Construction of universal calibration model for levofloxacin injections by fiber-optic transmittance-reflectance near-infrared spectroscopy

通过近红外漫反射光谱构建左氧氟沙星注射剂的通用型校正模型(英文)

The general strategy and method of constructing universal calibration model for levofioxacin injections by near-infrared spectroscopy have been investigated and discussed. Firstly, a constant-temperature homogeneous liquid calibration model for levofloxacin hydrochloride injections with the same composition but different active principal ingredient （API） content was established as the basic unit for universal model. Then, samples of levofloxacin hydrochloride injections containing propylene glycol or levofloxacin lactate injections were added to develop a primary constant-temperature liquid universal model. Temperature- amended final universal model was established to apply to samples under different temperatures. The final model was built from 61 calibration samples and 77 validation samples. The value of the root mean square error of cross validation （RMSECV） and coefficient of determination （r2） of leave-one-out cross-validation （LOOCV） were 0.792 and 0.9993, respectively, the root mean square error of prediction （RMSEP） of test set validation （TSV） was 0.87, and the average relative deviation was 1.44%. According to the ICH guidelines, the universal calibration model was evaluated. Based on the experimental statistical results, the recommended number of calibration samples for a constant-temperature homogeneous liquid quantitative model was no less than 15.

研究和讨论了近红外光谱法构建左氧氟沙星注射剂通用型模型的一般策略和方法。首先,对组分相同,但活性成分浓度不同的盐酸左氧氟沙星注射剂建立了恒温均质液体校正模型,并作为一个基本模型。第二步,将含有特殊组分的左氧氟沙星注射剂如丙二醇或者乳酸左氧氟沙星注射剂加入到模型中,形成初级恒温均质液体校正模型。第三步,针对不同温度下样本,建立了最终的变温通用型液体校正模型。所建立模型的最终参数是:建模样本数为61,验证集样本数为77,模型内部交叉验证的RMSECV和r^2分别为0.792和0.9993,外部验证的RMSEP为0.87,平均相对误差为1.44%。根据ICH指导原则对模型方法进行了评价并总结出了建立液体通用型模型的一般方法,同时根据实验结果,给出了建立组分一致浓度不同的均质定量模型的推荐样本量为不少于15个样品。