Researchers have developed a new nonparametric estimator designed to de-bias distribution estimation for survival data that has undergone local differential privacy (LDP). This method addresses the utility loss and statistical inference issues caused by the perturbation inherent in LDP, particularly for right-censored survival data. The proposed estimator simultaneously corrects for noise introduced in both event times and event indicators, theoretically establishing uniform consistency and asymptotic normality. Its performance was demonstrated through simulations and analysis of the METABRIC breast cancer cohort, showing effectiveness across various privacy budgets, censoring levels, and sample sizes. AI
IMPACT Enhances privacy-preserving techniques for sensitive clinical data analysis, potentially enabling more robust research with protected datasets.
RANK_REASON The cluster contains an academic paper detailing a new statistical methodology. [lever_c_demoted from research: ic=1 ai=0.7]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
