Targeted Data Fusion for Region-Specific Survival Effects in the AMP HIV Prevention Trials
Researchers have developed a novel federated learning approach to analyze HIV prevention trial data across different regions without sharing individual-level information. This method allows for the estimation of region-specific survival curves and causal contrasts, addressing heterogeneity in treatment efficacy observed in the Antibody Mediated Prevention (AMP) trials. The approach enhances privacy and improves inference precision by downweighting data sources not aligned with the target population. AI
IMPACT Introduces a privacy-preserving method for analyzing distributed datasets, potentially applicable to other sensitive health research.