## Course Website Locator: bio249-01

Harvard School of Public Health

The following course websites match your request:

### Spring 2009

#### Bayesian Methods in Biostatistics

Dr. C. Paciorek

5 credits

Lectures. Two 2-hour sessions each week.

General principles of the Bayesian approach, prior distributions, hierarchial models and modeling techniques, approximate inference, Markov chain Monte Carlo methods, model assessment and comparison. Bayesian approaches to GLMMs, multiple testing, nonparametrics, clinical trails, survival analysis.

BIO230 (Probability Theory and Applications I), BIO231 (Statistical Inference I), and BIO232 (Methods I), or the signature of the instructor is required. BIO233 (Methods II) will also be helpful for the second part of the course.

Course evaluations are an important method for feedback on the quality of course offerings. The submission of a course evaluation is a requirement for this course. Your grade for the course will be made available only after you have submitted responses to at least the first three questions of the on-line evaluation for this course.

### Fall 2007

#### Bayesian Methods in Biostatistics

Dr. C. Paciorek

5 credits

Lectures. Two 2-hour sessions each week.

Bayes' theorem is a trivial identity in probability theory, but it has profound consequences for the theory and practice of statistical inference and in decision theory. The course will introduce you to the Bayesian approach and philosophy and give you experience in applying the Bayesian approach to statistical inference with an emphasis on real data analysis. During the first two-thirds of the course, we will focus on the core of Bayesian statistics, with topics including Bayes' theorem, general principles (likelihood, exchange-ability, de Finetti's theorem), prior distributions, simple models, hierarchical models, methods of inference (exact, approximations, Monte Carlo strategies), Markov chain Monte Carlo (MCMC) computation, model diagnostics and model selection. The second part of the course will introduce the Bayesian approach to a range of important biostatistical models and situations including GLMs and GLMMs, high-dimensional data, clinical trials and meta-analysis, nonparametrics, survival analysis, missing data, and multiple testing.

Per the title, the course is focused on methodology; some limited theory will be covered and I will present some of the material in the context of real data, with an occasional case study. Bayesian inference relies on the posterior distribution, derived through integration, which can be challenging in complicated (and even simple) models. Since 1990 advances in computational techniques have allowed statisticians to simulate from the posterior distribution; so-called Markov chain Monte Carlo techniques are now the standard tool for fitting Bayesian models. As a result, the course will be heavily computational and there will be extensive discussion of computation, as well as a heavy dose of computing in the homework. Students are expected to be familiar with R and will be introduced to WinBUGS as well.

BIO230 (Probability Theory and Applications I), BIO231 (Statistical Inference I), and BIO232 (Methods I), or the signature of the instructor is required. BIO233 (Methods II) will also be helpful for the second part of the course.

Course evaluations are an important method for feedback on the quality of course offerings. The submission of a course evaluation is a requirement for this course. Your grade for the course will be made available only after you have submitted responses to at least the first three questions of the on-line evaluation for this course.