Distilling complex treatment effects into simple public health targeting rules
Conditional Average Treatment Effect Distillation for Hypothetical State-Intervention Targeting Under Budget Constraints
Machine Learning
Summary
Deciding who should receive health interventions like weight loss programs is challenging because benefits vary among individuals and resources are limited. The authors studied ways to simplify complex models that predict these benefits into easier-to-use rules while still targeting people effectively. They tested their approach on data about obesity, blood sugar, and smoking history, showing that simpler rules can closely match the performance of complex models. This method helps turn detailed predictions into practical decision tools for public health, although further real-world testing is needed.
What this means in practice
- •For public health planners: Create simple, interpretable rules to allocate health interventions under budget limits using complex causal effect models.
- •For health policy analysts: Evaluate hypothetical public-health intervention policies on observational data to balance resource use and individual benefit.
Authors
Vasundhara Acharya, Bülent Yener
Abstract
Personalized public-health targeting must capture heterogeneous intervention benefits while remaining interpretable under resource constraints. Flexible causal-effect models can be hard to operationalize, while coarse subgroup policies may lose individualized effect information. We study budget-constrained hypothetical state interventions in observational data without a logged historical decision policy. For obesity, elevated glucose, and lifetime smoking-history states, causal-structure-informed pretreatment covariates are prespecified, with capacity capped at 70% of those observed in the adverse state. We evaluate causal distillation where Causal Forest and Bayesian Causal Forest teachers estimate conditional average treatment effects (CATEs), which decision tree and random forest students approximate. We compare individualized policies with unsupervised K-means subgroup policies and supervised causal-tree policies with leaf counts matched to K-means cluster counts. Observational policies are assessed on held-out data using a doubly robust estimator, bootstrap uncertainty, and decision-focused teacher-student fidelity measures. Across PIMA obesity and glucose and NHANES smoking-history experiments, each distilled student was within 0.011 absolute utility of its teacher, with allocation Jaccard similarities of 0.738-0.937. K-means and causal-tree policies were also competitive, with the highest point-estimate policy varying by application. Prespecified paired comparisons were nonsignificant after Holm adjustment, while near-budget-boundary agreement was less stable than global fidelity for some teacher-student pairs. Findings support causal distillation for translating flexible effect models into simpler budgeted targeting rules, although observational estimates depend on causal-identification assumptions and require longitudinal or prospective validation.