《统计决策分析》课程授课教案（讲稿）Statistical Decision Analysis（英文讲义）

Coursedescription:Thegoal ofthis courseisto laya solidfoundationonethical,statisticalandlogisticalissuesaboutdesigningandanalyzingclinicaltrials.Theemphasisisonimportantmethodsofdesigningandanalyzingresponseadaptiveclinicaltrials.TopicsincludePhasellltrialrandomization,selectionbias,responseadaptivedesigns andtheirinference,largesampleinference,practicalissuesandopenproblemsPrerequisites:Goodknowledgeonprobabilityandstochasticprocesses,includinglimittheoremsandmartingaleslikelihoodbasedinferenceand asymptoticstatistics.福NIVERSITMANToaXikui Wang (University of Manitoba)Statistical Decision Analysis20134/198

Course description: The goal of this course is to lay a solid foundation on ethical, statistical and logistical issues about designing and analyzing clinical trials. The emphasis is on important methods of designing and analyzing response adaptive clinical trials. Topics include Phase III trial randomization, selection bias, response adaptive designs and their inference, large sample inference, practical issues and open problems. Prerequisites: Good knowledge on probability and stochastic processes, including limit theorems and martingales, likelihood based inference and asymptotic statistics. Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 4 / 198

Referencebooks:Clinicaltrialdesign-Bayesianandfrequentistadaptivemethods,byG.Yin,JohnWileyandSons,2012Randomizationinclinicaltrials-theoryandpractice,byW.FRosenbergerandJ.M.Lachin,JohnWileyandSons,2o02Thetheoryofresponse-adaptiverandomization inclinicaltrials,byF.HuandW.F.Rosenberger,JohnWileyandSons,2006Introductiontostatisticalmethodsforclinicaltrials,eds.byT.D.CookandD.L.DeMets,ChapmanandHall/CRC,2o08Evaluatingclinicalresearch(2nded.),byB.D.FurbergandC.D. Furberg, Springer, 2007Controversialstatisticalissuesinclinicaltrials,byS-C.Chow,ChapmanandHall,2011福UNIVERSITMANTToaKikui Wang (University of Manitoba)Statistical Decision Analysis20135/198

Reference books: Clinical trial design - Bayesian and frequentist adaptive methods, by G. Yin, John Wiley and Sons, 2012. Randomization in clinical trials - theory and practice, by W.F. Rosenberger and J.M. Lachin, John Wiley and Sons, 2002 The theory of response-adaptive randomization in clinical trials, by F. Hu and W.F. Rosenberger, John Wiley and Sons, 2006 Introduction to statistical methods for clinical trials, eds. by T.D. Cook and D.L. DeMets, Chapman and Hall/CRC, 2008 Evaluating clinical research (2nd ed.), by B.D. Furberg and C.D. Furberg, Springer, 2007 Controversial statistical issues in clinical trials, by S-C. Chow, Chapman and Hall, 2011 Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 5 / 198

Referencepapers:OX.Wang(2013).Banditprocessesandresponseadaptiveclinicaltrials:theartof explorationversusexploitation.MethodsandApplicationsof Statistics inClinicalTrialsVolume2:Planning,Analysis,andInferentialMethodsX.LiandX.Wang(2013).Responseadaptivedesignswithmisclassifiedresponses.CommunicationsinStatistics-TheoryandMethods42:2071-2083.X.LiandX.Wang(2012).Variance-penalizedresponse-3adaptiverandomization withmismeasurement.JournalofStatisticalPlanningandInference142:2128-2135.Y.Yi and Xikui Wang (2011). Comparison of Wald, score,and likelihoodratiotestsforresponseadaptivedesignsJournalofStatisticalTheoryandApplications10:553-569D.Tolussoand X.Wang (2011).Interval estimation for5responseadaptiveclinicaltrials.ComputationalStatisticsIVERSITLMANITONandDataAnalysis55:725-730.GkuiWang(Universityof Manitoba)Statistical Decision Analysis20136/198

