Research graph
References from Systematic benchmarking of evaluation paradigms, safety boundaries, and clinical reasoning gaps for multimodal large language models in rehabilitation. Local targets link to admitted publications; unresolved targets remain external evidence.