Estimating treatment effects with machine learning

2019/10/10 by K. John McConnell, Stephan Lindner · 12 citations
Mathematics · Medicine · #Advanced Causal Inference Techniques #Artificial Intelligence in Healthcare and Education #Sepsis Diagnosis and Treatment

paper · doi:10.1111/1475-6773.13212

Abstract

OBJECTIVE: To demonstrate the performance of methodologies that include machine learning (ML) algorithms to estimate average treatment effects under the assumption of exogeneity (selection on observables). DATA SOURCES: Simulated data and observational data on hospitalized adults. STUDY DESIGN: We assessed the performance of several ML-based estimators, including Targeted Maximum Likelihood Estimation, Bayesian Additive Regression Trees, Causal Random Forests, Double Machine Learning, and Bayesian Causal Forests, applying these methods to simulated data as well as data on the effects of right heart catheterization. PRINCIPAL FINDINGS: In Monte Carlo studies, ML-based estimators generated estimates with smaller bias than traditional regression approaches, demonstrating substantial (69 percent-98 percent) bias reduction in some scenarios. Bayesian Causal Forests and Double Machine Learning were top performers, although all were sensitive to high dimensional (>150) sets of covariates. CONCLUSIONS: ML-based methods are promising methods for estimating treatment effects, allowing for the inclusion of many covariates and automating the search for nonlinearities and interactions among variables. We provide guidance and sample code for researchers interested in implementing these tools in their own empirical work.

Citations

• Dealing with limited overlap in estimation of average treatment effects
• The central role of the propensity score in observational studies for causal effects
• Prediction Policy Problems
• Targeted Maximum Likelihood Estimation for Causal Inference in Observational Studies
• Double/Debiased/Neyman Machine Learning of Treatment Effects

Cited by

• A Framework for Assessing the Habitat Correlates of Spatially Explicit Population Trends
• Use of behavioral health care in Medicaid managed care carve-out versus carve-in arrangements. [europepmc]
• Assessing putative bias in prediction of anti-microbial resistance from real-world genotyping data under explicit causal assumptions. [europepmc]
• External control arm analysis: an evaluation of propensity score approaches, G-computation, and doubly debiased machine learning. [europepmc]
• Assessing the properties of patient-specific treatment effect estimates from causal forest algorithms under essential heterogeneity. [europepmc]
• Can supervised deep learning architecture outperform autoencoders in building propensity score models for matching? [europepmc]
• Do health care quality improvement policies work for all? Distributional effects by baseline quality in South Africa. [europepmc]
• Machine learning in causal inference for epidemiology. [europepmc]
• An overview of modern machine learning methods for effect measure modification analyses in high-dimensional settings. [europepmc]
• Prescriptive Predictors of Mindfulness Ecological Momentary Intervention for Social Anxiety Disorder: Machine Learning Analysis of Randomized Controlled Trial Data. [europepmc]
• A Causal Machine Learning Framework for Estimating the Impact of Cancer Diagnosis on Receipt of Advance Care Planning. [europepmc]
• Toward precision stroke rehabilitation: an integrated causal machine learning and clinician feedback approach. [europepmc]

Related

• Estimating and discovering heterogeneous treatment effects using machine learning in epidemiological studies: a practical guide
• Estimating Heterogeneous Treatment Effects With Real-World Health Data: A Scoping Review of Machine Learning Methods
• Estimating Heterogeneous Treatment Effects with Real-World Health Data – A Scoping Review of Machine Learning Methods
• Causal Machine Learning Methods for Estimating Personalised Treatment Effects -- Insights on validity from two large trials
• Predicting treatment effects from observational studies using machine learning methods: A simulation study