Bayesian Inference of Individualized Treatment Effects using Multi-task Gaussian Processes

By Ahmed M. Alaa, Mihaela van der Schaar. 2017

346 citations recorded by the source corpus.

Predicated on the increasing abundance of electronic health records, we investi- gate the problem of inferring individualized treatment effects using observational data. Stemming from the potential outcomes model, we propose a novel multi- task learning framework in which factual and counterfactual outcomes are mod- eled as the outputs of a function in a vector-valued reproducing kernel Hilbert space (vvRKHS). We develop a nonparametric Bayesian method for learning the treatment effects using a multi-task Gaussian process (GP) with a linear coregion- alization kernel as a prior over the vvRKHS. The Bayesian approach allows us to compute individualized measures of confidence in our estimates via pointwise credible intervals, which are crucial for realizing the full potential of precision medicine. The impact of selection bias is alleviated via a risk-based empirical Bayes method for adapting the multi-task GP prior, which jointly minimizes the empirical error in factual outcomes and the uncertainty in (unobserved) counter- factual outcomes. We conduct experiments on observational datasets for an inter- ventional social program applied to premature infants, and a left ventricular assist device applied to cardiac patients wait-listed for a heart transplant. In both experi- ments, we show that our method significantly outperforms the state-of-the-art.

Subfields: Artificial Intelligence, Statistics and Probability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Statistics and Probability

Citations in the corpus

Time Series Deconfounder: Estimating Treatment Effects over Time in the Presence of Hidden Confounders (2019, methodology): “Most of the previous methods for performing causal inference have focused on the static setting (Hill, 2011; Wager & Athey, 2017; Alaa & van der Schaar, 2017; Yoon et al., 2018), and less attention has been given to the time series setting.”
Estimating Counterfactual Treatment Outcomes over Time Through Adversarially Balanced Representations (2020, methodology): “…have been proposed for performing causal inference using observational data in the static setting (Johansson et al., 2016; Shalit et al., 2017; Alaa & van der Schaar, 2017; Li & Fu, 2017; Yoon et al., 2018; Alaa & van der Schaar, 2018; Yao et al., 2018) and only a few methods address the…”
Representation Balancing MDPs for Off-Policy Policy Evaluation (2018, background): “References [1] A.”
Forecasting Treatment Responses Over Time Using Recurrent Marginal Structural Networks (2018, background): “With the increasing prevalence of electronic health records, there has been much interest in the use of machine learning to estimate treatment effects directly from observational data [13, 41, 44, 2].”
Counterfactual risk assessments, evaluation, and fairness (2019, methodology): “Counterfactual modeling has been proposed for medical RAIs [1, 44, 45], and prior work has used counterfactual evaluation for off-policy learning in bandit settings [14].”
Estimating the Effects of Continuous-valued Interventions using Generative Adversarial Networks (2020, methodology): “Most of the methods developed in the causal inference literature focus on learning the counterfactual outcomes of discrete interventions, such as binary or categorical treatments2 [1–9].”
Causal machine learning for predicting treatment outcomes (2024, neutral): “This is also called overlap assumption.”
Appendix for “Learning Overlapping Representations for the Estimation of Individualized Treatment Effects” (2020, background): “In DKLITE, the negative log marginal likelihood L (Θ t ) can be rewritten as, where the second equality is achieved by [1,8,14,18,19].”
A Survey on Causal Inference (2020, methodology): “ures that belong in a region of poor overlap in the feature space between treatment and control group. The Bayesian method also can be extended under multi-task model. A nonparametric Bayesian method [3] uses a multi-task Gaussian process with a linear coregionalization kernel as a prior over the ve”
Representation Learning for Treatment Effect Estimation from Observational Data (2018, background): “Estimating the causal effect of an intervention/treatment at individual-level is an important problem that can benefit many domains including health care [12, 1], digital marketing [6, 34, 24], and machine learning [10, 37, 21, 23, 20].”
From Real‐World Patient Data to Individualized Treatment Effects Using Machine Learning: Current and Future Methods to Address Underlying Challenges (2020, methodology): “To create a synthetic observational dataset, parametric functions, such as polynomials or exponentials, are usually used to generate patient outcomes under different treatment options.(21,25) Confounding bias is introduced by assigning treatments according to Bernoulli or Categorical random variable”
Limits of Estimating Heterogeneous Treatment Effects: Guidelines for Practical Algorithm Design (2018, methodology): “Similar ﬁndings were reported for models based on Gaussian processes (Alaa & van der Schaar, 2017), and models based on deep neural networks (Shalit et al., 2017).”
Bayesian Nonparametric Causal Inference: Information Rates and Learning Algorithms (2017, background): “While more recent works have developed machine learning algorithms for estimating individualized treatment effects from observational data in the past few years [2], [5], [8], [14]– [19], the inference machinery built in most of these works seem to be rather ad-hoc.”