This research demonstrates how domain-specific fine-tuning can transform generic LLMs into specialized tools for radiation oncology workflows.

• Successfully improved LLM performance on treatment regimen generation and other radiation oncology tasks
• Showed that even open-source models can achieve clinically useful capabilities when properly fine-tuned
• Established a practical framework for adapting AI language models to specialized medical domains

The findings open new possibilities for AI-assisted clinical workflows in cancer treatment, potentially reducing physician workload while maintaining accuracy in critical treatment planning processes.

Fine-Tuning Open-Source Large Language Models to Improve Their Performance on Radiation Oncology Tasks