Big Data/Machine Learning/AI
Two-Step Pragmatic Subgroup Discovery for Heterogeneous Treatment Effects Analyses: Novel Framework Toward Enhanced Interpretability Toshiaki Komura* Toshiaki Komura Falco J. Bargagli-Stoffi Koichiro Shiba Kosuke Inoue
Background:
Effect heterogeneity analyses using causal machine learning algorithms have gained popularity in recent years. However, the interpretation of estimated individualized effects requires caution because insights from these data-driven approach might not be presented in a way that a human audience can reasonably understand. Thus, a practical framework that integrates advanced machine learning methods and decision-making remains critically needed to achieve effective scientific communication and implementation.
Development:
We introduce a 2-step framework to identify characteristics associated with substantial effect heterogeneity in a format human audiences can reasonably understand (Figure). The proposed framework applies distinct sets of covariates for i) estimation of individualized effects and ii) the discovery of subgroups that show effect heterogeneity based on highly interpretable if-then rules. By referring to existing metrics of interpretability, we describe how each step contributes to leveraging a theoretical advantage of machine learning models while creating an interpretable and practically relevant framework.
Application:
We applied the pragmatic subgroup discovery framework for the Look AHEAD trial to assess practically relevant, detailed, and comprehensive insights into the effect heterogeneities of intense lifestyle intervention for individuals with diabetes on cardiovascular mortality. Our analysis identified i) individuals with a history of cardiovascular disease and coronary artery bypass graft had the least benefit from the intervention (outcome risk [95% CI] = 17.67pp [-26.81, -8.54]), while ii) individuals with no history of CVD and age <60 received the highest benefit (outcome risk [95% CI] = -1.77pp [-1.28, 4.83]).
Conclusions:
The proposed framework can help researchers to discover insights into effect heterogeneity and present the results with enhanced interpretability. Our heterogeneous effects discovery approach could be a generic strategy to ensure effective implementation and scientific communication when applying machine learning algorithms in applied causal inference contexts.
