矿物药中的多种金属和非金属元素及其含量决定了其功效。 目前亟待建立普遍适用的、 非化学分析的、 以元素种类及其含量信息为基础的多种矿物药快速鉴别及定量质量评价方法。 EDS元素指纹图谱可以快速测定各类矿物药的元素种类及含量。 双指标等级序列个性化模式识别法则通过比较计算任意两个样品的共有(量)指标率及变异(量)指标率, 建立以各个样品为参考的个性化双指标序列, 然后再计算每个序列中所有样品的平均共有指标率及其标准偏差S, 构造相似尺度函数P≥+xS。 确定适合于所有样品最优分类的x和每个样品的个性化的显著相似样品特征序列, 将其作为分类标准实现最佳分类。 该法不仅适用于仅有共有峰, 同时也适用于既存共有峰又有变异峰的指纹图谱的量化分析。 测定了七种不同公司生产的蒙脱石散的EDS指纹图谱, 计算了不同样品指纹图谱之间峰面积(或主要活性元素Fe, Al, Ca, Mg, Si的含量)的双指标率, 当相似尺度x=0.5, 可对七种蒙脱石散产品进行高分辨鉴别及准确质量评价。 该法为矿物药的快速高分辨鉴别及定量质量评价提供了一种新途径。

EDS(energy dispersive spectrometer)element fingerprint spectra is able to quickly measure the kinds and the contents of elements in any mineral drug. In dual index grade sequence individualized pattern recognition method, common (quantity) index and variation (quantity) index ratios of any two samples’ fingerprint spectra are calculated, and the individualized dual index sequence of each sample is constructed relying on its own EDS fingerprint spectra as the reference. Then the mean common (quantity) index ratio and the standard deviation S of all samples in each sample’s individualized dual index sequence are computed. On this basis, for each sample, its own similarity scale function P≥+xS is built up. By this function, the optimum x suitable for optimized classification/cluster of all samples is determined, and the individualized characteristics sequence of one sample, to which samples in the individualized characteristics sequence are significantly similar, is decided also. Finally, depending on these individualized characteristics sequences, the optimized classification/cluster of all samples can be carried out perfectly without any prior knowledge related to them. This method is not only suitable for the quantitative analysis on fingerprint spectra being of only common peaks, but also fits for that being of both common and variant peaks. In this study, the EDS element fingerprint spectra of seven mineral drug montmorillonite powder samples from different companies were detected. Then common (quantity) index and variant (quantity) index ratios of peak area (or contents of majorly active element Fe, Al, Ca, Mg, Si) among different EDS fingerprint spectra were obtained. In the similarity scale function P≥+xS, when x=0.5, these seven mineral drug montmorillonite powder samples could be quickly identified with high resolution, be classified into two groups, and their quality could be evaluated precisely. In general EDS element fingerprint spectra combined with dual index grade sequence individualized pattern recognition method offers a novel approach to quickly identifying mineral drugs with high resolution and for accurately quantitative quality evaluation of them, based on atomic content information provide by EDS fingerprint spectra.