Multiple Causal Inference with Latent Confounding

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 treatments in the presence of unobserved confounding. We develop two assumptions based on shared confounding between treatments and independence of treatments given the confounder. Together, these assumptions lead to a confounder estimator regularized by mutual information. For this estimator, we develop a tractable lower bound. To recover treatment effects, we use the residual information in the treatments independent of the confounder. We validate on simulations and an example from clinical medicine.

[1]  Tobias Keck,et al.  Elevated blood urea nitrogen is an independent risk factor of prolonged intensive care unit stay due to acute necrotizing pancreatitis. , 2010, Journal of critical care.

[2]  John D. Storey,et al.  Testing for genetic associations in arbitrarily structured populations , 2014, Nature Genetics.

[3]  Alexandros G. Dimakis,et al.  CausalGAN: Learning Causal Implicit Generative Models with Adversarial Training , 2017, ICLR.

[4]  Ronald J. Williams,et al.  Simple Statistical Gradient-Following Algorithms for Connectionist Reinforcement Learning , 2004, Machine Learning.

[5]  H. Kang,et al.  Variance component model to account for sample structure in genome-wide association studies , 2010, Nature Genetics.

[6]  Radford M. Neal.  Pattern Recognition and Machine Learning , 2007, Technometrics.

[7]  Matthew J Press,et al.  Medicare's New Bundled Payments: Design, Strategy, and Evolution. , 2016, JAMA.

[8]  Ruslan Salakhutdinov,et al.  Evaluation methods for topic models , 2009, ICML '09.

[9]  Peter Szolovits,et al.  MIMIC-III, a freely accessible critical care database , 2016, Scientific Data.

[10]  Thomas M. Cover,et al.  Elements of Information Theory , 2005 .

[11]  Jennifer L. Hill,et al.  Bayesian Nonparametric Modeling for Causal Inference , 2011 .

[12]  Dustin Tran,et al.  Hierarchical Variational Models , 2015, ICML.

[13]  J. Pearl.  Causal inference in statistics: An overview , 2009 .

[14]  W. Shadish,et al.  Experimental and Quasi-Experimental Designs for Generalized Causal Inference , 2001 .

[15]  Ying Liu,et al.  FaST linear mixed models for genome-wide association studies , 2011, Nature Methods.

[16]  Ole Winther,et al.  Auxiliary Deep Generative Models , 2016, ICML.

[17]  Max Welling,et al.  Auto-Encoding Variational Bayes , 2013, ICLR.

[18]  L. Hydo,et al.  Relationship of systemic inflammatory response syndrome to organ dysfunction, length of stay, and mortality in critical surgical illness: effect of intensive care unit resuscitation. , 1999, Archives of surgery.

[19]  Max Welling,et al.  Markov Chain Monte Carlo and Variational Inference: Bridging the Gap , 2014, ICML.

[20]  Jan O. Friedrich,et al.  Medical admission order sets to improve deep vein thrombosis prophylaxis rates and other outcomes. , 2009, Journal of hospital medicine.

[21]  Simon C. Potter,et al.  Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls , 2007, Nature.

[22]  Daan Wierstra,et al.  Stochastic Backpropagation and Approximate Inference in Deep Generative Models , 2014, ICML.

[23]  Eleazar Eskin,et al.  Improved linear mixed models for genome-wide association studies , 2012, Nature Methods.

[24]  Joshua D. Angrist,et al.  Identification of Causal Effects Using Instrumental Variables , 1993 .

[25]  Robert H. Patrick,et al.  Hospital length of stay and probability of acquiring infection , 2010 .

[26]  Stefan Wager,et al.  Estimation and Inference of Heterogeneous Treatment Effects using Random Forests , 2015, Journal of the American Statistical Association.

[27]  M. McMullen,et al.  A unified mixed-model method for association mapping that accounts for multiple levels of relatedness , 2006, Nature Genetics.

[28]  A. Israeli,et al.  Management of severe hypokalemia in hospitalized patients: a study of quality of care based on computerized databases. , 2001, Archives of internal medicine.

[29]  Uri Shalit,et al.  Learning Representations for Counterfactual Inference , 2016, ICML.

[30]  David Barber,et al.  An Auxiliary Variational Method , 2004, ICONIP.

[31]  David M. Blei,et al.  The Blessings of Multiple Causes , 2018, Journal of the American Statistical Association.

[32]  D. Rubin,et al.  The central role of the propensity score in observational studies for causal effects , 1983 .

[33]  Sean Gerrish,et al.  Black Box Variational Inference , 2013, AISTATS.

[34]  Max Welling,et al.  Causal Effect Inference with Deep Latent-Variable Models , 2017, NIPS 2017.

[35]  Uri Shalit,et al.  Estimating individual treatment effect: generalization bounds and algorithms , 2016, ICML.

[36]  Martin Schumacher,et al.  Nosocomial Infection, Length of Stay, and Time-Dependent Bias , 2009, Infection Control & Hospital Epidemiology.

[37]  L. Andrews,et al.  An alternative strategy for studying adverse events in medical care , 1997, The Lancet.

[38]  Wei-Chun Lin,et al.  Preliminary Prospective Study to Assess the Effect of Early Blood Urea Nitrogen/Creatinine Ratio-Based Hydration Therapy on Poststroke Infection Rate and Length of Stay in Acute Ischemic Stroke. , 2015, Journal of stroke and cerebrovascular diseases : the official journal of National Stroke Association.

[39]  Bernhard Schölkopf,et al.  Nonlinear causal discovery with additive noise models , 2008, NIPS.

[40]  Dustin Tran,et al.  Implicit Causal Models for Genome-wide Association Studies , 2017, ICLR.