A Novel Approach for Data Integration with Multiple Heterogeneous Data Sources
By: Farimah Shamsi, Andriy Derkach
The integration of data from multiple sources is increasingly used to achieve larger sample sizes and enhance population diversity. Our previous work established that, under random sampling from the same underlying population, integrating large incomplete datasets with summary-level data produces unbiased parameter estimates. In this study, we develop a novel statistical framework that enables the integration of summary-level data with information from heterogeneous data sources by leveraging auxiliary information. The proposed approach estimates study-specific sampling weights using this auxiliary information and calibrates the estimating equations to obtain the full set of model parameters. We evaluate the performance of the proposed method through simulation studies under various sampling designs and illustrate its application by reanalyzing U.S. cancer registry data combined with summary-level odds ratio estimates for selected colorectal cancer (CRC) risk factors, while relaxing the random sampling assumption.
Similar Papers
Data Integration With Biased Summary Data via Generalized Entropy Balancing
Methodology
Helps studies use outside info without errors.
A Communication-Efficient Distributed Algorithm for Learning with Heterogeneous and Structurally Incomplete Multi-Site Data
Methodology
Helps doctors share patient data safely for better cures.
Improving prediction in M-estimation by integrating external information from heterogeneous populations
Methodology
Helps make predictions better using outside information.