• Researcher Profile

    Lee-Jen Wei, PhD

    Lee-Jen Wei, PhD
    Professor of Biostatistics, Harvard T.H. Chan School of Public Health

    Office phone: 617-432-2826
    Fax: 617-432-5619
    Email: wei@hsph.harvard.edu

    Preferred contact method: office phone

    Research Department

    Biostatistics and Computational Biology

    Area of Research

    Developing Quantitative Methods for Designing, Monitoring, and Analyzing Clinical Studies

    Dana-Farber Cancer Institute
    450 Brookline Avenue
    Mayer 232
    Boston, MA 02215


    Dr. Wei received his PhD in statistics from the University of Wisconsin-Madison in 1975. Since then, he has held academic appointments at the University of South Carolina, George Washington University, University of Michigan, University of Wisconsin, and Harvard University. He has also worked at the National Cancer Institute, and served as associate editor for the Journal of the American Statistical Association, Biometrics, and Communications in Statistics. Currently, he is director of the Bioinformatics Core of the Harvard School of Public Health.

    Recent Awards

    • Greenberg Distinguished Lectureship, University of North Carolina, Chapel Hill, 2001
    • Distinguished Alumni Award, Fu Jen University, 1999


    Developing Quantitative Methods for Designing, Monitoring, and Analyzing Clinical Studies

    Our research focuses on developing statistical methods for the design and analysis of clinical trials. In the late 1970s, Dr. Wei introduced the "urn design" for two-arm sequential clinical studies, a design that has since been used in several large-scale multicenter trials. Later he proposed a response adaptive design, currently utilized in several trials sponsored by private industry, to ameliorate problems arising from the conventional rule of 50-50 randomization treatment allocation in clinical studies. In the 1980s, he presented a monitoring scheme for the interim analysis for clinical trial data.

    We have also developed numerous methods for analyzing data with multiple outcomes or repeated measurements from study subjects, in particular, "multivariate Cox procedures" for handling multiple event times, as well as alternatives to the Cox proportional hazards model for analyzing survival observations.

    A critical issue in statistical inference is ascertaining whether the model used for the data is appropriate. Currently, we are developing graphical and numerical methods for checking the adequacy of the Cox proportional hazards model, other semi-parametric survival models, parametric models, and random effects models for repeated measurements. The new procedures are much less subjective than conventional methods based on ordinary residuals plots.

    Since the cost of computing has dropped, some intractable statistical problems can now be handled numerically. Presently, we are working on various resampling methods for quantile regression, rank regression, and regression models for censored data.

View Physician Directory

Find a Clinical Trial

Support Cancer

Give Now