Peter F. Thall’s Personal Website



Here is a picture of me being a biostatistician, with a pen, a pad, a computer, and everything.  This is what “science” looks like in real life.


A proper website seems to require a list of hobbies. In keeping with this dubious convention, here are some of mine:

Swimming; trading stock options; nonverbal canine communication; aggressively practicing humility; intergalactic exploration; watching nearly every movie ever made; gourmet cooking; giving terribly useful advice to people who seem to need it; staying in hotels in cities I probably wouldn’t visit if I weren’t attending  a conference; collecting not-too-expensive artwork. 

Curriculum Vitae   

For a copy of my CV, click here.


Book on Dose Finding Trials

Bayesian Designs for Phase I-II Clinical Trials. Yuan, Nguyen, and Thall.  Chapman & Hall/CRC Biostatistics Series, 2016.

Actual Reviews by Really Smart People

"This book is a must-read for students, statisticians, principal investigators and researchers who wish to apply innovative and more ethical designs for Phase I/II clinical trials. Several statisticians have previously proposed designs for dose-finding studies modelling the dose-toxicity and the dose-efficacy relationships. However such methods have been published in highly specialized statistical/biostatistical journals that are not very accessible nor comprehensible for non-initiated readers. To the best of my knowledge, no book has yet solely focused on the design of Phase I/II clinical trials, despite the fact that these studies represent 33% of all conducted trials (source: This excellent book offers a well written and a step by step guide to planning, conducting and analyzing Phase I/II clinical trials." —Sarah Zohar, The French National Institute of Health and Medical Research (Inserm), Paris

"Yuan, Nguyen and Thall are statisticians on the forefront of both theoretical statistics and practical implementation of adaptive trial designs, and have combined their knowledge and experience here to provide an exceptional textbook… A highlight of the text is a chapter on choosing priors, where the authors demonstrate that prior calibration is critical. Casual choice of priors can be disastrous in these trials (which have small cohorts and often small sample sizes), and Yuan et al. provide examples to demonstrate how poorly chosen priors can ruin operating characteristics. Complex trial designs are explained in a clear and sensible manner, making the arguments very plainly obvious as to the benefits of these more modern designs. The authors have pioneered adaptive approaches for dose-finding combining toxicity and efficacy trade-offs and present these and other designs that jointly model toxicity and efficacy… The writing style is conversational in places, making this text more enjoyable to read than many other statistics textbooks, and readers will appreciate the chapters that address practical problems often ignored in theoretical clinical trial texts: late onset and cumulative (toxicity) outcomes, molecularly targeted agents, and missing data in adaptive designs. Interactive software with a user-friendly interface is available for many of the designs, with illustrations in the text which demonstrate the implementation." —Elizabeth Garrett-Mayer, Professor of Biostatistics, Medical University of South Carolina

"This book covers almost every topic that you will need when designing Phase I, Phase II, and Phase I-II clinical trials. Each chapter is a treasure trove of wonderful new ideas, and contains examples - based on the authors' outstandingly broad experiences – that help the reader clearly understand the methodological aspects involved in clinical trials… This book is a "must-have" for every biostatistician involved in clinical trials." —Satoshi Morita, Department of Biomedical Statistics and Bioinformatics, Kyoto University



If you want to read some of my publications, click on the links below.  All of them are about statistics, medicine, or clinical trials, and most are astonishingly well written.  If you like one of them, be sure to tell your friends and relatives about it.

Random Measures With Aftereffects, Annals of Probability, 1978

A theorem on regular infinitely divisible Cox processes. Stochastic Processes and Their Applications 16: 205-210, 1983

Two-stage selection and testing designs for comparative clinical trials, Biometrika, 1988

Analysis of recurrent events: nonparametric methods for random interval count data.  JASA 83: 339-347, 1988

Poisson likelihood regression models for longitudinal interval count data. Biometrics 44: 197-209, 1988

Some covariance models for longitudinal count data with overdispersion. Biometrics 46: 657-671, 1990.

Test-based variable selection via cross-validation. J Computational and Graphical Stat 1: 41-61, 1992

Practical Bayesian guidelines for phase IIB clinical trials. Biometrics 50: 337-349, 1994

Bayesian sequential monitoring designs for single-arm clinical trials with multiple outcomes. Stat in Med 14:357-379, 1995

Estimating genomic category probabilities from fluorescent in situ hybridization counts with misclassification. JRSS-C (Applied Statistics) 45:431-446, 1996

Variable selection in regression via repeated data splitting.  J Comp and Graphical Stat, 6:416-434, 1997

Parametric likelihoods for multiple non-fatal competing risks and death. Stat in Med, 17:999-1016, 1998

Some extensions and applications of a Bayesian strategy for monitoring multiple outcomes in clinical trials.  Stat in Med, 17:1563-1580, 1998.

A strategy for dose finding and safety monitoring based on efficacy and adverse outcomes in phase I/II clinical trials.  Biometrics 54:251-264, 1998.

