LLMs as emerging tools for understanding and managing bone metastasis and cancer-induced bone pain

Artificial intelligence (AI) chatbots powered by large language models (LLMs) show promise in human-like responses, yet their utility for complex biomedical issues remains uncertain. We evaluated four leading LLMs, including ChatGPT-5.2, Claude 4.5 Sonnet, Gemini 3.0 Pro, and DeepSeek-R1, on their performance in addressing queries related to bone metastasis (BM) and cancer-induced bone pain (CIBP). Using 18 structured questions derived from bibliometric analysis, we assessed responses across research, clinical, and patient perspectives. Qualitatively rating based on accuracy, readability, completeness, conciseness, and empathy demonstrated that LLMs efficiently provide intuitive information regarding BM and CIBP. However, due to limitations such as data bias, weak interpretability, and ambiguous accountability, these tools should be regarded as supportive assistants under human experts' supervision. Although LLMs hold substantial promise for the medical field, it is essential to optimize model architecture and develop comprehensive evaluation standards for specialized domains.