Connectome-based predictive modeling is powerful approach to model individual differences in phenotypes using fMRI data. One limitation of this approach is that models are complex, making it challenging to target these models with interventions. Here, I will discuss our recent work in creating these models and developing potential approaches to target them.