Multivariate Tie-breaker Designs

TL;DR

This paper introduces a multivariate tie-breaker design framework that balances resource efficiency and statistical power, using convex optimization for treatment assignment, with applications demonstrated on medical triage data.

Contribution

It develops a D-optimality based approach for designing tie-breaker experiments with multivariate responses, incorporating economic and ethical constraints.

Findings

Optimal treatment probabilities can be computed via convex optimization.

Monotonicity constraints induce sparsity in treatment probability assignments.

The approach is demonstrated on real-world triage data from MIMIC-IV-ED.

Abstract

In a tie-breaker design (TBD), subjects with high values of a running variable are given some (usually desirable) treatment, subjects with low values are not, and subjects in the middle are randomized. TBDs are intermediate between regression discontinuity designs (RDDs) and randomized controlled trials (RCTs). TBDs allow a tradeoff between the resource allocation efficiency of an RDD and the statistical efficiency of an RCT. We study a model where the expected response is one multivariate regression for treated subjects and another for control subjects. We propose a prospective D-optimality, analogous to Bayesian optimal design, to understand design tradeoffs without reference to a specific data set. For given covariates, we show how to use convex optimization to choose treatment probabilities that optimize this criterion. We can incorporate a variety of constraints motivated by economic and ethical considerations. In our model, D-optimality for the treatment effect coincides with D-optimality for the whole regression, and, without constraints, an RCT is globally optimal. We show that a monotonicity constraint favoring more deserving subjects induces sparsity in the number of distinct treatment probabilities. We apply the convex optimization solution to a semi-synthetic example involving triage data from the MIMIC-IV-ED database.