A0731
Title: Data harmonization via regularized nonparametric mixing distribution estimation
Authors: Steven Wilkins-Reeves - Univeristy of Washington (Austria)
Yen-Chi Chen - University of Washington (United States)
Kwun Chuen Gary Chan - University of Washington (United States) [presenting]
Abstract: Data harmonization is the process of developing an equivalence between two measurements of a common domain. The problem is motivated by dementia research in which multiple neuropsychological tests have been used in practice to measure the same underlying cognitive ability, such as memory or attention. This statistical problem is connected to mixing distribution estimation, which is common in empirical Bayes approaches. A nonparametric latent trait model is introduced and studied; a method is developed that enforces the uniqueness of the regularized maximum likelihood estimator, showing how a nonparametric EM algorithm will converge weakly to its maximizer and its superior computational efficiency to off-the-shelf solvers is illustrated. The method is applied to the National Alzheimer's Coordination Center uniform data set, and it is shown that the method can be used to convert between score measurements and account for the measurement error. It is shown that this method outperforms standard techniques commonly used in dementia research.
