Artificial intelligence (AI)-enabled medical devices require governance models that adapt to continuous learning, evolving functionality, and shifting regulatory expectations. This study advances an adaptive governance framework for AI in digital health that integrates regulatory, ethical, and societal perspectives, with specific reference to the EU AI Act as the emerging regulatory benchmark. Rather than interpreting governance as a linear compliance trajectory, the framework reconceptualises it as an interconnected, feedback-driven system in which regulators, developers, clinicians, patients, and ethicists collaboratively shape oversight mechanisms. Each stakeholder contributes to iterative cycles of transparency, algorithmic validation, and accountability, ensuring that learning systems remain aligned with ethical norms and public interest. This approach underscores that innovation and governance are not opposing imperatives but co-dependent processes that reinforce trust and sustainability in healthcare AI ecosystems. To illustrate practical application, the framework is applied to a real-world scenario of adaptive AI diagnostics in remote patient monitoring, demonstrating how distributed audit loops and multi-stakeholder review boards can operationalise compliance while preserving agility and responsiveness. The framework further addresses the feedback dependencies that determine implementation success, acknowledging how regulatory delays, ethical tensions, and clinical adoption barriers interact dynamically. By linking ethical reflection, stakeholder inclusion, and regulatory adaptability, this work contributes to a more resilient paradigm for AI integration that evolves in sync with both regulatory standards and clinical innovation, ultimately supporting trustworthy deployment of AI-driven healthcare solutions.