Since the COVID-19 pandemic in 2020, telemedicine has become central to healthcare delivery, with physician use rising from 15.4% in 2019 to over 70% in 2024.1 Effective virtual care requires distinct adaptations not addressed by traditional training models, and inadequate communication is associated with reduced patient trust and satisfaction.2, 22 Current training frameworks, such as OSCEs and standardized patients, are designed for face-to-face interactions. However, these approaches do not account for the distinct interpersonal constraints of virtual care. This gap underscores the need for scalable, modality-specific training approaches tailored to telemedicine. This work describes the theoretical framework and design of TellyComm, an AI-enhanced, web-based training model developed to improve physician communication in telemedicine settings. This platform uses brief clinical video vignettes demonstrating effective and ineffective telemedicine behaviors. Each vignette includes embedded pause points prompting users to select appropriate communication strategies. Responses are evaluated using a large language model, guided by established frameworks such as Calgary–Cambridge, NURSE, and SPIKES. Planned evaluation includes pilot implementation within a population of physicians in communication-intensive specialties. Platforms like EQClinic demonstrate improvements in communication scores following virtual interventions. Strategies such as teach-back and deliberate eye contact are key to effective virtual care.10,13,20 However, these approaches have limited generalizability across specialties, rely on self-assessment, and underemphasize non-verbal communication differences in virtual settings. As telemedicine expands, the need for targeted communication training is increasingly evident. Conceptual models like TellyComm represent promising AI-supported approaches to improve both physician communication and patient health outcomes.
