Data Fusion for High-Resolution Estimation
Researchers have developed a novel data fusion method to improve the accuracy of high-resolution population health estimates. This technique combines unbiased, low-resolution data, such as aggregated administrative records, with potentially biased, high-resolution data from sources like online surveys. The proposed approach learns a distribution that is both consistent with the aggregated data and a model of sampling bias present in the high-resolution source, significantly reducing estimation bias compared to methods using single data sources. AI