基于数据信息熵探讨糖皮质激素诱导的阳虚或阴虚证候动物模型的状态特征

Mathematical analysis of characteristics of glucocorticoid-induced yang deficiency or yin deficiency syndrome in animal models based on information entropy theory

目的:运用信息熵理论对糖皮质激素诱导的中医阳虚或阴虚证候动物模型的生化指标进行数据挖掘,分析模型状态特征,为界定和阐释模型状态奠定基础。方法:收集4批实验共24组不同造模条件(造模剂种类、造模剂量和造模时间)下模型的生化指标数据作为数据源,通过计算标准分、信息增益值并结合等级级数分布图对不同模型状态进行分析。比较同一个指标在不同研究对象(造模剂种类、造模剂量和造模时间)时信息增益值大小以得知影响指标变化的最主要因素,比较同一研究对象不同指标时的信息增益值以得知指标对研究对象不同类别的区分能力大小。结果:当研究对象为造模剂种类时,绝大部分指标得到最大的信息增益值,提示造模剂种类是模型状态最主要的干预因素,其中血清三酰甘油(triglyceride,TG)、总胆固醇(total cholesterol,TC)、总蛋白(total protein,TP)、白蛋白(albumin,ALB)、丙氨酸氨基转移酶(alanine aminotransferase,ALT)、天门冬氨酸氨基转移酶(aspartate aminotransferase,AST)和皮质类固醇(corticosteroid,CS)的信息增益值较高,血浆环腺苷酸(cyclic adenosine monophosphate,cAMP)、睾酮(testosterone,T)和乳酸脱氢酶(lactate dehydrogenase,LDH)的信息增益值较低,提示TG、TC、TP、ALB、ALT、AST和CS较cAMP、T和LDH在两种模型中的差别更为明显,即前一组指标对两种模型状态的区分能力更强。受造模时间影响较多的指标是AST;造模剂剂量对氢化可的松模型中的TC和ALB影响较大,而对地塞米松模型所有指标变化均影响很小。结论:本研究采用的基于信息熵理论的数据处理方法能够综合考虑多批生物学数据,探讨指标变化的趋向性和稳定性。作者在分析造模条件对模型指标变化的干预程度及方向的基础上,提出了适合两种模型比较研究的合理指标及模型状态阐释的若干建议。

Objective：Based on information entropy theory,this study analyzes the experimental indicators of glucocorticoid-induced traditional Chinese medicine yang deficiency or yin deficiency syndrome in animal models,thus laying the foundation of defining and interpreting the model state.Methods：Data of biochemical indicators from 24 groups of animal models with different modeling conditions （type of modeling agent,dosage of modeling agent and modeling time） were collected. Information gain values for three study objects （type of modeling agent,dosage of modeling agent and modeling time） were calculated respectively after standardization,and then characteristics of yang deficiency or yin deficiency syndrome models were interpreted with these values and ranking map.Results：Greatest information gain values of most indicators were got when the study object was the type of modeling agent,which is the most important factor in the differentiation of model state. With this study object,triglyceride （TG）,total cholesterol （TC）,total protein （TP）,albumin （ALB）,alanine aminotransferase （ALT）,aspartate aminotransferase （AST）,and corticosteroid （CS） got larger information gain values than testosterone （T）,lactic dehydrogenase （LDH）,and cyclic adenosine monophosphate （cAMP）. This indicated that the former seven indicators may be significantly different between the two animal models induced by hydrocortisone and dexamethasone,respectively. In the study of the modeling time,AST may be affected more than others. In the study of the dosage of modeling agent,TC and ALB may be affected more in the model of hydrocortisone and no indicators were significantly affected by the dosage of modeling agent in the dexamethasone model.Conclusion：Mathematical method based on information entropy theory allows researchers to analyze experimental data in several experiments at the same time and to interpret the trend and stability of indicators. Based on this study and by analyzing the impacts of different modeling conditions on the indicators,elucidations of the characteristics of glucocorticoid-induced yang deficiency or yin deficiency syndrome in animal models and some biochemical indicators for model comparison are in need