Dataset Information

Using random forest to identify longitudinal predictors of health in a 30-year cohort study.

ABSTRACT: Due to the wealth of exposome data from longitudinal cohort studies that is currently available, the need for methods to adequately analyze these data is growing. We propose an approach in which machine learning is used to identify longitudinal exposome-related predictors of health, and illustrate its potential through an application. Our application involves studying the relation between exposome and self-perceived health based on the 30-year running Doetinchem Cohort Study. Random Forest (RF) was used to identify the strongest predictors due to its favorable prediction performance in prior research. The relation between predictors and outcome was visualized with partial dependence and accumulated local effects plots. To facilitate interpretation, exposures were summarized by expressing them as the average exposure and average trend over time. The RF model's ability to discriminate poor from good self-perceived health was acceptable (Area-Under-the-Curve = 0.707). Nine exposures from different exposome-related domains were largely responsible for the model's performance, while 87 exposures seemed to contribute little to the performance. Our approach demonstrates that ML can be interpreted more than widely believed, and can be applied to identify important longitudinal predictors of health over the life course in studies with repeated measures of exposure. The approach is context-independent and broadly applicable.

SUBMITTER: Loef B

PROVIDER: S-EPMC9209521 | biostudies-literature | 2022 Jun

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Using random forest to identify longitudinal predictors of health in a 30-year cohort study.

Loef Bette B Wong Albert A Janssen Nicole A H NAH Strak Maciek M Hoekstra Jurriaan J Picavet H Susan J HSJ Boshuizen H C Hendriek HCH Verschuren W M Monique WMM Herber Gerrie-Cor M GM

Scientific reports 20220620 1

Due to the wealth of exposome data from longitudinal cohort studies that is currently available, the need for methods to adequately analyze these data is growing. We propose an approach in which machine learning is used to identify longitudinal exposome-related predictors of health, and illustrate its potential through an application. Our application involves studying the relation between exposome and self-perceived health based on the 30-year running Doetinchem Cohort Study. Random Forest (RF) ...[more]

PMID: 35725920

Similar Datasets

Project description:ObjectiveReadmission rates after posterior cervical fusion (PCF) significantly impact patients and healthcare, with complication rates at 15%-25% and up to 12% 90-day readmission rates. In this study, we aim to test whether machine learning (ML) models that capture interfactorial interactions outperform traditional logistic regression (LR) in identifying readmission-associated factors.MethodsThe Optum Clinformatics Data Mart database was used to identify patients who underwent PCF between 2004-2017. To determine factors associated with 30-day readmissions, 5 ML models were generated and evaluated, including a multivariate LR (MLR) model. Then, the best-performing model, Gradient Boosting Machine (GBM), was compared to the LACE (Length patient stay in the hospital, Acuity of admission of patient in the hospital, Comorbidity, and Emergency visit) index regarding potential cost savings from algorithm implementation.ResultsThis study included 4,130 patients, 874 of which were readmitted within 30 days. When analyzed and scaled, we found that patient discharge status, comorbidities, and number of procedure codes were factors that influenced MLR, while patient discharge status, billed admission charge, and length of stay influenced the GBM model. The GBM model significantly outperformed MLR in predicting unplanned readmissions (mean area under the receiver operating characteristic curve, 0.846 vs. 0.829; p < 0.001), while also projecting an average cost savings of 50% more than the LACE index.ConclusionFive models (GBM, XGBoost [extreme gradient boosting], RF [random forest], LASSO [least absolute shrinkage and selection operator], and MLR) were evaluated, among which, the GBM model exhibited superior predictive performance, robustness, and accuracy. Factors associated with readmissions impact LR and GBM models differently, suggesting that these models can be used complementarily. When analyzing PCF procedures, the GBM model resulted in greater predictive performance and was associated with higher theoretical cost savings for readmissions associated with PCF complications.

Project description:BackgroundClostridiales and Bacteroidales are uniquely adapted to the gut environment and have co-evolved with their hosts resulting in convergent microbiome patterns within mammalian species. As a result, members of Clostridiales and Bacteroidales are particularly suitable for identifying sources of fecal contamination in environmental samples. However, a comprehensive evaluation of their predictive power and development of computational approaches is lacking. Given the global public health concern for waterborne disease, accurate identification of fecal pollution sources is essential for effective risk assessment and management. Here, we use random forest algorithm and 16S rRNA gene amplicon sequences assigned to Clostridiales and Bacteroidales to identify common fecal pollution sources. We benchmarked the accuracy, consistency, and sensitivity of our classification approach using fecal, environmental, and artificial in silico generated samples.ResultsClostridiales and Bacteroidales classifiers were composed mainly of sequences that displayed differential distributions (host-preferred) among sewage, cow, deer, pig, cat, and dog sources. Each classifier correctly identified human and individual animal sources in approximately 90% of the fecal and environmental samples tested. Misclassifications resulted mostly from false-positive identification of cat and dog fecal signatures in host animals not used to build the classifiers, suggesting characterization of additional animals would improve accuracy. Random forest predictions were highly reproducible, reflecting the consistency of the bacterial signatures within each of the animal and sewage sources. Using in silico generated samples, we could detect fecal bacterial signatures when the source dataset accounted for as little as ~ 0.5% of the assemblage, with ~ 0.04% of the sequences matching the classifiers. Finally, we developed a proxy to estimate proportions among sources, which allowed us to determine which sources contribute the most to observed fecal pollution.ConclusionRandom forest classification with 16S rRNA gene amplicons offers a rapid, sensitive, and accurate solution for identifying host microbial signatures to detect human and animal fecal contamination in environmental samples.

Dataset Information

Using random forest to identify longitudinal predictors of health in a 30-year cohort study.

Publications

Using random forest to identify longitudinal predictors of health in a 30-year cohort study.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets