MIMIC-Patient
收藏资源简介:
MIMIC-Patient数据集是从MIMICIII电子健康记录(EHRs)中构建的,旨在支持动态的、以患者为中心的模拟。数据集包含从MIMICIII临床数据库中提取的多样医疗信息,每个患者的信息被整理成一个结构化的JSON文件,包含人口统计学、诊断、处方、程序、实验室数据、心电图报告、超声心动图报告、放射学报告和出院摘要等。该数据集的构建过程分为两个阶段:数据选择和临床数据合并。数据选择阶段从MIMIC-III数据库中筛选出符合条件的住院记录,然后随机选取500个住院记录作为数据集。临床数据合并阶段将每个患者的相关数据提取并合并到一个JSON文件中,以便于分析和使用。MIMIC-Patient数据集旨在为动态、交互式的决策任务提供支持,并用于评估DynamiCare框架的有效性和可行性。
The MIMIC-Patient dataset is constructed from MIMIC-III electronic health records (EHRs), aiming to support dynamic, patient-centric simulations. The dataset contains diverse medical information extracted from the MIMIC-III clinical database. Each patient’s information is organized into a structured JSON file, covering demographics, diagnoses, prescriptions, procedures, laboratory data, electrocardiogram reports, echocardiogram reports, radiology reports, discharge summaries, and more. The construction of this dataset is divided into two stages: data selection and clinical data integration. In the data selection stage, eligible hospital admissions are screened from the MIMIC-III database, and then 500 hospital admissions are randomly selected to form the dataset. In the clinical data integration stage, relevant data for each patient is extracted and consolidated into a single JSON file to facilitate analysis and utilization. The MIMIC-Patient dataset is intended to support dynamic, interactive decision-making tasks, and is used to evaluate the effectiveness and feasibility of the DynamiCare framework.




