For multi-source data, blocks of variable information from certain sources are likely missing. Existing methods for handling missing data do not take structures of block-wise missing data into consideration. In this paper, we propose a Multiple Block-wise Imputation (MBI) approach, which incorporates imputations based on both complete and incomplete observations. Specifically, for a given missing pattern group, the imputations in MBI incorporate more samples from groups with fewer observed variables in addition to the group with complete observations. We propose to construct estimating equations based on all available information, and integrate informative estimating functions to achieve efficient estimators. We show that the proposed method has estimation and model selection consistency under both fixed-dimensional and high-dimensional settings. Moreover, the proposed estimator is asymptotically more efficient than the estimator based on a single imputation from complete observations only. In addition, the proposed method is not restricted to missing completely at random. Numerical studies and ADNI data application confirm that the proposed method outperforms existing variable selection methods under various missing mechanisms.
23 January 2020 in Journal of the American Statistical Association
In this paper we propose a heterogeneous modeling framework which achieves individual-wise feature selection and heterogeneous covariates’ effects subgrouping simultaneously. In contrast to conventional model selection approaches, the new approach constructs a separation penalty with multi-directional shrinkages, which facilitates individualized modeling to distinguish strong signals from noisy ones and selects different relevant variables for different individuals. Meanwhile, the proposed model identifies subgroups among which individuals share similar covariates’ effects, and thus improves individualized estimation efficiency and feature selection accuracy. Moreover, the proposed model also incorporates within-individual correlation for longitudinal data to gain extra efficiency. We provide a general theoretical foundation under a double-divergence modeling framework where the number of individuals and the number of individual-wise measurements can both diverge, which enables inference on both an individual level and a population level. In particular, we establish a strong oracle property for the individualized estimator to ensure its optimal large sample property under various conditions. An efficient ADMM algorithm is developed for computational scalability. Simulation studies and applications to post-trauma mental disorder analysis with genetic variation and an HIV longitudinal treatment study are illustrated to compare the new approach to existing methods.
14 November 2022 in Journal of the American Statistical Association
Recent advances in mobile health (mHealth) technology provide an effective way to monitor individuals’ health statuses and deliver just-in-time personalized interventions. However, the practical use of mHealth technology raises unique challenges to existing methodologies on learning an optimal dynamic treatment regime. Many mHealth applications involve decision-making with large numbers of intervention options and under an infinite time horizon setting where the number of decision stages diverges to infinity. In addition, temporary medication shortages may cause optimal treatments to be unavailable, while it is unclear what alternatives can be used. To address these challenges, we propose a Proximal Temporal consistency Learning (pT-Learning) framework to estimate an optimal regime that is adaptively adjusted between deterministic and stochastic sparse policy models. The resulting minimax estimator avoids the double sampling issue in the existing algorithms. It can be further simplified and can easily incorporate off-policy data without mismatched distribution corrections. We study theoretical properties of the sparse policy and establish finite-sample bounds on the excess risk and performance error. The proposed method is provided in our proximalDTR package and is evaluated through extensive simulation studies and the OhioT1DM mHealth dataset.
29 June 2017 in Journal of the American Statistical Association
In recent years, there has been a growing demand to develop efficient recommender systems which track users’ preferences and recommend potential items of interest to users. In this article, we propose a group-specific method to use dependency information from users and items which share similar characteristics under the singular value decomposition framework. The new approach is effective for the “cold-start” problem, where, in the testing set, majority responses are obtained from new users or for new items, and their preference information is not available from the training set. One advantage of the proposed model is that we are able to incorporate information from the missing mechanism and group-specific features through clustering based on the numbers of ratings from each user and other variables associated with missing patterns. In addition, since this type of data involves large-scale customer records, traditional algorithms are not computationally scalable. To implement the proposed method, we propose a new algorithm that embeds a back-fitting algorithm into alternating least squares, which avoids large matrices operation and big memory storage, and therefore makes it feasible to achieve scalable computing. Our simulation studies and MovieLens data analysis both indicate that the proposed group-specific method improves prediction accuracy significantly compared to existing competitive recommender system approaches. Supplementary materials for this article are available online.
21 May 2019 in Journal of the American Statistical Association
This work is motivated by multimodality breast cancer imaging data, which is quite challenging in that the signals of discrete tumor-associated microvesicles (TMVs) are randomly distributed with heterogeneous patterns. This imposes a significant challenge for conventional imaging regression and dimension reduction models assuming a homogeneous feature structure. We develop an innovative multilayer tensor learning method to incorporate heterogeneity to a higher-order tensor decomposition and predict disease status effectively through utilizing subject-wise imaging features and multimodality information. Specifically, we construct a multilayer decomposition which leverages an individualized imaging layer in addition to a modality-specific tensor structure. One major advantage of our approach is that we are able to efficiently capture the heterogeneous spatial features of signals that are not characterized by a population structure as well as integrating multimodality information simultaneously. To achieve scalable computing, we develop a new bi-level block improvement algorithm. In theory, we investigate both the algorithm convergence property, tensor signal recovery error bound and asymptotic consistency for prediction model estimation. We also apply the proposed method for simulated and human breast cancer imaging data. Numerical results demonstrate that the proposed method outperforms other existing competing methods.
11 July 2022 in Journal of the American Statistical Association
DNA methylation (DNAm) has been suggested to play a critical role in post-traumatic stress disorder (PTSD), through mediating the relationship between trauma and PTSD. However, this underlying mechanism of PTSD for African Americans still remains unknown. To fill this gap, in this article, we investigate how DNAm mediates the effects of traumatic experiences on PTSD symptoms in the Detroit Neighborhood Health Study (DNHS) (2008–2013) which involves primarily African Americans adults. To achieve this, we develop a new mediation analysis approach for high-dimensional potential DNAm mediators. A key novelty of our method is that we consider heterogeneity in mediation effects across subpopulations. Specifically, mediators in different subpopulations could have opposite effects on the outcome, and thus could be difficult to identify under a traditional homogeneous model framework. In contrast, the proposed method can estimate heterogeneous mediation effects and identifies subpopulations in which individuals share similar effects. Simulation studies demonstrate that the proposed method outperforms existing methods for both homogeneous and heterogeneous data. We also present our mediation analysis results of a dataset with 125 participants and more than CpG sites from the DNHS study. The proposed method finds three subgroups of subjects and identifies DNAm mediators corresponding to genes such as HSP90AA1 and NFATC1 which have been linked to PTSD symptoms in literature. Our finding could be useful in future finer-grained investigation of PTSD mechanism and in the development of new treatments for PTSD. Supplementary materials for this article are available online.