In theory, Just-in-Time Adaptive Interventions (JiTAIs) are a persuasive technology which promise to empower personal behavioral goals by optimizing treatments to situational context and user behaviors. This paper outlines open challenges facing the development of JiTAIs and discusses the use of modeling as a common ground between behavioral scientists designing interventions and software engineers building applications. We propose that Computational Human Behavior Modeling (CHBM) has the potential to (1) help create better behavioral theories, (2) enable real-time ideographic intervention optimization, and (3) facilitate more robust data analysis techniques. First, a small set of definitions are presented to clarify ambiguities and mismatches in terminology between these two areas. Next, existing modeling concepts are used to formalize a modeling paradigm designed to fit the needs JiTAI development methodology. Last, potential benefits and open challenges of this modeling paradigm are highlighted through examination of the model-development methodology, run-time user modeling, and modelbased data analysis.