xpertsystems/hccar004-sample

HCCAR004是一个合成的心房颤动（AFib）患者数据集，包含两个纵向表格：baseline（基线表）和visits（随访表）。基线表有150行（样本中），每行代表一个AFib患者，包含127个特征，覆盖9个临床模块：人口统计学、AFib概况、CHA2DS2-VASc和HAS-BLED评分、抗凝管理、心律控制、心电图/电生理学、生物标志物、设备监测和5年结局。随访表有1,800行，记录了12次季度随访（跨3年），在基线数据基础上对afib_burden_pct、nt_probnp_pg_ml和ventricular_rate_bpm_controlled三个字段进行了小幅扰动，适用于随访级分析和轨迹队列研究。数据集基于ESC 2020、AHA/ACC 2019等临床指南校准，并针对CHA2DS2-VASc验证、DOAC选择、导管消融成功率预测等多种使用案例设计。数据以CSV和Parquet格式提供，通过patient_id连接。

HCCAR004 is a synthetic atrial fibrillation (AFib) patient dataset consisting of two longitudinal tables: baseline and visits. The baseline table contains 150 rows (in the sample), each representing an AFib patient with 127 features across 9 clinical modules: demographics, AFib profile, CHA2DS2-VASc and HAS-BLED scores, anticoagulation management, rhythm control, ECG/electrophysiology, biomarkers, device monitoring, and 5-year outcomes. The visits table has 1,800 rows, recording quarterly visits over 12 visits × 3 years, with small perturbations on afib_burden_pct, nt_probnp_pg_ml, and ventricular_rate_bpm_controlled from baseline, useful for visit-level analytics and trajectory cohort analysis. The dataset is calibrated against clinical guidelines such as ESC 2020 and AHA/ACC 2019, and designed for use cases like CHA2DS2-VASc validation, DOAC selection ML, and catheter ablation success prediction. Data is provided in CSV and Parquet formats, joinable on patient_id.