A0852
Title: Semiparametric data fusion with external summary statistics
Authors: Wang Miao - Peking University (China) [presenting]
Abstract: Using external summary statistics to improve the efficiency of internal data analysis has attracted a lot of attention in recent years. Efficient estimation using external summary statistics has been developed for parametric models. We establish the theory for efficient estimation of a general functional under semiparametric or nonparametric models when combining internal individual-level data with external summary statistics. We propose an estimator that can achieve this efficiency bound. The estimator is also robust against the bias of external summary statistics when they are obtained with biased data whose distribution departs from the internal data. We illustrate with simulations and apply our methods to a Helicobacter pylori infection dataset.
