An information-theoretic approach for the evaluation of continuous surrogate for binary true endpoints based on causal inference

Abstract

This thesis pertains to the general topic of the statistical methods for the evaluation of surrogate endpoints. it also highlights the importance of the validation process such that the surrogate endpoint is reliable in predicting the clinical benefit of the treatment under evaluation. The discussion about some available statistical methodologies and measures for assessing the validity of surrogate endpoints is provided and further is describes an overview of the topics that are covered in this thesis.