Estimating Covariate-adjusted Survival Curve in Distributed Data Environment using Data Collaboration Quasi-Experiment

Abstract: In recent years, there has been an increasing demand for privacy-preserving survival analysis using integrated observational data from multiple institutions and data sources. In particular, estimating survival curves adjusted for covariates that account for confounding factors is essential for evaluating the effectiveness of medical treatments. While high-precision estimation of survival curves requires the collection of large amounts of individual-level data, sharing such data is challenging due to privacy concerns and, even if sharing were possible, the communication costs between institutions would be enormous. To address these challenges, this study proposes and evaluates a novel method that leverages an extended data collaboration quasi-experiment, to estimate covariate-adjusted survival curves.