Research Article
François Modave and Nav
Abstract
Informatics-based decision-making aids are becoming an essential component of clinical care from both a physician and patient perspective. Although additive approaches are used with a certain degree of success in a medical context, they often suffer from an inability to conveniently represent dependencies, which is certainly desirable in practice. To address this drawback, we present the concepts of non-additive measures, and non-additive integration, as well as Shapley values and interaction indices to a clinical framework, and show how they can be used to develop robust and reliable computing tools that support informed and shared decision-making. We also present an extension of these tools that allow us to manage the inherent uncertainty and imprecision of data, and help us address value clarification. To set ideas, we focus on presenting algorithms to improve shared decision-making for colorectal cancer screening, however, the framework presented here is general, and can be applied to a wide variety of clinical decision problems.