计算机与应用化学
計算機與應用化學
계산궤여응용화학
COMPUTERS AND APPLIED CHEMISTRY
2013年
7期
767-770
,共4页
Bootstrap方法%区间估计%指纹图谱%相似度阈值
Bootstrap方法%區間估計%指紋圖譜%相似度閾值
Bootstrap방법%구간고계%지문도보%상사도역치
Bootstrap method%interval estimation%fingerprints%similarity threshold
利用指纹图谱相似度评价中药质量时,色谱指纹图谱样本的数量有限。小样本条件下指纹图谱相似度值的分布难以满足正态分布或近似正态分布,不适合直接利用传统的区间估计方法来确定相似度的阈值。本文利用Bootstrap方法扩增样本,建立了小样本条件下的指纹图谱相似度阈值确定方法,并通过3个具体实例说明指纹图谱相似度阈值确定的具体过程。首先通过Bootstrap方法扩增样本,然后利用单侧区间估计方法确定相似度下限,得到相似度的阈值。以上3个实例说明了用该方法建立中药指纹图谱相似度阈值的合理性和可行性。此方法适用于由不同相似度计算方法所得相似度值。本文从理论、技术和应用的角度阐述了确定中药指纹图谱相似度阈值的方法,该方法将对指纹图谱技术的推广应用产生积极作用。
利用指紋圖譜相似度評價中藥質量時,色譜指紋圖譜樣本的數量有限。小樣本條件下指紋圖譜相似度值的分佈難以滿足正態分佈或近似正態分佈,不適閤直接利用傳統的區間估計方法來確定相似度的閾值。本文利用Bootstrap方法擴增樣本,建立瞭小樣本條件下的指紋圖譜相似度閾值確定方法,併通過3箇具體實例說明指紋圖譜相似度閾值確定的具體過程。首先通過Bootstrap方法擴增樣本,然後利用單側區間估計方法確定相似度下限,得到相似度的閾值。以上3箇實例說明瞭用該方法建立中藥指紋圖譜相似度閾值的閤理性和可行性。此方法適用于由不同相似度計算方法所得相似度值。本文從理論、技術和應用的角度闡述瞭確定中藥指紋圖譜相似度閾值的方法,該方法將對指紋圖譜技術的推廣應用產生積極作用。
이용지문도보상사도평개중약질량시,색보지문도보양본적수량유한。소양본조건하지문도보상사도치적분포난이만족정태분포혹근사정태분포,불괄합직접이용전통적구간고계방법래학정상사도적역치。본문이용Bootstrap방법확증양본,건립료소양본조건하적지문도보상사도역치학정방법,병통과3개구체실례설명지문도보상사도역치학정적구체과정。수선통과Bootstrap방법확증양본,연후이용단측구간고계방법학정상사도하한,득도상사도적역치。이상3개실례설명료용해방법건립중약지문도보상사도역치적합이성화가행성。차방법괄용우유불동상사도계산방법소득상사도치。본문종이론、기술화응용적각도천술료학정중약지문도보상사도역치적방법,해방법장대지문도보기술적추엄응용산생적겁작용。
The number of the samples is usually limited which chromatographic fingerprints similarity are used for the quality evaluation of traditional Chinese medicine(TCM). Fingerprints similarity distribution of the samples does not meet the normal distribution or approximate normal distribution because of the small sample size, and it is not suitable to determine the similarity threshold directly by interval estimation. A Method for determining the similarity threshold of fingerprints was proposed in this paper, and the determination process was illustrated by the chromatographic fingerprints similarity of 3 kinds of samples, which are Radix Salviae Miltiorrhizae, Sanhuang tablets and Smilax glabra Roxb. The number of samples was amplified by the Bootstrap method,and the lower limit of the chromatographic fingerprints similarity values was estimated by interval estimation. The effectiveness and feasibility of this method was confirmed by the above 3 examples,and the above method can be used to determine the threshold for the fingerprints similarity values calculated by different similarity measures. The method determining fingerprints similarity threshold based on the Bootstrap method and interval estimation is illustrated in this paper from the perspective of theory, technology and applications, which will have a positive impact on promoting the application of fingerprint technology in traditional Chinese medicine.