Another good option is the constraint model, which could theoretically be implemented into the agent based modeling system. Basically the constraint model takes a lot of inputs of different variables, then spits out what results could or could not happen. This could be an important way of testing how certain risk factors will affect the spread of disease. For example, students might be able to explore the effectiveness of bug nets or fogging on disease transmission, or look into how a particularly rainy rainy season might affect levels of standing water which might affect mosquito population. Creating these models would force students to think critically about the complex relationships between all the players in a given system and demonstrate how tiny changes in one area can have huge changes downstream in ways that wouldn't seem obvious at first.
I think one thing that could be a hurdle is not teaching students enough about Zika or about the mosquito vectors before setting them loose on model creation. A key part of this process that I think is somewhat glossed over is the research and legwork needed to gather the tools and build the foundation of successful models. This research could be tied in with the modeling process "What do we need to know about x to implement it into our model?" but I feel like it could be very easy to get caught up in the apparent success of modeling that we could gloss over some of the more foundational aspects of learning.