LLM–powered transcription and implementation of audiovisual radiology reports in elite soccer medicine: A pilot study and two-year clinical impact review

The aim of this study was to evaluate the integration of a large language model (LLM) GPT-4-powered transcription and translation of audiovisual radiology video reports within elite soccer medicine.The study examines the utility of this approach for enhancing diagnostic communication, streamlining clinical workflow, and improving interdisciplinary understanding among sports medicine stakeholders.A dual-phase study was conducted: (1) a pilot implementation of GPT-4 automatic transcription and multilingual translation of screen-recorded narrated video reports in elite soccer athletes, and (2) a retrospective two-year analysis of 530 audiovisual reports evaluating physician feedback on comprehension, decision-making, and workflow integration.GPT-4-generated transcriptions improved report accessibility, especially across language barriers, and enabled simultaneous video, text, and patient-friendly summaries within the EMR.Feedback from ordering physicians demonstrated enhanced comprehension (90%) and faster clinical decision-making (68%) for complex musculoskeletal injuries.However, limitations included integration challenges (29%) and time constraints for full video consumption (41%).Automated transcription and multilingual delivery of audiovisual radiology reports represent a transformative advancement in radiological communication for elite soccer medicine.Continued refinement, including structured summaries and EMR optimization, is essential for broader adoption.