Some methods for heterogeneous treatment effect estimation in high dimensions

By Scott Powers, Junyang Qian, Kenneth Jung, Alejandro Schuler, Nigam H. Shah, Trevor J. Hastie, Robert Tibshirani. 2017

244 citations recorded by the source corpus.

When devising a course of treatment for a patient, doctors often have little quantitative evidence on which to base their decisions, beyond their medical education and published clinical trials. Stanford Health Care alone has millions of electronic medical records that are only just recently being leveraged to inform better treatment recommendations. These data present a unique challenge because they are high dimensional and observational. Our goal is to make personalized treatment recommendations based on the outcomes for past patients similar to a new patient. We propose and analyze 3 methods for estimating heterogeneous treatment effects using observational data. Our methods perform well in simulations using a wide variety of treatment effect functions, and we present results of applying the 2 most promising methods to data from The SPRINT Data Analysis Challenge, from a large randomized trial of a treatment for high blood pressure.

Subfields: Economics and Econometrics, 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 Economics and Econometrics

1. Ahmed M. Alaa
2. Dimitris J. Bertsimas
3. Mihaela van der Schaar
4. Stefan Wager
5. Xinkun Nie

In Statistics and Probability

1. Mihaela van der Schaar
2. Ahmed M. Alaa
3. Jack Dunn
4. Stefan Wager
5. Xinkun Nie
6. Susan Athey
7. Nishanth Mundru
8. Dimitris J. Bertsimas

Citations in the corpus

• Quasi-oracle estimation of heterogeneous treatment effects (2017, methodology): “…2009), BART (Hahn, Murray, and Carvalho, 2017; Hill, 2011), random forests (Athey, Tibshirani, and Wager, 2018; Wager and Athey, 2017), boosting (Powers et al., 2018), neural networks (Shalit, Johansson, and Sontag, 2017), etc., as well as combinations thereof (K¨unzel et al., 2017; Luedtke and…”
• Optimal trees for prediction and prescription (2018, unknown): “[101], who develop nonlinear methods, e.”
• Limits of Estimating Heterogeneous Treatment Effects: Guidelines for Practical Algorithm Design (2018, methodology): “R , and estimated the CATE as ^ T ( x ) = f ( x; 1) (cid:0) f ( x; 0) (Hill, 2011; Johansson et al., 2016; Powers et al., 2017).”
• Optimal Prescriptive Trees (2019, methodology): “Additional methods that build on causal forests are proposed in the recent work of Powers et al. (2017), who develop nonlinear methods to provide better estimates of the personalized average treatment effect, E[Y(1)”
• Bayesian Nonparametric Causal Inference: Information Rates and Learning Algorithms (2017, methodology): “We compared our algorithm with various causal models and standard machine learning benchmarks which we list in what follows: ♣ Tree-based methods (BART [8], [16], [22], causal forests (CF) [7], [36], ♠ Balancing counterfactual regression (balancing neural networks (BNN) [18], and counterfactual regr”