Dataset Information

Using Innovative Machine Learning Methods to Screen and Identify Predictors of Congenital Heart Diseases.

ABSTRACT: Objective: Congenital heart diseases (CHDs) are associated with an extremely heavy global disease burden as the most common category of birth defects. Genetic and environmental factors have been identified as risk factors of CHDs previously. However, high volume clinical indicators have never been considered when predicting CHDs. This study aimed to predict the occurrence of CHDs by considering thousands of variables from self-reported questionnaires and routinely collected clinical laboratory data using machine learning algorithms. Methods: We conducted a birth cohort study at one of the largest cardiac centers in China from 2011 to 2017. All fetuses were screened for CHDs using ultrasound and cases were confirmed by at least two pediatric cardiologists using echocardiogram. A total of 1,127 potential predictors were included to predict CHDs. We used the Explainable Boosting Machine (EBM) for prediction and evaluated the model performance using area under the Receive Operating Characteristics (ROC) curves (AUC). The top predictors were selected according to their contributions and predictive values. Thresholds were calculated for the most significant predictors. Results: Overall, 5,390 mother-child pairs were recruited. Our prediction model achieved an AUC of 76% (69-83%) from out-of-sample predictions. Among the top 35 predictors of CHDs we identified, 34 were from clinical laboratory tests and only one was from the questionnaire (abortion history). Total accuracy, sensitivity, and specificity were 0.65, 0.74, and 0.65, respectively. Maternal serum uric acid (UA), glucose, and coagulation levels were the most consistent and significant predictors of CHDs. According to the thresholds of the predictors identified in our study, which did not reach the current clinical diagnosis criteria, elevated UA (>4.38 mg/dl), shortened activated partial thromboplastin time (<33.33 s), and elevated glucose levels were the most important predictors and were associated with ranges of 1.17-1.54 relative risks of CHDs. We have developed an online predictive tool for CHDs based on our findings that may help screening and prevention of CHDs. Conclusions: Maternal UA, glucose, and coagulation levels were the most consistent and significant predictors of CHDs. Thresholds below the current clinical definition of "abnormal" for these predictors could be used to help develop CHD screening and prevention strategies.

SUBMITTER: Qu Y

PROVIDER: S-EPMC8777022 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:The relationship between increased cerebral spinal fluid (CSF) ventricular compartments, structural and microstructural dysmaturation, and executive function in patients with congenital heart disease (CHD) is unknown. Here, we leverage a novel machine-learning data-driven technique to delineate interrelationships between CSF ventricular volume, structural and microstructural alterations, clinical risk factors, and sub-domains of executive dysfunction in adolescent CHD patients. We trained random forest regression models to predict measures of executive function (EF) from the NIH Toolbox, the Delis-Kaplan Executive Function System (D-KEFS), and the Behavior Rating Inventory of Executive Function (BRIEF) and across three subdomains of EF - mental flexibility, working memory, and inhibition. We estimated the best parameters for the random forest algorithm via a randomized grid search of parameters using 10-fold cross-validation on the training set only. The best parameters were then used to fit the model on the full training set and validated on the test set. Algorithm performance was measured using root-mean squared-error (RMSE). As predictors, we included patient clinical variables, perioperative clinical measures, microstructural white matter (diffusion tensor imaging- DTI), and structural volumes (volumetric magnetic resonance imaging- MRI). Structural white matter was measured using along-tract diffusivity measures of 13 inter-hemispheric and cortico-association fibers. Structural volumes were measured using FreeSurfer and manual segmentation of key structures. Variable importance was measured by the average Gini-impurity of each feature across all decision trees in which that feature is present in the model, and functional ontology mapping (FOM) was used to measure the degree of overlap in feature importance for each EF subdomain and across subdomains. We found that CSF structural properties (including increased lateral ventricular volume and reduced choroid plexus volumes) in conjunction with proximate cortical projection and paralimbic-related association white matter tracts that straddle the lateral ventricles and distal paralimbic-related subcortical structures (basal ganglia, hippocampus, cerebellum) are predictive of two-specific subdomains of executive dysfunction in CHD patients: cognitive flexibility and inhibition. These findings in conjunction with combined RF models that incorporated clinical risk factors, highlighted important clinical risk factors, including the presence of microbleeds, altered vessel volume, and delayed PDA closure, suggesting that CSF-interstitial fluid clearance, vascular pulsatility, and glymphatic microfluid dynamics may be pathways that are impaired in CHD, providing mechanistic information about the relationship between CSF and executive dysfunction.