Treatment Comparisons Based on Two-Dimensional Safety and Efficacy Alternatives in Oncology Trials, Biometrics 1999

Decision Theoretic Designs for Phase II Clinical Trials with Multiple Outcomes, Biometrics 1999

Accrual strategies for phase I trials with delayed patient outcome, Statistics in Medicine, 1999

Decision Theoretic Designs for Phase II Clinical Trials with Multiple Outcomes, Biometrics 1999

Evaluating multiple treatment courses in clinical trials.  Stat in Med, 19: 1011-1028, 2000

Approximate Bayesian Evaluation of Multiple Treatment Effects, Biometrics 56: 213-219, 2000

Decision-Theoretic Designs for Pre-Phase II Screening Trials in Oncology, Biometrics 2001

Dose-Finding Based on Feasibility and Toxicity in T-cell Infusion Trials, Biometrics 2001

Selecting therapeutic strategies based on efficacy and death in multi-course clinical trials.  JASA, 97:29-39, 2002

Ethical issues in oncology biostatistics.  Stat Meth in Medical Res, 11:429-448, 2002

Seamlessly Expanding a Randomized Phase II Trial to Phase III, Biometrics, 2002

Adaptive Decision Making in a Lymphocyte Infusion Trial, Biometrics, 2002

Monitoring the Rates of Composite Events with Censored Data in Phase II Clinical Trials, Biometrics, 2002

Busulfan systemic exposure relative to regimen-related toxicity and acute graft vs. host disease; defining a therapeutic window for IVBuCy2 in chronic myelogenous leukemia.  Biology of Blood and Marrow Transplantation, 8:477-485, 2002

Hierarchical Bayesian approaches to phase II trials in diseases with multiple subtypes.  Statistics in Medicine, 22: 763-780, 2003

Dose-finding with two agents in phase I oncology trials.  Biometrics, 59:487-496, 2003

Comparison of 100-day mortality rates associated with IV busulfan and cyclophosphamide versus other preparative regimens in allogeneic bone marrow transplant for chronic myelogenous leukemia: Bayesian sensitivity analyses of confounded treatment and center effects. Bone Marrow Transplantation, 33: 1191-1199, 2004.

Dose-finding based on multiple toxicities in a soft tissue sarcoma trial.  JASA, 99:26-35, 2004

Dose-finding based on efficacy-toxicity trade-offs. Biometrics, 60:684-693, 2004.

Determining a Maximum-Tolerated Schedule of a Cytotoxic Agent, Biometrics 61, 335–343, 2005

Monitoring event times in early phase clinical trials: some practical issues. Clinical Trials. 2:467-478, 2005

Adaptive dose selection using efficacy-toxicity trade-offs: illustrations and practical considerations. J Biopharm Stat 16:623-638, 2006

A Geometric Approach to Comparing Treatments for Rapidly Fatal Diseases, Biometrics 2006

Continuous Bayesian adaptive randomization based on event times with covariates.  Stat in Med, 25:55-70, 2006

Simultaneously optimizing dose and schedule of a new cytotoxic agent. Clinical Trials, 4:113-124, 2007

Bayesian and frequentist two-stage treatment strategies based on sequential failure times subject to interval censoring. Stat in Med  26:4687-4702, 2007.

Adaptive Therapy for Androgen-Independent Prostate Cancer: A Randomized Selection Trial of Four Regimens, JNCI, 99:1613-1622, 2007

Phase I/II study of gemtuzumab ozogamicin added to fludarabine, melphalan and allogeneic hematopoietic stem cell transplantation for high-risk CD33 positive myeloid leukemias and myelodysplastic syndrome.  Leukemia. 22:258-264, 2008

Once daily IV busulfan and fludarabine (IV Bu-Flu) compares favorably with IV busulfan and cyclophosphamide (IV BuCy2) as pretransplant conditioning therapy in AML/MDS. Biology of Blood and Marrow Transplantation. 14:672-684, 2008

A review of phase 2-3 clinical trial designs. Lifetime Data Analysis. 14:37-53, 2008

Determining the effective sample size of a parametric prior.  Biometrics. 64:595-602, 2008.

Accounting for patient heterogeneity in phase II clinical trials. Stat in Med  27:2802-2815, 2008

Monitoring late onset toxicities in phase I trials using predicted risks.  Biostatistics.  9:442-457, 2008

Patient-Specific Dose Finding Based on Bivariate Outcomes and Covariates, Biometrics, 2008

Bayesian adaptive model selection for optimizing group sequential clinical trials. Stat in Med,  27:5586-5604, 2008.

Bayesian hierarchical mixture model for platelet derived growth factor receptor phosphorylation to improve estimation  of progression-free survival in prostate cancer. JRSS-C, 59:19-34, 2010.

Utility-based optimization of combination therapy using ordinal toxicity and efficacy in phase I/II trials. Biometrics. 66:532-540, 2010

Evaluating the impact of prior assumptions in Bayesian biostatistics.  Statistics in Biosciences. 2:1-17, 2010

Bayesian models and decision algorithms for complex early phase clinical trials. Statistical Science. 25:227-244, 2010

Defining and ranking effects of individual agents based on survival times of cancer patients treated with combination chemotherapies. Stat in Med, 30:1777-1794, 2011.

