A0996
Title: All else being equal: Implications of measurement error for precision medicine and health equity
Authors: Michael Wallace - University of Waterloo (Canada) [presenting]
Abstract: Precision medicine describes the tailoring of treatment decisions to individual-level characteristics. Dynamic treatment regimes operationalize precision medicine through sequences of decision rules which take patient-level data as input and output treatment recommendations. Estimation of decision rules that optimize some outcome across a population based on observational data is a large - and expanding - area of the literature. A common assumption within this framework (as well as in the broader causal inference literature) is that observed data are measured without error, which, in reality, is seldom the case. Moreover, measurement error poses some unique challenges within the context of precision medicine, such as when there is nonadherence to personalized treatment regimes or when treatment decisions are based on error-prone variates. In addition to mis-estimated optimal decision rules, a further concern arises in the context of health equity. Namely, an individual who identifies in one social (or other type of) group may be disproportionately affected if the results of an analysis based on error-prone data are implemented. The challenges are discussed and explored at the interface of precision medicine and measurement error, as well as their potential implications for health equity. In addition to theoretical results, illustrative examples are demonstrated via simulation and an R Shiny app.
