Dataset Information

The identification of complex interactions in epidemiology and toxicology: a simulation study of boosted regression trees.

ABSTRACT: BACKGROUND:There is a need to evaluate complex interaction effects on human health, such as those induced by mixtures of environmental contaminants. The usual approach is to formulate an additive statistical model and check for departures using product terms between the variables of interest. In this paper, we present an approach to search for interaction effects among several variables using boosted regression trees. METHODS:We simulate a continuous outcome from real data on 27 environmental contaminants, some of which are correlated, and test the method's ability to uncover the simulated interactions. The simulated outcome contains one four-way interaction, one non-linear effect and one interaction between a continuous variable and a binary variable. Four scenarios reflecting different strengths of association are simulated. We illustrate the method using real data. RESULTS:The method succeeded in identifying the true interactions in all scenarios except where the association was weakest. Some spurious interactions were also found, however. The method was also capable to identify interactions in the real data set. CONCLUSIONS:We conclude that boosted regression trees can be used to uncover complex interaction effects in epidemiological studies.

SUBMITTER: Lampa E

PROVIDER: S-EPMC4120739 | biostudies-literature | 2014 Jul

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

The identification of complex interactions in epidemiology and toxicology: a simulation study of boosted regression trees.

Lampa Erik E Lind Lars L Lind P Monica PM Bornefalk-Hermansson Anna A

Environmental health : a global access science source 20140704

<h4>Background</h4>There is a need to evaluate complex interaction effects on human health, such as those induced by mixtures of environmental contaminants. The usual approach is to formulate an additive statistical model and check for departures using product terms between the variables of interest. In this paper, we present an approach to search for interaction effects among several variables using boosted regression trees.<h4>Methods</h4>We simulate a continuous outcome from real data on 27 e ...[more]

PMID: 24993424

Similar Datasets

Project description:Boosted Regression Trees (BRT) is one of the modelling techniques most recently applied to biodiversity conservation and it can be implemented with presence-only data through the generation of artificial absences (pseudo-absences). In this paper, three pseudo-absences generation techniques are compared, namely the generation of pseudo-absences within target-group background (TGB), testing both the weighted (WTGB) and unweighted (UTGB) scheme, and the generation at random (RDM), evaluating their performance and applicability in distribution modelling and species conservation. The choice of the target group fell on amphibians, because of their rapid decline worldwide and the frequent lack of guidelines for conservation strategies and regional-scale planning, which instead could be provided through an appropriate implementation of SDMs. Bufo bufo, Salamandrina perspicillata and Triturus carnifex were considered as target species, in order to perform our analysis with species having different ecological and distributional characteristics. The study area is the "Gran Sasso-Monti della Laga" National Park, which hosts 15 Natura 2000 sites and represents one of the most important biodiversity hotspots in Europe. Our results show that the model calibration ameliorates when using the target-group based pseudo-absences compared to the random ones, especially when applying the WTGB. Contrarily, model discrimination did not significantly vary in a consistent way among the three approaches with respect to the tree target species. Both WTGB and RDM clearly isolate the highly contributing variables, supplying many relevant indications for species conservation actions. Moreover, the assessment of pairwise variable interactions and their three-dimensional visualization further increase the amount of useful information for protected areas' managers. Finally, we suggest the use of RDM as an admissible alternative when it is not possible to individuate a suitable set of species as a representative target-group from which the pseudo-absences can be generated.

Project description:BackgroundClinical models to predict first trimester viability are traditionally based on multivariable logistic regression (LR) which is not directly interpretable for non-statistical experts like physicians. Furthermore, LR requires complete datasets and pre-established variables specifications. In this study, we leveraged the internal non-linearity, feature selection and missing values handling mechanisms of machine learning algorithms, along with a post-hoc interpretability strategy, as potential advantages over LR for clinical modeling.MethodsThe dataset included 1154 patients with 2377 individual scans and was obtained from a prospective observational cohort study conducted at a hospital in London, UK, from March 2014 to May 2019. The data were split into a training (70%) and a test set (30%). Parsimonious and complete multivariable models were developed from two algorithms to predict first trimester viability at 11-14 weeks gestational age (GA): LR and light gradient boosted machine (LGBM). Missing values were handled by multiple imputation where appropriate. The SHapley Additive exPlanations (SHAP) framework was applied to derive individual explanations of the models.ResultsThe parsimonious LGBM model had similar discriminative and calibration performance as the parsimonious LR (AUC 0.885 vs 0.860; calibration slope: 1.19 vs 1.18). The complete models did not outperform the parsimonious models. LGBM was robust to the presence of missing values and did not require multiple imputation unlike LR. Decision path plots and feature importance analysis revealed different algorithm behaviors despite similar predictive performance. The main driving variable from the LR model was the pre-specified interaction between fetal heart presence and mean sac diameter. The crown-rump length variable and a proxy variable reflecting the difference in GA between expected and observed GA were the two most important variables of LGBM. Finally, while variable interactions must be specified upfront with LR, several interactions were ranked by the SHAP framework among the most important features learned automatically by the LGBM algorithm.ConclusionsGradient boosted algorithms performed similarly to carefully crafted LR models in terms of discrimination and calibration for first trimester viability prediction. By handling multi-collinearity, missing values, feature selection and variable interactions internally, the gradient boosted trees algorithm, combined with SHAP, offers a serious alternative to traditional LR models.

Dataset Information

The identification of complex interactions in epidemiology and toxicology: a simulation study of boosted regression trees.

Publications

The identification of complex interactions in epidemiology and toxicology: a simulation study of boosted regression trees.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets