Dataset Information

Multimodel inference for biomarker development: an application to schizophrenia.

ABSTRACT: In the present study, to improve the predictive performance of a model and its reproducibility when applied to an independent data set, we investigated the use of multimodel inference to predict the probability of having a complex psychiatric disorder. We formed training and test sets using proteomic data (147 peptides from 77 proteins) from two-independent collections of first-onset drug-naive schizophrenia patients and controls. A set of prediction models was produced by applying lasso regression with repeated tenfold cross-validation to the training set. We used feature extraction and model averaging across the set of models to form two prediction models. The resulting models clearly demonstrated the utility of a multimodel based approach to make good (training set AUC?>?0.80) and reproducible predictions (test set AUC?>?0.80) for the probability of having schizophrenia. Moreover, we identified four proteins (five peptides) whose effect on the probability of having schizophrenia was modified by sex, one of which was a novel potential biomarker of schizophrenia, foetal haemoglobin. The evidence of effect modification suggests that future schizophrenia studies should be conducted in males and females separately. Future biomarker studies should consider adopting a multimodel approach and going beyond the main effects of features.

SUBMITTER: Cooper JD

PROVIDER: S-EPMC6370882 | biostudies-literature | 2019 Feb

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Multimodel inference for biomarker development: an application to schizophrenia.

Cooper Jason D JD Han Sung Yeon Sarah SYS Tomasik Jakub J Ozcan Sureyya S Rustogi Nitin N van Beveren Nico J M NJM Leweke F Markus FM Bahn Sabine S

Translational psychiatry 20190211 1

In the present study, to improve the predictive performance of a model and its reproducibility when applied to an independent data set, we investigated the use of multimodel inference to predict the probability of having a complex psychiatric disorder. We formed training and test sets using proteomic data (147 peptides from 77 proteins) from two-independent collections of first-onset drug-naive schizophrenia patients and controls. A set of prediction models was produced by applying lasso regress ...[more]

PMID: 30745560

Similar Datasets

Project description:BackgroundArboviral diseases are major global public health threats. Yet, our understanding of infection risk factors is, with a few exceptions, considerably limited. A crucial shortcoming is the widespread use of analytical methods generally not suited for observational data--particularly null hypothesis-testing (NHT) and step-wise regression (SWR). Using Mayaro virus (MAYV) as a case study, here we compare information theory-based multimodel inference (MMI) with conventional analyses for arboviral infection risk factor assessment.Methodology/principal findingsA cross-sectional survey of anti-MAYV antibodies revealed 44% prevalence (n?=?270 subjects) in a central Amazon rural settlement. NHT suggested that residents of village-like household clusters and those using closed toilet/latrines were at higher risk, while living in non-village-like areas, using bednets, and owning fowl, pigs or dogs were protective. The "minimum adequate" SWR model retained only residence area and bednet use. Using MMI, we identified relevant covariates, quantified their relative importance, and estimated effect-sizes (? ± SE) on which to base inference. Residence area (?(Village)? = ?2.93 ± 0.41; ?(Upland)?=?-0.56 ± 0.33, ?(Riverbanks)? = ?-2.37 ± 0.55) and bednet use (??=?-0.95 ± 0.28) were the most important factors, followed by crop-plot ownership (?? = ?0.39 ± 0.22) and regular use of a closed toilet/latrine (??=?0.19 ± 0.13); domestic animals had insignificant protective effects and were relatively unimportant. The SWR model ranked fifth among the 128 models in the final MMI set.Conclusions/significanceOur analyses illustrate how MMI can enhance inference on infection risk factors when compared with NHT or SWR. MMI indicates that forest crop-plot workers are likely exposed to typical MAYV cycles maintained by diurnal, forest dwelling vectors; however, MAYV might also be circulating in nocturnal, domestic-peridomestic cycles in village-like areas. This suggests either a vector shift (synanthropic mosquitoes vectoring MAYV) or a habitat/habits shift (classical MAYV vectors adapting to densely populated landscapes and nocturnal biting); any such ecological/adaptive novelty could increase the likelihood of MAYV emergence in Amazonia.

Project description:ObjectiveTo explore and apply multimodel inference to test the relative contributions of latent genetic, environmental and direct causal factors to the covariation between two variables with data from the classical twin design by estimating model-averaged parameters.MethodsBehavior genetics is concerned with understanding the causes of variation in phenotypes and the causes of covariation between two or more phenotypes. Two variables may correlate as a result of genetic, shared environmental or unique environmental factors contributing to variation in both variables. Two variables may also correlate because one or both directly cause variation in the other. Furthermore, covariation may result from any combination of these sources, leading to 25 different identified structural equation models. OpenMx was used to fit all these models to account for covariation between two variables collected in twins. Multimodel inference and model averaging were used to summarize the key sources of covariation, and estimate the magnitude of these causes of covariance. Extensions of these models to test heterogeneity by sex are discussed.ResultsWe illustrate the application of multimodel inference by fitting a comprehensive set of bivariate models to twin data from the Virginia Twin Study of Psychiatric and Substance Use Disorders. Analyses of body mass index and tobacco consumption data show sufficient power to reject distinct models, and to estimate the contribution of each of the five potential sources of covariation, irrespective of selecting the best fitting model. Discrimination between models on sample size, type of variable (continuous versus binary or ordinal measures) and the effect size of sources of variance and covariance.ConclusionsWe introduce multimodel inference and model averaging approaches to the behavior genetics community, in the context of testing models for the causes of covariation between traits in term of genetic, environmental and causal explanations.

Project description:6-min walk tests (6MWT) are routinely performed in patients with chronic obstructive pulmonary disease (COPD). Oxygen uptake ([Formula: see text]) kinetics during 6MWT can be modeled and derived parameters provide indicators of patients' exercise capacity. Post-exercise [Formula: see text] recovery also provides important parameters of patients' fitness which has not been extensively investigated in COPD. Several nonlinear regression models with different underlying biological assumptions may be suitable for describing recovery kinetics. Multimodel inference (model averaging) can then be used to capture the uncertainty in considering several models. Our aim was to apply multimodel inference in order to better understand the physiological underpinnings of [Formula: see text] recovery after 6MWT in patients with COPD. 61 patients with COPD (stages 2 to 4) were included in this study. Oxygen kinetics during 6MWT were modeled using nonlinear regression. Three statistical approaches (mixed-effects, meta-analysis and weighted regression) were compared in order to summarize estimates obtained from multiple kinetics. The recovery phase was modeled using 3 distinct equations (log-logistic, Weibull 1 and Weibull 2). Three models were fitted to the set of 61 kinetics. A significant model-averaged difference of 40.39 sec (SE = 17.1) in the time to half decrease of [Formula: see text] level ([Formula: see text]) was found between stage 2 and 4 (p = 0.0178). In addition, the Weibull 1 model characterized by a steeper decrease at the beginning of the recovery phase showed some improvement of goodness of fit when fitted to the kinetics of patients with stage 2 COPD in comparison with the 2 other models. Multimodel inference was successfully used to model [Formula: see text] recovery after 6MWT in patients with COPD. Significant model-averaged differences in [Formula: see text] were found between moderate and very severe COPD patients. Furthermore, specific patterns of [Formula: see text] recovery could be identified across COPD stages.

Dataset Information

Multimodel inference for biomarker development: an application to schizophrenia.

Publications

Multimodel inference for biomarker development: an application to schizophrenia.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets