New method estimates causal effects of multiple versions of treatment.
problem Ignoring multiple versions of treatment leads to biased causal effect estimates.
method Mixture-of-Experts framework for estimating version-specific causal effects.
result Effective method for estimating causal effects of latent versions.
M3E2 neural network estimates multiple treatment effects.
problem Estimating effects of multiple treatments simultaneously.
method Multi-task learning neural network model for multiple treatments, continuous and binary.
result M3E2 outperforms baselines in synthetic datasets.
Estimates the effect of time-varying treatments using machine learning.
problem Estimating the impact of time-varying treatments over multiple periods.
method Difference-in-Differences framework with double/debiased machine learning.
result Higher vaccination rates reduce COVID-19 mortality after several weeks.
New method identifies proxies for causal effects on multiple outcomes.
problem Estimating causal effects in scenarios with multiple outcomes and treatments.
method Causal discovery method leveraging multiple outcomes as proxies for each treatment effect.
result Parallel studies of multiple outcomes can assist in causal identification.
Method estimates dynamic treatment effects using machine learning and g-estimation.
problem Estimating treatment effects over time with multiple treatments and potential future outcomes.
method Double/debiased machine learning framework for dynamic treatment effects, extending Neyman orthogonal cross-fitted g-estimation. result Provides finite sample guarantees and allows for non-linear effect heterogeneity and high-dimensional parameterizations.
The paper compares methods for estimating heterogeneous treatment effects using multiple randomized trials.
problem Estimating heterogeneous treatment effects reliably and precisely with a single dataset is challenging.
method Non-parametric approaches for estimating heterogeneous treatment effects using data from multiple trials.
result Methods that directly allow for heterogeneity of the treatment effect across trials perform better than those that do not.
SDD improves DD for estimating treatment effects by adjusting for confounding.
problem Estimating treatment effects in observational studies with confounding.
method Synthesized Difference in Differences (SDD) using RCT data to infer correct slopes.
result SDD achieves state-of-the-art performance across synthetic and real datasets.
TCFimt forecasts causal effects of multiple interventions from individual data.
problem Estimating causal effects of temporal multi-interventions from individual data.
method TCFimt uses adversarial tasks in seq2seq framework to alleviate bias and contrastive learning to decouple effects.
result TCFimt outperforms state-of-the-art methods in predicting future outcomes and choosing optimal treatments.
Proposes a generalized causal tree for handling multiple treatments in uplift modeling.
problem Handling multiple treatments in uplift modeling.
method Generalizes causal tree algorithm to handle multiple discrete and continuous-valued treatments.
result Demonstrates improved performance over existing methods in experiments and real data examples.
Proposes methods for learning optimal dynamic treatment regimes robust to unconfoundedness violations.
problem Estimating optimal dynamic treatment regimes using historical observational data when unconfoundedness is violated.
method Utilizes proximal causal inference framework to propose three nonparametric identification methods, a (K+1)-robust method, and establish a semiparametric efficiency bound.
result Establishes the (K+1)-robust method for learning optimal dynamic treatment regimes, validating its efficiency and multiple robustness through numerical experiments.
NCoRE learns counterfactual representations for combined treatments.
problem Estimating individual response to multiple simultaneous interventions.
method Neural conditional representation with modulators for cross-treatment interactions.
result NCoRE significantly outperforms existing methods in counterfactual treatment effect estimation.
RAMEN corrects observational data biases for multiple environments.
problem Bias in observational data for causal inference.
method RAMEN algorithm that leverages heterogeneity of multiple data sources.
result RAMEN produces unbiased treatment effect estimates.
The paper studies causal effects of multiple treatments in healthcare databases with rare outcomes.
problem Estimating causal effects of multiple treatments in healthcare databases with rare outcomes.
method The paper designs three sets of simulations and compares the operating characteristics of three types of methods: Bayesian Additive Regression Trees (BART), regression adjustment on multivariate spline of generalized propensity scores (RAMS), and inverse probability of treatment weighting (IPTW) with multinomial logistic regression or generalized boosted models.
result BART and RAMS provide lower bias and mean squared error compared to IPTW methods.
Causal inference from observational data requires assumptions. These assumptions range from measuring confounders to identifying instruments. Traditionally, causal inference assumptions have focused on estimation of effects for a single treatment. In this work, we construct techniques for estimation with multiple treat…
Uplift modeling is an emerging machine learning approach for estimating the treatment effect at an individual or subgroup level. It can be used for optimizing the performance of interventions such as marketing campaigns and product designs. Uplift modeling can be used to estimate which users are likely to benefit from …
MDCN improves treatment effect estimation in multicenter observational studies.
problem Incongruities in multicenter observational studies due to center-specific protocols and treatment reactions.
method MDCN learns a new feature embedding to address selection bias and strengthen information sharing between similar centers.
result MDCN provides more accurate treatment insights for new, unobserved centers compared to existing methods.
System interprets complex treatment effects for personalized policies.
problem Complex, hard-to-understand treatment effect models.
method Scalable, interpretable personalized experimentation system.
result Learned explanations and generated interpretable policies.
Meta-analysis improves personalized treatment rules across multiple sites.
problem Lack of generalizability in learning individualized treatment rules across different medical sites.
method Developed a method for individual-level meta-analysis of ITRs, borrowing sign-coherency information between sites.
result Jointly learned site-specific ITRs with improved generalizability.
New method estimates individual treatment effects using domain generalization.
problem Estimating causal individual treatment effects from observational data with treatment bias.
method Invariant Risk Minimization (IRM) framework to learn predictors invariant to domain-dependent factors.
result IRM-based ITE estimator shows gains over classical regression approaches in settings with pronounced support mismatch.
Bayesian model ranks treatments in multi-response experiments.
problem Identifying the best treatment among competing ideal properties.
method Bayesian approach with Markov Chain Monte Carlo algorithm.
result Reliable inference of treatment ranks in practice.
Framework integrates mental disorder measurements for personalized treatment.
problem Optimizing treatment for mental disorders with latent mental states and heterogeneity.
method Measurement theory and multi-layer neural network for complex treatment effects.
result Learned treatment policies outperform alternatives on heterogeneous treatment effects.
New method estimates effects of multiple nutrients on blood glucose.
problem Estimating physiological response to multiple nutrient treatments.
method Convolution-based multi-output Gaussian process model.
result Improved prediction accuracy and better interpretation of individual nutrient effects.
While sample sizes in randomized clinical trials are large enough to estimate the average treatment effect well, they are often insufficient for estimation of treatment-covariate interactions critical to studying data-driven precision medicine. Observational data from real world practice may play an important role in a…
New method selects best HTE estimator without ground-truth treatment effects.
problem Selecting best HTE estimator from multiple candidates.
method Cross-fitted, exponentially weighted test statistic with two-way sample splitting.
result Empirically, reliable error control and reduced false selections.
Treatment effects can be estimated from observational data as the difference in potential outcomes. In this paper, we address the challenge of estimating the potential outcome when treatment-dose levels can vary continuously over time. Further, the outcome variable may not be measured at a regular frequency. Our propos…
Paper tackles estimating individual treatment effects from observational data.
problem Estimating the difference between outcomes with and without treatment from single observation.
method Formulated as inference from hidden variables, uses a model of four causal populations, proposes ECM algorithm.
result ECM algorithm provides better performance compared to baseline methods on synthetic and real-world data.
CRL approach improves understanding of heterogeneous treatment effects in complex diseases.
problem Estimating heterogeneous treatment effects in complex diseases.
method Causal rule learning (CRL) workflow consisting of rule discovery, selection, and analysis.
result CRL outperforms other methods in providing interpretable estimates of HTE.
The causal effect of a treatment can vary from person to person based on their individual characteristics and predispositions. Mining for patterns of individual-level effect differences, a problem known as heterogeneous treatment effect estimation, has many important applications, from precision medicine to recommender…
Method constructs prediction intervals for time-varying individual treatment effects.
problem Accurately quantify uncertainty of individual treatment effects across multiple decision points.
method Conformal inference techniques for time-varying ITEs with weaker assumptions.
result Guaranteed lower bound for coverage dependent on data non-exchangeability.
GraphITE estimates individual effects of graph-structured treatments.
problem Estimating individual effects of complex treatment structures.
method Graph neural networks and Hilbert-Schmidt Independence Criterion regularization.
result GraphITE outperforms baselines in estimating treatment effects for large numbers of treatments.
Develops a sparsity-inducing Bayesian Causal Forest for estimating heterogeneous treatment effects.
problem Estimating heterogeneous treatment effects using observational data with varying degrees of sparsity.
method Introduces a sparsity-inducing version of Bayesian Causal Forests with additional priors to adjust covariate weights.
result Improves adaptability to sparse data generating processes and uncovering moderating factors driving heterogeneity.
Estimates treatment effects in panel data with general intervention patterns.
problem Estimating average treatment effects in panel data with heterogeneous treatment effects.
method Extends synthetic control framework to allow rate-optimal recovery of average treatment effects for general intervention patterns.
result First rate-optimal guarantees for general intervention patterns in estimating average treatment effects.
CRN model estimates treatment effects over time using adversarial balancing.
problem Estimating treatment effects over time in medical settings.
method Adversarial domain balancing to remove time-varying confounders.
result CRN achieves lower error in estimating counterfactuals and treatment timing.
Proposes a model to estimate effects of multiple related treatments.
problem Estimating effects of many related treatments in observational data.
method Customized ridge regression to reduce noise and MSE.
result Significantly reduces MSE for individual sub-treatments while allowing reconstruction of aggregated treatment effects.
The paper develops deep learning models for personalized treatment rules in survival analysis.
problem Deriving optimal treatment rules for bivariate survival outcomes in randomized trials.
method Adaptive prediction-powered learning using deep neural networks and stochastic policies.
result Maximizes joint survival probability beyond fixed time points (t1,t2). Estimating the long-term effects of treatments is of interest in many fields. A common challenge in estimating such treatment effects is that long-term outcomes are unobserved in the time frame needed to make policy decisions. One approach to overcome this missing data problem is to analyze treatments effects on an int…
Inverse classification uses an induced classifier as a queryable oracle to guide test instances towards a preferred posterior class label. The result produced from the process is a set of instance-specific feature perturbations, or recommendations, that optimally improve the probability of the class label. In this work…
The paper develops methods to estimate optimal treatment sequences under policy constraints.
problem Estimating the best sequence of treatments over multiple stages for individuals.
method Empirical welfare maximization approach, solving treatment assignment sequentially or simultaneously.
result Established convergence rates and upper bounds for estimation methods.
Synthetic interventions extend SC method to multiple treatments.
problem Evaluating the impact of multiple treatments in panel data.
method Low-rank tensor factor model for latent factors of treatments.
result Consistent and asymptotically normal estimators for synthetic interventions.
New method combines multiple datasets to estimate ATE with valid confidence intervals.
problem Combining multiple observational datasets to estimate ATE with valid confidence intervals.
method Prediction-powered inferences to shrink CIs and provide valid CIs.
result Valid confidence intervals for ATE from multiple datasets.
ICA accurately estimates treatment effects even with confounders.
problem Estimating treatment effects in the presence of confounding variables.
method Uses Independent Component Analysis (ICA) to identify latent sources and estimate mixing coefficients.
result Linear ICA can consistently estimate multiple treatment effects, even with Gaussian confounders, and is more sample-efficient than Orthogonal Machine Learning (OML).
The complexity of human cancer often results in significant heterogeneity in response to treatment. Precision medicine offers potential to improve patient outcomes by leveraging this heterogeneity. Individualized treatment rules (ITRs) formalize precision medicine as maps from the patient covariate space into the space…
An optimal dynamic treatment regime (DTR) consists of a sequence of decision rules in maximizing long-term benefits, which is applicable for chronic diseases such as HIV infection or cancer. In this paper, we develop a novel angle-based approach to search the optimal DTR under a multicategory treatment framework for su…
Meta-learners estimate CATE from multiple environments with partial identification.
problem Estimating CATE from observational data across multiple environments with violations of causal assumptions.
method Adapt IV literature for partial identification, propose model-agnostic meta-learners.
result Meta-learners effectively estimate CATE bounds across various experiments.
A novel framework synthesizes treatment data across sites using optimal transport.
problem Estimating treatment effects across different sites with varying conditions.
method Distributional causal inference, Optimal Transport for alignment of control group distributions.
result Synthetic treatment group data aligns with true target distribution under general conditions.
Optimal biomarker combinations for treatment-selection can be derived by minimizing total burden to the population caused by the targeted disease and its treatment. However, when multiple biomarkers are present, including all in the model can be expensive and hurt model performance. To remedy this, we consider feature …
Developed a new algorithm to improve dynamic treatment regimens.
problem Non-convergence of Q-learning-based Q-shared algorithm in dynamic treatment regimens.
method Penalized Q-shared algorithm to address convergence issues.
result The penalized Q-shared algorithm converges and outperforms the original in various settings.
Develops methods for near-optimal personalized treatment recommendations.
problem Assigning optimal treatments to patients based on individual characteristics.
method Outcome weighted learning framework to estimate near-optimal alternative individualized treatment recommendations (A-ITR).
result Consistency of proposed methods and upper bound for risk between optimal and estimated recommendations.