Proactive service migration is critical for ensuring Quality of Experience (QoE) in Mobile Edge Computing (MEC). However, existing methods are often intent-agnostic; by relying solely on physical mobility, they ignore the dynamic user interests that fundamentally drive service needs, leading to significant resource wastage. To address this, we propose FUSE, a novel framework that synergistically fuses user intent prediction (instantiated via Point-of-Interest forecasting) with trajectory prediction. The core of FUSE is a dual-adaptive decision engine that translates behavioral uncertainty into an optimal migration policy, determining not only what and where to migrate, but also the ideal quantity and path. Experiments on real-world datasets show that FUSE achieves a superior balance between service availability and resource efficiency, significantly outperforming state-of-the-art baselines.