LLM-TB-VQA
收藏资源简介:
LLM-TB-VQA数据集是一个详细的医学依从性视觉问答(VQA)数据集,包含806个自定义标注的结核病(TB)药物监测视频,由临床专家标注。数据集涵盖了积极、消极和模糊的依从性案例。该数据集旨在通过分析患者的面部、药物、摄入水量和吞咽动作等视觉特征,与标题中的相关医疗概念进行关联,以促进视觉和语言特征表示的校准,并提高多模态交互。数据集分为训练集和验证集,以评估模型在药物依从性识别、问答和行为分析方面的性能。
The LLM-TB-VQA dataset is a comprehensive medical adherence visual question answering (VQA) dataset. It contains 806 manually annotated tuberculosis (TB) medication monitoring videos, with annotations completed by clinical experts. The dataset covers positive, negative, and ambiguous adherence cases. This dataset aims to promote the calibration of visual and linguistic feature representations and enhance multimodal interaction by analyzing visual features such as patients' facial expressions, medications, water intake volume, and swallowing movements, and associating these features with relevant medical concepts in the corresponding titles. The dataset is divided into training and validation subsets to evaluate model performance on medication adherence recognition, question answering, and behavior analysis.

- 1AdCare-VLM: Leveraging Large Vision Language Model (LVLM) to Monitor Long-Term Medication Adherence and Care佐治亚大学 · 2025年



