Generative AI in telemedicine: clinical validation, ethics, and implementation to optimize remote diagnosis and adherence.

Abstract

This mixed-methods study investigates the implementation of Generative Artificial Intelligence (GenAI) in telemedicine, focusing on optimizing remote diagnosis and therapeutic adherence. Through a systematic review, expert interviews, and two experimental studies, it validates GenAI as an effective "copilot," improving physicians' diagnostic accuracy by 15% and efficiency by 30%, while AI-generated care plans significantly increase patient adherence and understanding. However, the core findings identify three critical and interdependent pillars for responsible adoption: Validation (requiring evidence of net clinical benefit beyond technical accuracy), Ethics (where explainability is revealed as a clinical facilitator, not merely a moral imperative), and Implementation (demanding a human-centric redesign of workflows). The proposed V-E-I Framework is an iterative model integrating these pillars, arguing that success depends on advancing them simultaneously and in balance. The conclusion emphasizes that the future of GenAI in healthcare is not guaranteed by technology alone, but by responsible governance that prioritizes safety, equity, and real clinical utility over mere innovation.