Synth-SBDH
收藏资源简介:
Synth-SBDH是一个由麻省大学阿默斯特分校信息与计算机科学学院和美国退伍军人事务部联合创建的合成数据集,专注于社会和行为健康决定因素(SBDH)。该数据集包含8,767个样本,覆盖15个SBDH类别,并提供详细的注释理由。创建过程中,利用大型语言模型(LLM)GPT-4生成数据,并通过人工监督确保数据质量。Synth-SBDH的应用领域广泛,旨在通过自动化提取SBDH信息,改善患者护理和人群健康管理,特别是在资源有限的环境中显示出其有效性。
Synth-SBDH is a synthetic dataset jointly created by the School of Information and Computer Sciences at the University of Massachusetts Amherst and the United States Department of Veterans Affairs, focusing on Social and Behavioral Determinants of Health (SBDH). The dataset contains 8,767 samples covering 15 SBDH categories, and provides detailed annotation rationales. During its development, the large language model (LLM) GPT-4 was utilized to generate the data, with human supervision employed to ensure data quality. Synth-SBDH has a wide range of application scenarios, aiming to improve patient care and population health management through automated extraction of SBDH information, and its effectiveness has been particularly demonstrated in resource-constrained environments.

- 1Synth-SBDH: A Synthetic Dataset of Social and Behavioral Determinants of Health for Clinical Text麻省大学阿默斯特分校信息与计算机科学学院 · 2024年



