基于BP-ANN模型的香附效应成分筛选

目的: 分析香附不同炮制品的HPLC指纹图谱与药理效应数据,应用偏最小二乘法(PLS)和BP神经网络模型(BP-ANN)对香附不同炮制品中HPLC指纹图谱共有峰的峰面积与调经止痛药理效应进行关联,筛选香附主要效应成分。 方法: 采用PLS进行数据处理,通过MATLAB中的神经网络工具箱,建立BP-ANN模型,计算出各个因素的平均影响值(MIV),根据MIV大小列出各因素对应变量(药理效应)影响的相对重要性顺位,筛选香附治疗痛经的有效成分。 结果: 11组香附样品中存在16个共有峰,假设各峰的成分分别为X1,X2,X3,X4,X5,X6,X7,X8,X9,X10,X11,X12,X13,X14,X15,X16。香附对缩宫素所致小鼠痛经模型的扭体反应抑制率影响的主要效应成分排序为X13>X15>X7>X16;对缩宫素诱导小鼠离体子宫痉挛性收缩模型的肌张力抑制率影响的主要效应成分排序为X15>X13>X5>X16。 结论: 通过应用PLS和BP-ANN模型进行HPLC指纹图谱共有峰与药理效应关联性分析,可进行香附主要效应成分的筛选。

Screening of Effective Ingredients from Cyperi Rhizoma Based on BP-ANN Model

Objective: Through analysis HPLC fibgerprint and pharmacological effect data of different processed products of Cyperi Rhizoma,application PLS and BP-ANN model to associate total peak area of HPLC fingerprint and pharmacodynamics effect of anti-dysmenorrhea from different processed products of Cyperi Rhizoma,then to screen major effective ingredients of Cyperi Rhizoma. Method: Using PLS for data processing,BP-ANN model was established by the neural network toolbox in MATLAB,MIV of various factors were calculated,relative importance sequence of various factors for corresponding variable(pharmacological effect) was listed by according to MIV,then to screen major effective ingredients of anti-dysmenorrhea in Cyperi Rhizoma. Result: There were sixteen common peaks in eleven groups of Cyperi Rhizoma samples,compositions of each peak were assumed X1,X2,X3,X4,X5,X6,X7,X8,X9,X10,X11,X12,X13,X15,X16.Sequence of major effective ingredients for effect of Cyperi Rhizoma on writhing inhibition rate in mice dysmenorrhea model was X13,X15,X7,X16,sequence of major effective ingredients on muscle tension inhibition rate in mice isolated uterine contraction model was X15,X13,X5,X16,which were induced by oxytocin. Conclusion: Through relevance analysis of HPLC fingerprint total peak and pharmacodynamics effect by PLS and BP-ANN model,it was able to screen major effective ingredients in Cyperi Rhizoma.