Applications of Generative AI in the Field of Thyroidology: a Narrative Review

Generative artificial intelligence (GenAI), particularly large language models (LLMs), has rapidly advanced in capability and accessibility.In thyroidology, where diagnosis, management, and patient education involve complex multimodal data, GenAI presents promising opportunities for clinical workflow support.This narrative review explores current applications, strengths, and limitations of GenAI technologies in thyroid medicine.This narrative review summarizes studies examining GenAI applications in thyroidology, including risk stratification, report structuring, patient communication, multidisciplinary decision-making, thyroid function test interpretation, and medical image generation.LLMs demonstrated potential in interpreting thyroid ultrasound reports and stratifying risk using Thyroid Imaging Reporting and Data System features, with ChatGPT achieving 86.7% sensitivity for high-risk nodules.Prompt engineering and hybrid artificial intelligence models improved guideline alignment and interpretability.In patient communication, GenAI tools showed high factual accuracy and readability, though limitations remained in empathy and individualized contextualization.For thyroid function test interpretation, models exhibited acceptable performance in overt disease but remained unreliable in subtle cases.Synthetic image generation using diffusion models enabled photorealistic visualizations of thyroid-related phenotypes for education and communication.GenAI is poised to augment various aspects of thyroid care, but current models require further refinement, clinical validation, and human oversight before integration into frontline clinical practice.Ongoing improvements in prompt design, domain-specific fine-tuning, and multimodal reasoning will be essential to realize GenAI's full potential in thyroidology.