PhysB: a framework for designing conversational agents to support physical activity in adolescents.

Conversational Agents (CAs) are a promising tool for enhancing adolescents' physical activity behaviours. To effectively fulfil their role, these agents must address the diverse range of barriers adolescents encounter and provide effective, evidence-based solutions. To achieve this, a theoretical framework titled PhysB is proposed, which is grounded in theory and evidence, allowing for the identification and categorisation of barriers through natural language input. The PhysB framework outlines the relationship between the process of identifying barriers and the cognitive system driving the CA's interaction with adolescents. It incorporates key constructs from various theoretical models and frameworks and draws from behaviour change theory. The PhysB framework's main function is to inform the design and evaluation of personalised, evidence-based conversational AI interventions to promote positive behaviour change in adolescents. The PhysB framework's broad applicability extends to practitioners, policymakers, researchers, and designers involved in designing conversational AI interventions.