On estimation of covariance function for functional data with detection limits

Article Statistics & Probability

Fast estimators for the mean function for functional data with detection limits

Haiyan Liu et al.

Summary: In many studies on disease progression, the restricted detection limits of biomarkers lead to informative missing values. Current approaches either ignore this problem or use a global approach for estimating the mean function, which is time-consuming and may not be realistic. In this study, we propose novel estimators for the mean function in the presence of detection limits for both unbalanced sparse and dense data. We derive the asymptotic properties of the estimators and compare them with existing methods through simulations and a real data application. Our methods appear to perform well, particularly for dense data where they are computationally faster than existing methods.

STAT (2022)

添加到收藏夹

Article Mathematical & Computational Biology

Functional principal component analysis for longitudinal data with informative dropout

Haolun Shi et al.

Summary: This study introduces a method for informatively missing longitudinal data called informatively missing functional principal component analysis (imFunPCA), which incorporates information from both observed and missing data points based on the likelihood of the data. The approach uses a regression-based orthogonal approximation method for computing the functional principal components. Simulation studies demonstrate the superior performance of the proposed method compared to conventional approaches in the presence of informative missingness.

STATISTICS IN MEDICINE (2021)

添加到收藏夹

Article Statistics & Probability