xpertsystems/hccar003-sample

HCCAR003是一个合成高血压和心血管风险数据集（样本预览），由XpertSystems.ai的Synthetic Data Factory创建，专注于医疗保健/心脏病学领域。这是一个纵向高血压队列数据集，包含季度访问级别的记录，覆盖7个临床模块：血压监测（诊室、家庭、动态血压监测24小时/白天/夜间及夜间下降模式、中心主动脉压、脉搏波速度、增强指数、血压变异性）、抗高血压药物（最多4个药物槽位，8种药物类别，包括ACE抑制剂、ARB、CCB、噻嗪类、β受体阻滞剂等）、生活方式因素（DASH饮食、膳食钠摄入、体力活动METs、睡眠、吸烟、饮酒、压力）、生物标志物（完整血脂谱、hs-CRP、肌酐、eGFR和KDIGO CKD分期、UACR和微量/大量白蛋白尿、电解质、血糖、HbA1c、BNP、肌钙蛋白、ASCVD 10年风险、Framingham风险评分）、终末器官损伤（左心室肥厚、左心室质量指数、LVEF、E/e比率、颈动脉内膜中层厚度、斑块、视网膜病变分级、白质病变体积、腔隙性梗死）以及MACE结果（心肌梗死、缺血性/出血性卒中、短暂性脑缺血发作、心力衰竭住院、新发心房颤动、心血管死亡、全因死亡、研究退出）。数据集基于多个临床指南进行校准，包括ACC/AHA 2017高血压指南、ACC/AHA 2013合并队列方程、KDIGO 2012 CKD分类、ABPM工作组/ESH 2013建议、AHA/ACC顽固性高血压科学声明和ESH/ESC 2018高血压指南。样本预览包含150名患者，约12次季度访问（超过3年），共约1,800条访问记录，数据以CSV和Parquet格式提供。数据集适用于多种用例，如血压控制预测、动态血压监测解释、药物反应预测、MACE生存分析等，但存在一些局限性，例如ASCVD风险计算错误、eGFR使用旧公式等。HCCAR003 is a synthetic hypertension and cardiovascular risk dataset (sample preview) created by XpertSystems.ai’s Synthetic Data Factory, focusing on the healthcare/cardiology field. This is a longitudinal hypertension cohort dataset containing quarterly visit-level records, covering 7 clinical modules: blood pressure monitoring (office, home, 24-hour/daytime/nighttime ambulatory blood pressure monitoring with nocturnal dipping pattern, central aortic pressure, pulse wave velocity, augmentation index, blood pressure variability), antihypertensive medications (up to 4 medication slots, 8 drug categories including ACE inhibitors, ARBs, CCBs, thiazides, beta-blockers, etc.), lifestyle factors (DASH diet, dietary sodium intake, physical activity METs, sleep, smoking, alcohol consumption, stress), biomarkers (comprehensive lipid profile, hs-CRP, creatinine, eGFR and KDIGO CKD staging, UACR and micro/macroalbuminuria, electrolytes, blood glucose, HbA1c, BNP, troponin, 10-year ASCVD risk, Framingham risk score), end-organ damage (left ventricular hypertrophy, left ventricular mass index, LVEF, E/e ratio, carotid intima-media thickness, plaque, retinopathy grading, white matter lesion volume, lacunar infarction), and MACE outcomes (myocardial infarction, ischemic/hemorrhagic stroke, transient ischemic attack, heart failure hospitalization, new-onset atrial fibrillation, cardiovascular death, all-cause death, study withdrawal). The dataset is calibrated based on multiple clinical guidelines, including the ACC/AHA 2017 Hypertension Guideline, ACC/AHA 2013 Pooled Cohort Equations, KDIGO 2012 CKD Classification, ABPM Working Group/ESH 2013 Recommendations, AHA/ACC Scientific Statement on Resistant Hypertension, and ESH/ESC 2018 Hypertension Guideline. The sample preview includes 150 patients, with approximately 12 quarterly visits per patient (spanning over 3 years), totaling around 1,800 visit records. The data is provided in CSV and Parquet formats. This dataset is applicable to various use cases, such as blood pressure control prediction, ambulatory blood pressure monitoring interpretation, drug response prediction, MACE survival analysis, etc. However, it has some limitations, such as errors in ASCVD risk calculation and use of outdated formulas for eGFR.