检测数据准确性 Test data accuracy

gaowenhua

什么是检测数据准确性？

准确性（Accuracy） 是指检测结果与真实值（或参考值）之间的接近程度。

• 准确性高 = 测量值接近真实值
• 准确性低 = 存在系统误差（偏差）
  

✅ 与“精密度”（Precision）区别：

• 精密度：重复测量的一致性（随机误差）
• 准确性：接近真值的程度（系统误差 + 随机误差）
  

What is testing data accuracy?
Accuracy refers to the degree of closeness between the detection result and the true value (or reference value).


High accuracy=measured values close to true values
Low accuracy=presence of systematic error (bias)
✅  Difference from "Precision":


Precision: Consistency of repeated measurements (random error)
Accuracy: The degree to which it approaches the true value (systematic error+random error)

gaowenhua

影响检测数据准确性的主要因素因素类别具体原因
人员因素操作不规范、读数错误、样品处理不当
设备因素仪器未校准、老化、分辨率不足
方法因素方法不适用、标准曲线错误、干扰物质存在
环境因素温湿度波动、振动、电磁干扰
样品因素样品不均匀、污染、保存不当
标准物质使用过期或不合格的标准品/对照品



The main factors affecting the accuracy of detection data
Specific reasons for factor categories
Personnel factors such as improper operation, incorrect readings, and improper sample handling
Equipment factors: instruments not calibrated, aging, insufficient resolution
Method factors: Inappropriate method, incorrect standard curve, presence of interfering substances
Environmental factors such as temperature and humidity fluctuations, vibration, and electromagnetic interference
Sample factors: Uneven sample, contamination, and improper storage
Expired or unqualified standard/reference materials used as reference materials

gaowenhua

仪器校准与期间核查

• 定期校准：由有资质的机构对检测设备进行周期性校准（如每年一次），并获取校准证书。
• 期间核查：在校准周期内，使用标准物质或控制样定期检查仪器稳定性。
• 标识管理：贴上“合格/准用/停用”标签，防止误用未校准设备。
  

📌 示例：pH计每天使用前用标准缓冲液校准。

---

2. 使用标准物质与对照实验

• 使用有证标准物质（CRM） 进行比对测试，验证方法准确性。
• 开展加标回收实验（Spiking Recovery）：
  • 在样品中加入已知浓度的标准物质
  • 检测后计算回收率
  • 回收率应在合理范围（如90%~110%）
    
  
  

🔍 回收率 = （检测值 - 原始值） / 加入量 × 100%

---

3. 方法验证与确认

新方法或变更方法时，必须进行方法学验证，评估其准确性：

验证指标说明
准确度通过加标回收或与标准方法比对验证
精密度重复性、再现性测试
线性范围标准曲线的线性相关系数（R² ≥ 0.99）
检出限（LOD）与定量限（LOQ）方法能检测到的最低浓度
特异性是否受干扰物质影响

📌 依据标准：ISO/IEC 17025、ICH Q2、CNAS-CL01等。

---

4. 参与能力验证与实验室间比对

• 能力验证（PT, Proficiency Testing）：参加外部机构组织的盲样测试，评估实验室检测能力。
• 实验室间比对：与同行实验室对同一样品进行检测，比较结果一致性。
• 结果评价：Z值 ≤ 2 为满意，2 < Z ≤ 3 为可疑，Z > 3 为不满意。
  

---

5. 实施质量控制（QC）程序

在日常检测中加入质量控制样，实时监控准确性：

QC手段说明
空白样检查试剂或环境是否污染
平行样同一样品测两次，评估精密度
控制样（Control Sample）已知浓度的标准样，监控长期准确性
质控图（Levey-Jennings图）绘制控制样结果趋势，发现漂移或失控

✅ 规则示例：Westgard规则（如1₂s、1₃s、2₂s等）判断是否失控。

Instrument calibration and periodic verificationRegular calibration: Qualified institutions conduct periodic calibration of testing equipment (such as once a year) and obtain calibration certificates.Periodic verification: During the calibration cycle, use standard substances or control samples to regularly check the stability of the instrument.Identification management: Attach "qualified/approved/discontinued" labels to prevent misuse of uncalibrated equipment.📌  Example: The pH meter is calibrated with standard buffer solution before daily use.2. Use standard substances and control experimentsUse certified reference materials (CRM) for comparative testing to verify the accuracy of the method.Conduct Spiking Recovery experiment:Add standard substances of known concentration to the sampleCalculate the recovery rate after testingThe recovery rate should be within a reasonable range (such as 90%~110%)🔍  Recovery rate=(detection value - original value)/amount added x 100%3. Method validation and confirmationWhen new or changed methods are used, methodological validation must be conducted to evaluate their accuracyVerification indicator descriptionAccuracy is verified through spiked recovery or comparison with standard methodsPrecision repeatability and reproducibility testingThe linear correlation coefficient of the linear range standard curve (R ² ≥ 0.99)The lowest concentration that can be detected by the detection limit (LOD) and quantification limit (LOQ) methodsIs specificity affected by interfering substances📌  According to standards such as ISO/IEC 17025, ICH Q2, CNAS-CL01, etc.4. Participate in proficiency testing and inter laboratory comparisonProficiency Testing (PT): Participate in blind sample testing organized by external organizations to evaluate laboratory testing capabilities.Inter laboratory comparison: Testing the same sample with peer laboratories to compare the consistency of the results.Result evaluation: A Z-value ≤ 2 is considered satisfactory, 2<Z ≤ 3 is considered suspicious, and Z>3 is considered unsatisfactory.5. Implement Quality Control (QC) proceduresAdd quality control samples to daily testing to monitor accuracy in real-time:Explanation of QC MethodsBlank sample check for contamination of reagents or environmentMeasure the same sample twice in parallel to evaluate the precisionControl Sample: A standard sample with known concentration for monitoring long-term accuracyQuality control chart (Levey Jennings chart) shows the trend of control sample results and identifies drift or loss of control✅  Rule example: Westgard rules (such as 1 ₂ s, 1 ∝ s, 2 ₂ s, etc.) determine whether the control is out of control.

gaowenhua

三级审核制度

• 检测人员自审 → 组长复核 → 质量负责人批准
• 重点检查：原始记录、计算过程、标准依据、质控结果
  

2. 异常数据处理流程

• 标识可疑数据
• 查阅原始记录和仪器日志
• 必要时重新检测或复测
• 记录偏差原因和纠正措施（CAPA）
  

Three level review system
Self review by testing personnel → review by team leader → approval by quality manager
Key inspection: original records, calculation process, standard basis, quality control results
2. Abnormal data processing flow
Identify suspicious data
Refer to original records and instrument logs
Retest or retest if necessary
Record the reasons for deviations and corrective actions (CAPA)

gaowenhua

人员培训与能力建设

• 定期培训检测人员的操作技能和标准理解能力
• 实施上岗考核和年度能力评估
• 建立技术人员档案（培训记录、考核结果、授权范围）
  

Personnel training and capacity building
Regularly train testing personnel on their operational skills and understanding of standards
Implement on-the-job assessment and annual competency evaluation
Establish technical personnel files (training records, assessment results, authorization scope)

whping

谢谢分享