STAR

该数据集来源于器官采购与移植网络（OPTN）登记在等待名单上的患者，在去除所有身份标识信息并对日期进行随机偏移处理后实现了匿名化。数据集包含了患者的生物医学信息以及器官/捐赠者的相关信息，并跟踪记录了移植后的随访结果。此外，该数据集还包括了受者与器官/捐赠者的特征，如种族和性别等敏感属性。其中，41.8%的受者出现了移植物衰竭。该数据集的规模涉及160,360名受者，其任务是预测移植物衰竭。This dataset is sourced from waitlisted patients registered with the Organ Procurement and Transplantation Network (OPTN), and has been anonymized by removing all personally identifiable information and applying random offsets to dates. It contains the biomedical information of patients, relevant data of organs and donors, as well as tracked post-transplant follow-up results. Additionally, the dataset includes characteristics of recipients, organs and donors, such as sensitive attributes like race and gender. A total of 41.8% of the recipients experienced graft failure. The dataset covers 160,360 recipients, and its target task is to predict graft failure.

搜集汇总

背景与挑战

背景概述

STAR是一个用于迁移学习的模式引导对话数据集，包含多个JSON格式的对话文件，每个文件记录用户和向导的交互事件及场景信息。该数据集旨在支持对话系统的迁移学习研究，并附带相关论文和MIT许可证。