Towards integration-preserving customization of just-in-time adaptive interventions with composite clabjects in RDF and SHACL

Abstract

Just-in-time adaptive interventions (JITAIs) aim at health-promoting behavior change of individuals. Moving the development and evaluation of JITAIs beyond custom implementations for each specific use case will require integration-preserving customization, i.e., adaptation to different studies and participants without compromising integration for data analysis. For this purpose we develop a multilevel modeling (MLM) approach that builds on two-level structural conceptual models with composition and specialization extended by Cardelli power types yielding hierarchies of composite clabjects. We show the practical applicability of the approach through modeling of an example study on JITAIs for a digital health intervention, and demonstrate an RDF- and SHACL-based implementation.