Digital health AI systems are often evaluated solely based on model performance, but their real-world impact depends on how their design choices shape the experiences, decisions, and workflows of patients and clinicians. We will discuss HCI issues related to the design, evaluation, and deployment of AI-enabled systems in digital health. For example, we will discuss how "easy to implement" AI engineering decisions that are usually default parameters in Python machine learning code can have dramatic effects on the user experience of patients and clinicians. We will discuss how the choice of evaluation metrics to optimize for can affect downstream user experience, and why the traditional data science practice of focusing on precision, recall, and F1-score is misaligned with the user experience goals of clinically-oriented systems. We will discuss how we can leverage AI to optimize the user experience of mobile health systems. Ultimately, this talk will highlight how HCI approaches can ensure that digital health AI-enabled solutions are aligned with the realities of patient and clinician use.