Evaluation of fine-tuned small language models (SLMs) vs large language models (LLMs) in modeled patient conversations for lung cancer screening.