目的 建立基于β-容度容忍区间(β-CTI)的血塞通注射液近红外(NIR)定量分析不确定度评估方法。方法 采用"I×J×K"型析因实验设计法获取NIR分析校正和验证实验数据,通过PLS算法建立NIR定量模型,通过β-CTI估计NIR定量分析不确定度,并与常规β-期望容忍区间(β-ETI)进行比较。采用lnU=a+b lnY模型对相对扩展不确定度(U)与相应质量浓度(Y)之间的关系进行拟合,设置不确定度可接受限(λ)=20%,并绘制不确定度决策曲线。结果 在考察的7个质量浓度水平中,除1.485 mg/mL外,由β-ETI估计的不确定度均低于由β-CTI估计的不确定度。由β-ETI和β-CTI确定的临界可接受不确定度质量浓度分别为4.867、5.686 mg/mL。结论 β-CTI可充分估计NIR定量不确定度,可用于NIR定量结果准确性和可靠性评价。

Objective To establish the uncertainty evaluation for the quantitative analysis on Xuesaitong Injection by near infrared (NIR) spectroscopy based on β-content tolerance interval (β-CTI). Methods The data of NIR calibration and validation sets were obtained by "I × J × K" full factorial experimental design. The NIR quantification model was established through the PLS algorithm. Based on the validation data, the β-CTI was introduced to estimate the quantitative analysys on uncertainty of NIR, which was compared with the regular β-expectation tolerance interval (β-ETI). Results Among the seven designed concentration levels, the uncertainty values estimated by the β-ETI are smaller than that estimated by the β-CTI, except for the concentration level of 1.485 mg/mL. Then, the relationship between the relative expanded uncertainty (U) and corresponding concentration (Y) was fitted by lnU= a + b lnY model. When the uncertainty acceptance limit (λ) was set to 20% and the decision profile of uncertainty was plotted, the critical concentration for β-ETI and β-CTI with the acceptable uncertainty could be read as 4.867 and 5.686 mg/mL, respectively. Conlusion This study has proved that β-CTI can be used to adequately estimate the uncertainty of quantitative analysis by NIR, which provides a new potential tool for the assessment of the accuracy and reliability of NIR spectroscopy.

国家自然科学基金资助项目(81403112);北京市重点实验室建设项目(中药基础与新药研究实验室,2014)