Most current adaptive gamification approaches use what is often called a "static" adaptation approach-i.e. game elements are adapted to users once, generally before using the gamified tool, based on a static user profile. On the other hand, dynamic adaptation proposes to adapt game elements based on user behaviour in real time, reacting to variations in user engagement. In this paper, we propose an adaptation framework using an initial static adaptation based on learner profiles, and a dynamic adaptation that uses learning analytics to refine the static adaptation recommendations. The adaptation system is able to observe various learning analytics to estimate learner engagement, to compare to that of other learners, and then to signal to teachers learners that require a change in their gamified environment. We propose a protocol for a future study to test our approach in real conditions, and provide some recommendations for future directions.