF ICE-CREAM RECOMMENDATION USING COMBINATORIAL RANKING OF AHP AND MONTE CARLO AHP 1 AKASH RAMESHWAR LADDHA, 2 RAHUL RAGHVENDRA JOSHI, 3 Dr.PEETI MULAY 1 M.Tech, Department of Computer

More information

Are Drinkers Prone to Engage in Risky Sexual Behaviors?

Are Drinkers Prone to Engage in Risky Sexual Behaviors? Amercan Internatonal Journal of Socal Scence Vol. 2 No. 5; July 2013 Are Drnkers Prone to Engage n Rsky Sexual Behavors? Ana Isabel Gl Lacruz Zaragoza Unversty Department of Busness Organzaton and Management

More information

Are National School Lunch Program Participants More Likely to be Obese? Dealing with Identification

Are National School Lunch Program Participants More Likely to be Obese? Dealing with Identification Are Natonal School Lunch Program Partcpants More Lkely to be Obese? Dealng wth Identfcaton Janet G. Peckham Graduate Student, Clemson Unversty (jgemml@clemson.edu) Jaclyn D. Kropp Assstant Professor, Clemson

More information