Maximizing Intervention Effectiveness

Abstract

Frequently, policy makers seek to roll out an intervention previously proven effective in a research study, perhaps subject to resource constraints. However, because different subpopulations may respond differently to the same treatment, there is no a priori guarantee that the intervention will be as effective in the targeted population as it was in the study. How then should policy makers target individuals to maximize intervention effectiveness? We propose a novel robust optimization approach that leverages evidence typically available in a published study. Our model can be easily optimized in minutes for realistic instances with off-the-shelf software and is flexible enough to accommodate a variety of resource and fairness constraints. We compare our approach with current practice by proving performance guarantees for both approaches, which emphasize their structural differences. We also prove an intuitive interpretation of our model in terms of regularization, penalizing differences in the demographic distribution between targeted individuals and the study population. Although the precise penalty depends on the choice of uncertainty set, we show that for special cases we can recover classical penalties from the covariate matching literature on causal inference. Finally, using real data from a large teaching hospital, we compare our approach to common practice in the particular context of reducing emergency department utilization by Medicaid patients through case management. We find that our approach can offer significant benefits over common practice, particularly when the heterogeneity in patient response to the treatment is large. This paper was accepted by Chung-Piaw Teo, optimization.

Public health risks arising from food supply chains: Challenges and opportunities

“…However, new data-driven approaches using "small data" have been emerging that offer a different approach for understanding and estimating causal effects. Gupta et al (2020) consider the problem of allocating a binary intervention among individuals in a population, subject to a budget constraint. In Gupta et al (2020), the authors' goal is to determine the subpopulation that would be most impacted by the treatment and propose a new method that uses only summary statistics that are often found in published studies.…”

“…Gupta et al (2020) consider the problem of allocating a binary intervention among individuals in a population, subject to a budget constraint. In Gupta et al (2020), the authors' goal is to determine the subpopulation that would be most impacted by the treatment and propose a new method that uses only summary statistics that are often found in published studies. Although Gupta et al (2020) focuses on a healthcare setting, the methods could be applied to a variety of settings where large data sets are not available, but summary data is widely available.…”

Data Analytics in Operations Management: A Review

“…The paper shows that, compared to the existing allocation policy at the time, the proposed policy can satisfy the same fairness criteria while resulting in an 8% increase in aggregate quality-adjusted life years. Gupta et al (2017) consider the problem of targeting individuals for a treatment or intervention. Clinical research studies often establish that a treatment has some sort of aggregate benefit in a large population.…”

The foundations of cost-sensitive causal classification

“…In order to improve upon the performance of conventional classification models, cost-sensitive classification models can be adopted to take into account the cost and benefit of the various possible outcomes, or causal classification models can be adopted to take into account the causal effect of an action that is taken on the outcome of interest. The objective in this paper is to merge cost-sensitive and causal classification so as to practically support optimization of resource allocation and maximize intervention effectiveness (Gupta et al 2020).…”