CGM 血糖算法优化 - 溶液模拟数据收集建议

1. 背景

本方案采用标准葡萄糖溶液模拟的方式收集数据，用于优化校准曲线、温度补偿和漂移修正算法。

---

2. 溶液配制

配制不同浓度的葡萄糖标准溶液，覆盖 CGM 监测范围（2.0-25.0 mmol/L）：

---

3. 需要收集的数据类别

3.1 传感器基础数据（每次读数）

3.2 温度特性测试数据

测试温度点：20°C, 25°C, 30°C, 33°C, 37°C, 40°C, 45°C

3.3 时间漂移测试数据

测试周期：14 天（模拟完整使用周期）

3.4 重复性/精度测试数据

CV% (变异系数) 采集方法：

1. 固定条件：同一探针、同一溶液浓度、同一温度
2. 重复测量：连续读取 5-10 次电流值
3. 计算公式：CV% = (标准差 / 平均值) × 100%
4. 记录方式：前 N-1 行 cv_percent 留空，第 N 行计算后填入

示例（同一探针在 5.6 mmol/L 溶液中测 5 次）：

计算：平均值 = 85.16，标准差 = 0.36，CV% = (0.36/85.16) × 100% = 0.42%

判断标准：

3.5 干扰物质测试数据

常见干扰物质及测试浓度：

测试方法：

1. 先在纯葡萄糖溶液中读取稳态电流（current_no_interferent）
2. 加入干扰物质，等待 5-10 分钟至稳态
3. 记录存在干扰时的电流（raw_current）
4. 计算偏差：current_deviation = (current_with - current_without) / current_without × 100%

---

4. CSV 数据记录模板（重点）

4.1 主实验记录表

表头格式：

test_id,sensor_id,batch_id,timestamp,day_of_experiment,solution_bg,solution_actual_bg,raw_current,temperature,impedance,notes

数据示例：

test_id,sensor_id,batch_id,timestamp,day_of_experiment,solution_bg,solution_actual_bg,raw_current,temperature,impedance,notes
TEST-001,CGM-SIM-001,BATCH-202505,2026-05-11T08:00:00Z,1,5.6,5.5,85.3,37.0,5.1,稳态
TEST-002,CGM-SIM-001,BATCH-202505,2026-05-11T09:00:00Z,1,10.0,10.2,152.8,37.0,5.2,稳态
TEST-003,CGM-SIM-001,BATCH-202505,2026-05-11T10:00:00Z,1,2.0,2.1,32.5,37.0,5.0,稳态
TEST-004,CGM-SIM-001,BATCH-202505,2026-05-11T11:00:00Z,1,13.9,14.1,210.5,37.0,5.3,稳态

Excel 手动记录表格：

4.2 温度测试记录表

表头格式：

test_id,sensor_id,target_bg,test_temp,raw_current,steady_state,notes

数据示例：

test_id,sensor_id,target_bg,test_temp,raw_current,steady_state,notes
TEMP-001,CGM-SIM-001,5.6,20.0,62.1,TRUE,
TEMP-001,CGM-SIM-001,5.6,25.0,71.5,TRUE,
TEMP-001,CGM-SIM-001,5.6,30.0,78.3,TRUE,
TEMP-001,CGM-SIM-001,5.6,37.0,85.3,TRUE,
TEMP-001,CGM-SIM-001,5.6,40.0,88.7,TRUE,
TEMP-001,CGM-SIM-001,5.6,45.0,93.2,FALSE,未稳态

4.3 漂移测试记录表

表头格式：

test_id,sensor_id,solution_bg,hour_of_test,raw_current,temperature,notes

数据示例：

test_id,sensor_id,solution_bg,hour_of_test,raw_current,temperature,notes
DRIFT-001,CGM-SIM-001,5.6,0,85.3,37.0,开始测试
DRIFT-001,CGM-SIM-001,5.6,24,83.1,37.0,
DRIFT-001,CGM-SIM-001,5.6,48,81.5,37.0,
DRIFT-001,CGM-SIM-001,5.6,72,79.8,37.0,
DRIFT-001,CGM-SIM-001,5.6,168,76.2,37.0,
DRIFT-001,CGM-SIM-001,5.6,336,72.5,37.0,测试结束

4.4 重复性测试记录表

表头格式：

test_id,sensor_id,solution_bg,repeat_count,raw_current,cv_percent,notes

数据示例：

test_id,sensor_id,solution_bg,repeat_count,raw_current,cv_percent,notes
REPEAT-001,CGM-SIM-001,5.6,1,85.3,,
REPEAT-001,CGM-SIM-001,5.6,2,84.9,,
REPEAT-001,CGM-SIM-001,5.6,3,85.7,,
REPEAT-001,CGM-SIM-001,5.6,4,85.1,,
REPEAT-001,CGM-SIM-001,5.6,5,84.8,0.42,计算CV%

4.5 干扰物质测试记录表

表头格式：

test_id,sensor_id,solution_bg,interferent_name,interferent_conc,raw_current,current_no_interferent,current_deviation,notes

数据示例：

test_id,sensor_id,solution_bg,interferent_name,interferent_conc,raw_current,current_no_interferent,current_deviation,notes
INTERF-001,CGM-SIM-001,5.6,尿酸,5.0,87.2,85.3,2.2,
INTERF-001,CGM-SIM-001,5.6,抗坏血酸,10.0,89.5,85.3,4.9,
INTERF-001,CGM-SIM-001,5.6,对乙酰氨基酚,15.0,86.1,85.3,0.9,
INTERF-001,CGM-SIM-001,5.6,水杨酸,20.0,85.8,85.3,0.6,
INTERF-001,CGM-SIM-001,5.6,甘露醇,30.0,85.5,85.3,0.2,

Excel 手动记录表格：

4.6 探针校准参数记录表

表头格式：

sensor_id,batch_id,factory_slope,factory_intercept,calibrated_slope,calibrated_intercept,calibration_r_squared,calibration_points,calibration_date,calibration_method

数据示例：

sensor_id,batch_id,factory_slope,factory_intercept,calibrated_slope,calibrated_intercept,calibration_r_squared,calibration_points,calibration_date,calibration_method
CGM-SIM-001,BATCH-202505,0.065,-0.5,0.068,-0.3,0.998,8,2026-05-11T08:00:00Z,linear
CGM-SIM-002,BATCH-202505,0.062,-0.4,0.064,-0.2,0.997,8,2026-05-11T09:00:00Z,segmented

Excel 手动记录表格：