Reference papers: 1 X. Wang (2013). Bandit processes and response adaptive clinical trials: the art of exploration versus exploitation. Methods and Applications of Statistics in Clinical Trials, Volume 2: Planning, Analysis, and Inferential Methods 2 X. Li and X. Wang (2013). Response adaptive designs with misclassified responses. Communications in Statistics - Theory and Methods 42: 2071-2083. 3 X.Li and X. Wang (2012). Variance-penalized response￾adaptive randomization with mismeasurement. Journal of Statistical Planning and Inference 142: 2128-2135. 4 Y. Yi and Xikui Wang (2011). Comparison of Wald, score, and likelihood ratio tests for response adaptive designs. Journal of Statistical Theory and Applications 10: 553-569. 5 D. Tolusso and X. Wang (2011). Interval estimation for response adaptive clinical trials. Computational Statistics and Data Analysis 55: 725-730. Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 6 / 198

Referencepapers:OY.YiandXikuiWang(2009).Responseadaptivedesignswithavariance-penalizedcriterion.BiometricalJournal51:763-773.Y.YiandXikuiWang(2008).Asymptotically efficientestimationinresponseadaptivetrials.Journalof StatisticalPlanningandInference138:2899-2905.Y.YiandXikuiWang(2007).Statisticalinferenceof3adaptivedesignswithbinaryresponses.CommunicationsinStatistics-TheoryandMethods36:1643-1653.Y.YiandXikuiWang(2007).Goodness-of-fittestforresponseadaptivedesigns.StatisticsandProbabilityLetters77:1014-1020.D.PullmanandX.Wang(2001).Adaptivedesigns,5informed consent,and the ethics ofresearch.Controlled鞋UNIVERSITClinicalTrials22:203-210IMANITOSXikuiWang(UniversityofManitoba)Statistical Decision Analysis20137/198

Reference papers: 1 Y. Yi and Xikui Wang (2009). Response adaptive designs with a variance-penalized criterion. Biometrical Journal 51: 763-773. 2 Y. Yi and Xikui Wang (2008). Asymptotically efficient estimation in response adaptive trials. Journal of Statistical Planning and Inference 138: 2899-2905. 3 Y. Yi and Xikui Wang (2007). Statistical inference of adaptive designs with binary responses. Communications in Statistics - Theory and Methods 36: 1643-1653. 4 Y. Yi and Xikui Wang (2007). Goodness-of-fit test for response adaptive designs. Statistics and Probability Letters 77: 1014-1020. 5 D. Pullman and X. Wang (2001). Adaptive designs, informed consent, and the ethics of research. Controlled Clinical Trials 22: 203-210. Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 7 / 198

医药使人生病Medicinemakespeopleill数学使人悲伤mathematicsmakesthemsadand theology makes them sinful.神学使人有罪MartinLuther(1483-1546)NIVERSITMANITONXikui Wang (University of Manitoba)Statistical Decision Analysis20138/198

Medicine makes people ill, mathematics makes them sad and theology makes them sinful. 医药使人生病 数学使人悲伤 神学使人有罪 Martin Luther (1483 - 1546) Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 8 / 198

Clinical trials and response adaptive designsClinical trials are regarded as the most reliable andefficient wayto evaluatethe efficacy and safety ofnewmedical interventions.1.Clinical experimentationonhumansubjectsrequiresacareful balancingactbetweenthecollectivegood(collectiveethics)andthe benefits ofthe individuals inthetrial(individualethics),particularlyindesperatemedicalsituations.2.Responseadaptivedesigns representamajoradvancementinclinicaltrialmethodologythathelpsbalancetheseethicalissuesandimproveefficiencywithoutunderminingthevalidityand integrity of the clinical research.They use accumulatinginformation fromthetrialtodeliberatelybiastreatmentrandomizationin ordertoassignmorepatientsto thepotentiallybettertreatment.Xikui Wang (University of Manitoba)Statistical DecisionAnalysis20139/198

Clinical trials and response adaptive designs Clinical trials are regarded as the most reliable and efficient way to evaluate the efficacy and safety of new medical interventions. 1. Clinical experimentation on human subjects requires a careful balancing act between the collective good (collective ethics) and the benefits of the individuals in the trial (individual ethics), particularly in desperate medical situations. 2. Response adaptive designs represent a major advancement in clinical trial methodology that helps balance these ethical issues and improve efficiency without undermining the validity and integrity of the clinical research. They use accumulating information from the trial to deliberately bias treatment randomization in order to assign more patients to the potentially better treatment. Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 9 / 198

