Abstract

Population-based cancer registry databases are critical resources to bridge the information gap that results from a lack of sufficient statistical power from primary cohort data with small to moderate sample size. This presentation focuses on the strategic use of cancer registry data (i.e., source cohort) to augment the precision of individualized risk estimates for patients with inflammatory breast cancer (IBC) at The University of Texas MD Anderson Cancer Center (i.e., target cohort). Challenges, including potentially incompatible external aggregate information and unignorable uncertainty in external data, , will be discussed and addressed. Our proposed methods yield more precise individualized risk estimation compared to relying solely on the target cohort, and meanwhile exhibit robustness against cohort heterogeneity, outperforming methods that directly combine target and source data.