Advanced Causal Inference Techniques

High-Dimensional Mediation Analysis with Network Mediators: Applications to Pediatric Acute Lymphoblastic Leukemia.

The paper develops a high-dimensional mediation analysis method for brain connectivity networks and applies it to pediatric ALL data, finding significant mediation effects of brain connectivity on visual processing speed.

MELODY: Mediation Analysis in Logistic Regression for High-Dimensional Mediators and a Binary Outcome

MELODY quantifies total mediation effects in logistic regression for high-dimensional mediators and binary outcomes, outperforming traditional methods in non-rare disease scenarios.

CoxMDS: Multiple Data Splitting for High-dimensional Mediation Analysis with Survival Outcomes in Epigenome-wide Studies

CoxMDS identifies causal mediators in DNA methylation studies with survival outcomes, maintaining FDR control and higher power than existing methods.

HILAMA: High-dimensional multi-omic mediation analysis with latent confounding

HILAMA is a method for high-dimensional mediation analysis with latent confounding, showing improved FDR control and power in simulations and applied to ADNI proteomics-radiomics data.

High-dimensional confounding in causal mediation: a comparison study of double machine learning and regularized partial correlation network

This paper compares double machine learning and regularized partial correlation network methods for high-dimensional confounding in causal mediation analyses using a human microbiome study.

multimedia: Multimodal Mediation Analysis of Microbiome Data

The paper introduces the multimedia R package for mediation analysis of microbiome data, demonstrating its ability to uncover mechanistic interactions in two case studies.

Through the lens of causal inference: Decisions and pitfalls of covariate selection

The paper highlights pitfalls in covariate selection and advocates for principled approaches to improve statistical validity and reproducibility in data analysis.

MR-SPLIT: a novel method to address selection and weak instrument bias in one-sample Mendelian randomization studies

MR-SPLIT reduces bias in one-sample Mendelian randomization by addressing weak instruments and selection bias through sample-splitting and cross-fitting.

Rethinking animal attrition in preclinical research: expressing causal mechanisms of selection bias using directed acyclic graphs

The paper uses directed acyclic graphs to model selection bias from animal attrition in preclinical research, showing how collider stratification bias can distort causal effect estimates.

MASH: Mediation Analysis of Survival Outcome and High-dimensional Omics Mediators with Application to Complex Diseases

MASH identifies high-dimensional mediators in survival analysis, showing cigarette smoking mediates 51.1% of coronary heart disease risk via five metabolites.

A machine learning based approach towards high-dimensional mediation analysis

This paper introduces a machine learning method for high-dimensional mediation analysis and applies it to fMRI data to identify brain patterns mediating relationships between stimuli and behavioral outcomes.

GIVE Statistic for Goodness of Fit in Instrumental Variables Models with Application to COVID Data

The paper introduces the GIVE statistic to assess goodness of fit in instrumental variables models and applies it to analyze the impact of BCG vaccine on COVID-19 mortality.

Efficient Estimation of Indirect Effects in Case-Control Studies Using a Unified Likelihood Framework

The paper introduces a unified likelihood framework for mediation analysis in case-control studies to improve efficiency of indirect effect estimates by leveraging exposure-mediator relationships.

High Dimensional Mediation Analysis with Applications to Causal Gene Identification

The paper develops MedFix and MedMix for high-dimensional mediation analysis, showing improved accuracy and robustness in causal gene identification using a mouse F2 diabetes dataset.

The Triangulation WIthin A STudy (TWIST) framework for causal inference within Pharmacogenetic research

The paper introduces the TWIST framework for causal inference in pharmacogenetics, showing how to combine different estimation strategies to improve robustness and efficiency in estimating genetically moderated treatment effects.

A Deep Learning Semiparametric Regression for Adjusting Complex Confounding Structures

This paper introduces a deep learning semiparametric regression method to adjust complex confounding structures in observational studies, showing reduced bias and mean squared error in parameter estimates.

Mendelian randomisation for mediation analysis: current methods and challenges for implementation

Multivariable and two-step Mendelian randomisation recovered direct and indirect effects with little bias in simulations with confounding or measurement error, conditions that biased phenotypic regression mediation.

Large-Scale Hypothesis Testing for Causal Mediation Effects with Applications in Genome-wide Epigenetic Studies

The paper develops a new statistical test, DACT, to detect mediation effects in genome-wide epigenetic studies and applies it to identify DNA methylation sites mediating smoking's effect on lung function.

Assessing exposure effects on gene expression

The paper compares regression, IPW, and g-formula methods for estimating exposure effects on gene expression, applying them to assess smoking's impact on adipose tissue gene expression.

Estimating and testing the microbial causal mediation effect with high-dimensional and compositional microbiome data

The paper introduces a causal mediation model for high-dimensional microbiome data to estimate and test microbial causal effects in a three-factor study design.

Compositional Mediation Analysis for Microbiome Studies

The paper develops a sparse compositional mediation model to estimate causal effects in microbiome studies and applies it to investigate fat intake's effect on BMI through gut microbiome composition.

Estimating heterogeneous treatment effects by balancing heterogeneity and fitness

The paper proposes a multi-objective splitting procedure to estimate heterogeneous treatment effects by balancing fitness and heterogeneity criteria, achieving better performance than existing methods.

Semiparametric methods for estimation of a non-linear exposure-outcome relationship using instrumental variables with application to Mendelian randomization

This paper introduces two semiparametric IV methods to estimate non-linear exposure-outcome relationships in Mendelian randomization and applies them to study BMI's effect on blood pressure.