Sarcouncil Journal of Multidisciplinary

Abstract: Background: The introduction of Google DeepMind’s Gemini series represents a paradigm shift in medical informatics, transitioning from text-centric Large Language Models (LLMs) to natively multimodal Large Multimodal Models (LMMs). This bibliometric study aimed to quantify the integration, research trends, and academic impact of Gemini in the medical sector from its inception to 2025. Methods: A bibliometric analysis was performed using data from PubMed, Scopus, and Web of Science. The search strategy focused on keywords such as "Google Gemini" and "multimodal AI medicine" for the period of 2023–2025. Analytical tools, including VOSviewer and CiteSpace, were used to map publication velocity, geographic contributions, and keyword co-occurrence. Results: The scientific output increased significantly from 12 publications in 2023 to 126 in 2025. The United States leads global production (35%), followed by China (27%) and the United Kingdom (14%). The primary research themes included Clinical Decision Support (28%) and Medical Imaging (24%). While initial research focused on benchmarking against GPT-4, recent trends have shown a pivot toward "clinical-vision" tasks, such as interpreting histology and radiology images, alongside longitudinal record analysis via Gemini’s long-context window. Conclusion: Gemini is rapidly establishing a dominant role in multimodal diagnostics and large-scale EHR processing. Despite its potential to alleviate administrative burdens and enhance diagnostic accuracy, challenges regarding interpretability (i.e., "black box" nature), ethical bias, and the need for large-scale real-world clinical validation persist. Gemini is poised to fundamentally shape the future of interdisciplinary, AI-driven healthcare.