Large Language Models for Generating Interactive Simulations in Anatomy Education

Introduction
Human anatomy forms the foundation of medical education and has traditionally been taught through cadaveric dissection and prosected specimens. Over time, anatomy teaching has evolved with the integration of imaging modalities such as X-ray, CT, and MRI, as well as digital tools including three-dimensional (3D) software, virtual reality, and augmented reality. More recently, artificial intelligence (AI) and large language models (LLMs) have been explored for applications such as personalised learning, tutoring systems, and assessment support. However, the potential use of LLMs to generate interactive simulation-based learning resources remains underexplored.

Methods
This study investigated whether LLM platforms can support the development of interactive web-based anatomy simulations from structured expert input. Three models—Codex, Gemini, and Claude—were evaluated for their ability to translate anatomy knowledge into code capable of generating interactive simulation modules. Generated outputs were refined and implemented within a simple web-based environment. The simulations were then qualitatively reviewed by anatomy educators to assess clarity, clinical relevance, and educational usability.

Results
All three LLM platforms demonstrated the ability to convert structured anatomical knowledge into functional interactive simulation scripts. The generated outputs successfully supported interactive learning tasks linking clinical signs to underlying anatomical mechanisms. While LLMs showed strong capability in generating schematic representations and interactive logic, they remain limited in producing accurate anatomical imagery for complex visual simulations. Educator feedback highlighted the potential value of these tools for developing innovative learning resources.

Conclusions
LLMs show promising potential as tools for generating interactive simulation-based learning materials in anatomy education. When combined with expert oversight, this approach may offer a scalable method for educators to develop structured simulations while promoting responsible and innovative use of AI in medical teaching.