CourseobjectivesUponcompletingthecourse,youshouldbeabletounderstandthefundamental principlesandmethodsof(responseadaptive)clinicaltrials,appreciatethecomplexethicalissuesofclinicaltrials,understandthestrengthsandweaknessesofvariousmethodsofmedicalresearch(suchasepidemiologicalstudies and clinical trials),criticallyappraisethedifferentgoalsandmethodsofrandomization in Phase Ill clinical trials,layaintroductoryfoundationforresearchintheareaofclinicaltrials,especiallyonresponseadaptivedesigns.福NIVERSITIMANITToaXikui Wang (University of Manitoba)Statistical Decision Analysis201310/198

Course objectives Upon completing the course, you should be able to understand the fundamental principles and methods of (response adaptive) clinical trials, appreciate the complex ethical issues of clinical trials, understand the strengths and weaknesses of various methods of medical research (such as epidemiological studies and clinical trials), critically appraise the different goals and methods of randomization in Phase III clinical trials, lay a introductory foundation for research in the area of clinical trials, especially on response adaptive designs. Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 10 / 198

教学大纲1.临床试验的基础：历史及主要内容，方法和意义2.临床试验的道德问题：群体道德及个人道德3.临床试验的随机分配：分类及选择误差4.反应自适用试验设计的基本问题和基本方法5.反应自适用随机分配的统计设计：参数方法6.带有测量误差的反应自适用随机分配7.反应自适用随机分配的统计分析8.反应自适用随机分配的未来展望UNIVERSITY1.IMANITOSXikui Wang (University of Manitoba)Statistical Decision Analysis11/1982013

教学大纲 1. 临床试验的基础：历史及主要内容，方法和意义 2. 临床试验的道德问题：群体道德及个人道德 3. 临床试验的随机分配：分类及选择误差 4. 反应自适用试验设计的基本问题和基本方法 5. 反应自适用随机分配的统计设计：参数方法 6. 带有测量误差的反应自适用随机分配 7. 反应自适用随机分配的统计分析 8. 反应自适用随机分配的未来展望 Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 11 / 198

OutlineFoundation of clinical trials: history,important goals andissuesEthical issues of clinical trials:collective ethics and individuaethicsRandomization in clinicaltrials:classification and selectionbiasFundamental questions andmethods of responseadaptivedesignsStatisticaldesignofresponseadaptiverandomizationparametric methodsResponseadaptiverandomizationwithmeasurementerrors7Statistical inference of response adaptive randomizationThefuture of response adaptiverandomization蓝JNIVERSITYLMANITOSAXikui Wang(UniversityofManitoba)12/198Statistical DecisionAnalysis2013

Outline 1 Foundation of clinical trials: history, important goals and issues 2 Ethical issues of clinical trials: collective ethics and individual ethics 3 Randomization in clinical trials: classification and selection bias 4 Fundamental questions and methods of response adaptive designs 5 Statistical design of response adaptive randomization: parametric methods 6 Response adaptive randomization with measurement errors 7 Statistical inference of response adaptive randomization 8 The future of response adaptive randomization Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 12 / 198

1.Foundation of clinicaltrialsAccording to the New Oxford DictionaryAdiseaseis"a disorderof structure orfunction inahuman,animal,orplant,especiallyonethatproducesspecificsigns orsymptomsorthataffectsaspecificlocationandisnotsimplyadirectresultofphysicalinjury"(http://oxforddictionaries.com/)(Origin:MiddleEnglish,inthesenselackofease',dis-ease)Factsaboutdisease:Thingsthatare inevitable inlife:tax,diseaseanddeathTypesofillness:acute(headache,nausea,etc.)andchronic(cancer,heartdisease,hypertension,etc.)Formsofdisease:physicalandmental福JNIVERSITYLMANITOSXikui Wang (University of Manitoba)Statistical Decision Analysis13/1982013

1. Foundation of clinical trials According to the New Oxford Dictionary A disease is “a disorder of structure or function in a human, animal, or plant, especially one that produces specific signs or symptoms or that affects a specific location and is not simply a direct result of physical injury.” (http://oxforddictionaries.com/) (Origin: Middle English, in the sense ‘lack of ease’, dis-ease) Facts about disease: Things that are inevitable in life: tax, disease and death Types of illness: acute (headache, nausea, etc.) and chronic (cancer, heart disease, hypertension, etc.) Forms of disease: physical and mental Xikui Wang (University of Manitoba) Statistical Decision Analysis 2013 13 / 198