Abstract:
|
Indices of risk reduction are commonly used to assess the effectiveness of treatment. Increase in chance of recovery is a logically equivalent counterpart to risk reduction. The recovery model emphasizes the distinction between capability of spontaneous recovery and sensitivity to compared treatments. Exploiting this may help improve prediction of treatment response, facilitate search for biomarkers, and reduce risk for adverse effects via eliminating unnecessary prescriptions. Spontaneous and treatment-induced recoveries are not phenotypically distinguishable; thus we introduce a hidden variable model. The (unobservable) proportion of patients sensitive to treatment is not equal to the (observable) proportion of "responders"; we introduce a logical framework for this relationship, which constrains the priors for Bayesian analysis. Clinical, experimental, and epidemiological data will be presented to motivate the hidden variable architecture. We focus on the interface between the clinical and statistical aspects of effectiveness assessment, and on the development of a conceptual, structural, and logical framework.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.