Project description:ObjectiveTo identify predictors of prediabetes using feature selection and machine learning on a nationally representative sample of the US population.Materials and methodsWe analyzed n = 6346 men and women enrolled in the National Health and Nutrition Examination Survey 2013-2014. Prediabetes was defined using American Diabetes Association guidelines. The sample was randomly partitioned to training (n = 3174) and internal validation (n = 3172) sets. Feature selection algorithms were run on training data containing 156 preselected exposure variables. Four machine learning algorithms were applied on 46 exposure variables in original and resampled training datasets built using 4 resampling methods. Predictive models were tested on internal validation data (n = 3172) and external validation data (n = 3000) prepared from National Health and Nutrition Examination Survey 2011-2012. Model performance was evaluated using area under the receiver operating characteristic curve (AUROC). Predictors were assessed by odds ratios in logistic models and variable importance in others. The Centers for Disease Control (CDC) prediabetes screening tool was the benchmark to compare model performance.ResultsPrediabetes prevalence was 23.43%. The CDC prediabetes screening tool produced 64.40% AUROC. Seven optimal (≥ 70% AUROC) models identified 25 predictors including 4 potentially novel associations; 20 by both logistic and other nonlinear/ensemble models and 5 solely by the latter. All optimal models outperformed the CDC prediabetes screening tool (P < 0.05).DiscussionCombined use of feature selection and machine learning increased predictive performance outperforming the recommended screening tool. A range of predictors of prediabetes was identified.ConclusionThis work demonstrated the value of combining feature selection with machine learning to identify a wide range of predictors that could enhance prediabetes prediction and clinical decision-making.

Project description:BackgroundPredicting birth weight and identifying its risk factors are clinically important. This study aims to use interpretable machine learning to predict birth weight and identity important predictors.MethodsThis prospective cohort study was conducted in Tongzhou Maternal and Child Health Care Hospital of Beijing, China, recruiting pregnant women between June 2018 and February 2019. We used 24 features to predict infant birth weight, including gestational age, mother's age, parity, history of macrosomia delivery, pre-pregnancy body mass index (BMI), height, father's BMI, lifestyle (diet, physical activity, smoking), and biomarker (fasting glucose and lipids) features. Study outcome was birth weight of infant. We used 8 supervised learning models including 4 individual [linear regression, ridge regression, lasso regression, support vector machines regression (SVR)], and 4 ensemble estimators (random forest, AdaBoost, gradient boosted trees, and voting ensemble for regression) to predict birth weight. Model accuracy was measured by root mean squared error (RMSE) of 10-fold cross validation on the training set and RMSE of prediction on the test set. We used permutation importance algorithm to understand the prediction from the models and what affected them.ResultThis study included 4,754 mother-child dyads. RMSEs were lower in voting ensemble for regression, linear regression, and SVR than random forest, AdaBoost, and gradient boosted tree. The 5 most important predictors for infant birth weight were gestational age, fetal sex, preterm birth, mother's height, and pre-pregnancy BMI. After adding ultrasound-measured indicators of fetal growth into predictors, mother's height and pre-pregnancy BMI remained the most important predictors in predicting the outcome.ConclusionMother's height and pre-pregnancy BMI were identified as important predictors for infant birth weight. Interpretable machine learning is a promising tool in the prediction of birth weight.

Project description:BackgroundAlthough different quality controls have been applied at different stages of the sample preparation and data analysis to ensure both reproducibility and reliability of RNA-seq results, there are still limitations and bias on the detectability for certain differentially expressed genes (DEGs). Whether the transcriptional dynamics of a gene can be captured accurately depends on experimental design/operation and the following data analysis processes. The workflow of subsequent data processing, such as reads alignment, transcript quantification, normalization, and statistical methods for ultimate identification of DEGs can influence the accuracy and sensitivity of DEGs analysis, producing a certain number of false-positivity or false-negativity. Machine learning (ML) is a multidisciplinary field that employs computer science, artificial intelligence, computational statistics and information theory to construct algorithms that can learn from existing data sets and to make predictions on new data set. ML-based differential network analysis has been applied to predict stress-responsive genes through learning the patterns of 32 expression characteristics of known stress-related genes. In addition, the epigenetic regulation plays critical roles in gene expression, therefore, DNA and histone methylation data has been shown to be powerful for ML-based model for prediction of gene expression in many systems, including lung cancer cells. Therefore, it is promising that ML-based methods could help to identify the DEGs that are not identified by traditional RNA-seq method.ResultsWe identified the top 23 most informative features through assessing the performance of three different feature selection algorithms combined with five different classification methods on training and testing data sets. By comprehensive comparison, we found that the model based on InfoGain feature selection and Logistic Regression classification is powerful for DEGs prediction. Moreover, the power and performance of ML-based prediction was validated by the prediction on ethylene regulated gene expression and the following qRT-PCR.ConclusionsOur study shows that the combination of ML-based method with RNA-seq greatly improves the sensitivity of DEGs identification.

Dataset Information

Using Innovative Machine Learning Methods to Screen and Identify Predictors of Congenital Heart Diseases.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets