NLPCC 2026 DA-MIVQA: Difficulty-Aware Medical Video QA Benchmark

Following the CMIVQA, MMI-VQA, and M4IVQA challenges from 2023–2025, NLPCC 2026 presents the Difficulty-Aware Medical Instructional Video Question Answering (DA-MIVQA) shared task. This benchmark explicitly grades questions by the type and complexity of evidence needed: simple questions can be answered from subtitle text alone, while complex questions require visual grounding, procedural understanding, and integration across multiple modalities. The dataset is sourced from public medical instructional channels and covers diverse scenarios including first aid, emergency response, rehabilitation, nursing, and general medical education. All queries are manually annotated with difficulty labels to enable fine-grained evaluation of AI systems.

The challenge comprises three tracks: Difficulty-Aware Temporal Answer Grounding in Single Video (DA-TAGSV), Difficulty-Aware Video Corpus Retrieval (DA-VCR), and Difficulty-Aware Temporal Answer Grounding in Video Corpus (DA-TAGVC). The paper (21 pages, 1 figure, 5 tables) provides a comprehensive overview including task motivation, dataset statistics, evaluation protocol, participant summary, competition results, and representative system architectures. DA-MIVQA pushes medical AI assessment beyond simple retrieval by requiring systems to recognize when to rely on text versus when to cross-reference visual and temporal cues—a critical step for real-world clinical decision support.