Teaching Clinicians to Speak AI: A Novel Approach to Prompt Engineering for Medical Scribe Technology (Preprint)

BACKGROUND Physical therapy, like other sectors of U.S. healthcare, faces increasing documentation burden and growing demand for patient-centered care. AI scribes offer a promising solution by automating clinical documentation, but their effectiveness depends heavily on clinicians’ ability to guide these systems through well-designed prompts. Currently, physical therapists lack structured training in this competency. OBJECTIVE This Viewpoint introduces and standardizes the SCRIBE Framework, a systematic, six-step approach to prompt engineering, designed to help physical therapists use AI scribes effectively, ethically, and safely. METHODS The framework was developed through synthesis of literature on prompt engineering, healthcare AI adoption, and documentation standards in rehabilitation. It was refined for relevance to clinical workflows in physical therapy and allied health professions. RESULTS The SCRIBE Framework outlines six practical steps for prompt design: Set the Specialist Role, Clarify Context and Sources, Require Structured Response, Include Clinical Logic, Build in Boundaries, and Establish Voice and Style. Together, these steps enable therapists to create prompts that generate accurate, compliant, and clinically valid documentation. The framework addresses efficiency, ethical risks, and professional communication standards. CONCLUSIONS The integration of AI scribes into physical therapy is inevitable. Prompt engineering represents a new professional competency that shifts therapists from passive users to active conductors of AI documentation. Adoption of the SCRIBE Framework can reduce administrative burden, enhance clinical accuracy, and ensure that AI technologies support patient-centered care in ways that are sustainable and ethically sound.