Dataset Information

Selection of latent variables for multiple mixed-outcome models.

ABSTRACT: Latent variable models have been widely used for modeling the dependence structure of multiple outcomes data. However, the formulation of a latent variable model is often unknown a priori, the misspecification will distort the dependence structure and lead to unreliable model inference. Moreover, multiple outcomes with varying types present enormous analytical challenges. In this paper, we present a class of general latent variable models that can accommodate mixed types of outcomes. We propose a novel selection approach that simultaneously selects latent variables and estimates parameters. We show that the proposed estimator is consistent, asymptotically normal and has the oracle property. The practical utility of the methods is confirmed via simulations as well as an application to the analysis of the World Values Survey, a global research project that explores peoples' values and beliefs and the social and personal characteristics that might influence them.

SUBMITTER: Zhou L

PROVIDER: S-EPMC5026194 | biostudies-literature | 2014 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Selection of latent variables for multiple mixed-outcome models.

Zhou Ling L Lin Huazhen H Song Xinyuan X Li Y I YI

Scandinavian journal of statistics, theory and applications 20140402 4

Latent variable models have been widely used for modeling the dependence structure of multiple outcomes data. However, the formulation of a latent variable model is often unknown <i>a priori</i>, the misspecification will distort the dependence structure and lead to unreliable model inference. Moreover, multiple outcomes with varying types present enormous analytical challenges. In this paper, we present a class of general latent variable models that can accommodate mixed types of outcomes. We p ...[more]

PMID: 27642219

Similar Datasets

Project description:BackgroundTraumatic brain injuries (TBI) are associated with high risk of morbidity and mortality. Early outcome prediction in patients with TBI require reliable data input and stable prognostic models. The aim of this investigation was to analyze different CT classification systems and prognostic calculators in a representative population of TBI-patients, with known outcomes, in a neurointensive care unit (NICU), to identify the most suitable CT scoring system for continued research.Materials and methodsWe retrospectively included 158 consecutive patients with TBI admitted to the NICU at a level 1 trauma center in Sweden from 2012 to 2016. Baseline data on admission was recorded, CT scans were reviewed, and patient outcome one year after trauma was assessed according to Glasgow Outcome Scale (GOS). The Marshall classification, Rotterdam scoring system, Helsinki CT score and Stockholm CT score were tested, in addition to the IMPACT and CRASH prognostic calculators. The results were then compared with the actual outcomes.ResultsGlasgow Coma Scale score on admission was 3-8 in 38%, 9-13 in 27.2%, and 14-15 in 34.8% of the patients. GOS after one year showed good recovery in 15.8%, moderate disability in 27.2%, severe disability in 24.7%, vegetative state in 1.3% and death in 29.7%. When adding the variables from the IMPACT base model to the CT scoring systems, the Stockholm CT score yielded the strongest relationship to actual outcome. The results from the prognostic calculators IMPACT and CRASH were divided into two subgroups of mortality (percentages); ≤50% (favorable outcome) and > 50% (unfavorable outcome). This yielded favorable IMPACT and CRASH scores in 54.4 and 38.0% respectively.ConclusionThe Stockholm CT score and the Helsinki score yielded the closest relationship between the models and the actual outcomes in this consecutive patient series, representative of a NICU TBI-population. Furthermore, the Stockholm CT score yielded the strongest overall relationship when adding variables from the IMPACT base model and would be our method of choice for continued research when using any of the current available CT score models.

Dataset Information

Selection of latent variables for multiple mixed-outcome models.

Publications

Selection of latent variables for multiple mixed-outcome models.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets