Sensory intervention is an emerging type of digital intervention pioneered by the Human-Computer Interaction (HCI) and the Ubiquitous Computing Community (UbiComp). Despite the proliferation of digital and mobile behavioral health applications, the real-world impact of these interventions remains limited. Users often fail to engage with these interventions during distress when support is most needed. Sensory interventions address this challenge by achieving effectiveness with minimal effort. Across four studies, this dissertation examines the end-to-end translational pipeline for sensory interventions. It begins by examining the translation process of offset heart rate biofeedback, showcasing two studies that bridge an in-laboratory proof-of-concept to a closed-loop intervention system evaluated in the wild. Then, it returns to the initial stage of intervention development, applying the lessons learned during the translation pipeline to inform in-lab studies. Together, this dissertation offers both conceptual and technical advances sensory interventions. It synthesizes three key evaluation aspects, specific efficacy, mechanism, and system design implications, for sensory intervention researchers to investigate using their existing methodology. It showcases how anchoring research on these three aspects can support translation and expanding the impact of sensory interventions. In doing so, this dissertation envisions a future in which behavioral health support is seamlessly integrated into everyday life, empowering individuals with timely, unobtrusive, and effective tools in difficult times.