Evaluating the Use of Generalized Dynamic Weighted Ordinary Least Squares for Individualized HIV Treatment Strategies
Résumé
Dynamic treatment regimes (DTR) are a statistical paradigm in personalized medicine which aims to optimize the outcome of long-term treatments. At its simplest, a DTR can optimize for a decision rule which consists of a single treatment; such DTRs are called individualized treatment rules (ITR) and they are often used in optimizing short-term rather than long-term rewards. DTR estimation methods such as dynamic weighted ordinary least squares (dWOLS) offer desirable theoretical advantages such as double robustness of parameter estimates in the decision rules. A recent extension of dWOLS called generalized dWOLS can now accommodate categorical treatments in the estimation of optimal treatment strategies. An application of this novel method will be demonstrated on HIV-infected individuals called low immunological responders (LIR), who are characterized by their low CD4+ T cell counts despite receiving antiretroviral treatment. The administration of Interleukin 7 (IL-7) injections has been shown to increase the concentration of CD4 cells in LIRs, but the optimal number of injections has yet to be determined. In this project, an ITR will be devised to maximize the duration where the CD4+ load is above a healthy threshold (500 cells/µL) while preventing the administration of unnecessary injections.
Origine : Fichiers produits par l'(les) auteur(s)