For the healthcare industry, predictive analytics offer revolutionary benefits for improving patient outcomes, reducing hospital readmissions, and lowering treatment costs. The increasing adoption of electronic health records allows the modeling of laboratory results, medications, and socio-economic data, as well as mental health, among others. We emphasize the opportunities that generative models offer for predictive healthcare analytics and the necessity for healthcare analytics to contextualize data relationships. We analyze predictive models, understand our contextual data relationships, interpret our results, expose them, and understand why models are learning certain relationships. We make use of benchmark data and case studies to illustrate our points. Our discussion concludes by offering a framework and a departure point for future related research.