血糖测量数据集

1、数据采集：通过医卫通AI一体机智能建档，在用户授权下依法合规采集用户的空腹血糖数据。原始数据包括唯一设备标识（健康设备ID）、系统生成的匿名用户标识（健康设备监测用户ID）、空腹血糖值（单位：mmol/L）、测量时间戳（精确到分钟，即“血糖采集时间”）及系统自动生成的状态标记（处理异常标识）。
2、数据处理：1）以“健康设备ID”和“血糖采集时间”为联合主键，对原始数据进行清洗，剔除格式错误或明显超出生理范围的异常值（如血糖<2.0或>30.0 mmol/L）；2）计算多维衍生指标：个体维度——日均/周均空腹血糖、血糖控制达标率（3.9–6.1 mmol/L）、高血糖事件频次（≥7.0 mmol/L）、低血糖风险事件（≤3.9 mmol/L）；设备维度——日均服务人数、设备使用频次、异常上报率；群体维度——区域糖尿病筛查阳性率、空腹血糖异常人群分布；3）生成结构化、标准化的血糖测量数据集，支持趋势分析、风险预警与模型训练。
3、数据应用：输出高质量的多用户血糖时序数据，用于生成个人血糖趋势报告、触发高/低血糖自动预警、辅助医生调整治疗方案、评估社区糖尿病规范管理率及训练糖尿病并发症

1. Data Collection: Fasting blood glucose data is collected legally and compliantly via the Yiweitong AI integrated machine for intelligent medical record filing with user authorization. The raw data includes: unique device identifier (health device ID), system-generated anonymous user ID (health device monitoring user ID), fasting blood glucose value (unit: mmol/L), measurement timestamp accurate to the minute (referred to as "blood glucose collection time"), and system-generated status mark (processing exception flag).
2. Data Processing: 1) Clean the raw data with "health device ID" and "blood glucose collection time" as the composite primary key, eliminating outliers with format errors or values obviously outside the physiological range (e.g., blood glucose <2.0 or >30.0 mmol/L); 2) Calculate multi-dimensional derived indicators:
- Individual dimension: daily/weekly average fasting blood glucose, blood glucose control compliance rate (3.9–6.1 mmol/L), frequency of hyperglycemia events (≥7.0 mmol/L), hypoglycemia risk events (≤3.9 mmol/L);
- Device dimension: daily average number of served users, device usage frequency, abnormal reporting rate;
- Population dimension: regional diabetes screening positive rate, distribution of people with abnormal fasting blood glucose;
3) Generate a structured and standardized blood glucose measurement dataset that supports trend analysis, risk early warning and model training.
3. Data Application: Output high-quality multi-user time-series blood glucose data, which is used to generate individual blood glucose trend reports, trigger automatic early warnings for hyperglycemia and hypoglycemia, assist doctors in adjusting treatment plans, evaluate the standardized diabetes management rate in communities, and train models for diabetes complications.