Optimizing the concentration and bolus of a drug delivered by continuous infusion. Biometrics. 67:1638-1646, 2011

A hybrid geometric phase II-III clinical trial design based on treatment failure time and toxicity. J Stat Planning and Inf. 142:944-955, 2012

Estimating progression-free survival in pediatric brain tumor patients when some progression statuses are unknown. JRSS-C, 61:135-149, 2012

Evaluation of viable dynamic treatment regimes in a sequentially randomized trial of advanced prostate cancer (with discussion). JASA.  107:493-508, (with discussion, pages 509-517; rejoinder, pages 518-520), 2012.

Adaptive randomization to improve utility-based dose-finding with bivariate ordinal outcomes. J Biopharmaceutical Stat. 22:785-801, 2012.

Prior effective sample size in conditionally independent hierarchical models. Bayesian Analysis. 7:591-614, 2012.

Bayesian adaptive dose-finding based on efficacy and toxicity. J. Statistical Research. 14:187-202, 2012.

Evaluating joint effects of induction-salvage treatment regimes on overall survival in acute leukemia.  JRSS-C. 62:67-83, 2013.


Using joint utilities of the times to response and toxicity to adaptively optimize schedule-dose regimes. Biometrics 69:673-682, 2013.


Using data augmentation to facilitate conduct of phase I/II clinical trials with delayed outcomes.  J American Statistical Assoc.  109:525-536, 2014.


Optimizing sedative dose in preterm infants undergoing treatment for respiratory distress syndrome. J American Statistical Assoc.   109:931-943, 2014.


Using effective sample size for prior calibration in Bayesian phase I-II dose-finding.  Clinical Trials. 11:657-666, 2014.

A modified adaptive Lasso for identifying interactions in the Cox model with the heredity constraint. Statistics and Probability Letters.  93:126-133, 2014.

Optimization of multi-stage dynamic treatment regimes utilizing accumulated data. Statistics in Medicine, 34:3424-3443, 2015

SMART design, conduct, and analysis in oncology. In Dynamic Treatment Regimes in Practice: Planning Trials and Analyzing Data for Personalized Medicine. E Moodie and M. Kosorok (eds), SIAM. In press.

Some caveats for outcome adaptive randomization in clinical trials. In Modern Adaptive Randomized Clinical Trials: Statistical, Operational, and Regulatory Aspects. A. Sverdlov (ed). Taylor & Francis. 2015.

Bayesian dose-finding in two treatment cycles based on the joint utility of efficacy and toxicity. J American Statistical Assoc.  110:711-722, 2015.

Bayesian nonparametric estimation of targeted agent effects on biomarker change to predict clinical outcome. Biometrics. 71:188-197, 2015.

Comparing radiation modalities for esophageal cancer based on total toxicity burden and disease-free survival. J Royal Statistical Society, Series C (Applied Statistics) 65:273-297, 2016.

A Decision Theoretic Phase I-II Design for Ordinal Outcomes in Two Cycles. Biostatistics   17:304-319, 2016.

Bayesian nonparametric estimation for dynamic treatment regimes with sequential transition times J. American Statistical Assoc. with discussion) 111:921-950, 2016.


Utility-based designs for randomized comparative trials with discrete outcomes. Statistics in Medicine.  35:4285-4305, 2016.


Robust treatment comparison based on utilities of semi-competing risks in non-small-cell lung cancer. J American Statistical Assoc. 112:11-23, 2017.


A decision-theoretic comparison of treatments to resolve air leaks after lung surgery based on nonparametric modeling. Bayesian Analysis. 12(3):639-652, 2017.


Parametric dose standardization for two-agent combinations in a phase I-II trial with ordinal outcomes. J Royal Statistical Society, Series C (Applied Statistics). 66:201-224, 2017.


A simulation study of methods for selecting subgroup-specific doses in phase I trials.  Pharmaceutical Statistics. 16:143-156, 2017.  


Bayesian variable selection for a semi-competing risks model with three hazard functions. Computational Statistics and Data Analysis. 112:170-185, 2017.


Clinical trial design as a decision problem.  Applied Stochastic Models in Business and Industry.  (Invited, special issue in honor of Kathryn Chaloner) 33:296-301, 2017.


A simulation study of outcome adaptive randomization in multi-arm clinical trials. Clinical Trials. 14:432-440, 2017.


Bayesian nonparametric statistics:  A new toolkit for discovery in cancer research.    Pharmaceutical Statistics.  16:414-423, 2017.


A Bayesian machine learning method for optimizing dynamic treatment regimes. Revised for J American Statistical Assoc  In press.

Bayesian treatment comparison using parametric mixture priors computed from elicited histograms Statistical Methods in Medical Research. In press.


A utility-based design for randomized comparative trials with ordinal outcomes and prognostic subgroups. Biometrics.  In press.

Optimizing natural killer cell doses for heterogeneous cancer patients based on multiple event times. J Royal Statistical Society, Series C (Applied Statistics). In press.