Advanced Topics in Case-Control and Cohort Studies

EPI 215 Subject & Catalog Number

Description

This topics course explores foundational epidemiologic concepts and statistical methods for predictive model building, prediction evaluation, and practical implications of clinical risk assessment tool implementation, along with advanced topics including missing data, multiple testing, and reproducible research. Concepts in predictive model building are explored by extension of standard linear, logistic and Cox regression modeling frameworks to high-dimensional data via stepwise and penalized regression methods. Principles of prediction validation are introduced including sensitivity/specificity, calibration and discrimination, and validation methods such as cross-validation and bootstrapping. Other topics covered include methods for missing data such as multiple imputation, multiplicity correction methods for high dimensional clinical research, and principles of reproducible research. Throughout, emphasis will be on applications and interpretation of results, with limited introduction to theory that underlies these techniques. Familiarity with R or SAS is expected.