CN1876172A - BP neural net model method for preparing pharmaceutical composition - Google Patents
BP neural net model method for preparing pharmaceutical composition Download PDFInfo
- Publication number
- CN1876172A CN1876172A CN 200610083340 CN200610083340A CN1876172A CN 1876172 A CN1876172 A CN 1876172A CN 200610083340 CN200610083340 CN 200610083340 CN 200610083340 A CN200610083340 A CN 200610083340A CN 1876172 A CN1876172 A CN 1876172A
- Authority
- CN
- China
- Prior art keywords
- network
- flavour
- sample
- model
- pharmaceutical composition
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Medicines That Contain Protein Lipid Enzymes And Other Medicines (AREA)
Abstract
The invention discloses the drug compound with BP neural net model. The method comprises the following steps: building non-linear mapping model of dosage and drug-effect with BP neural net; defining BP neural net model parameters; building BP neural net model non-linear mapping model. The method completely and fully utilizes the laboratory data, and uses computer to replace a great amount of animal experiment works. According to the BP neural net model, the method forecasts matching relation.
Description
Technical field
The present invention relates to a kind of medicine series compositions, preparation method and its usage, particularly the pharmaceutical composition and the new pharmaceutical use of the preparation of BP neural network model.
Background technology
The normal Return Law of using is carried out modeling in the medical science, but needs data to satisfy certain precondition during Return Law modeling, must satisfy independence, linearity, normal state, the neat condition of variance in data as linear regression, just can use; Whether regression effect directly depends on model suitable, so after the modelling, need the reasonability of regression model basic assumption is checked.When using nonlinear regression, Model Selection, preparation are analyzed, initial value is determined, conversion, the parametric solution of parameter and variable, also has the assessment of model correction, comparison and check, convergence and match, all there are multiple difficult point and problem, these all can have influence on the effect to the nonlinear data match, even can have influence on the foundation of model.A large amount of studies show that, between the flavour of a drug dose of compound recipe and pharmacodynamics, there is complicated non-linear relation, still not having proper model at present for this non-linear relation can use for reference, when this just makes method modelings such as using traditional nonlinear regression, determine in Model Selection, initial value, aspects such as parameter transformation is found the solution, model testing exist and determine difficulty, thereby influence the effect of Nonlinear Modeling.Utilize this method can filter out the pharmaceutical composition of optimization.
Summary of the invention
The objective of the invention is to disclose a kind of new pharmaceutical composition and new medicine use of former crude drug from banxia xiexin decoction and class side thereof.The object of the invention also is to provide a kind of new method with the screening of BP cerebellar model arithmetic computer, preparation pharmaceutical composition of the present invention.The present invention seeks to be achieved through the following technical solutions:
One, the new medicine use of single medicinal material
(1) Rhizoma Pinelliae has application in the medicine that suppresses pepsin activity effect, gastric acid inhibitory secretory action, antagonism bowel movement effect or collaborative gastric emptying in preparation.
Wherein said gastric acid inhibitory secretory action is meant and contains suitable Rhizoma Pinelliae crude drug amount 5-25g in the pharmaceutical units preparation, contains suitable Rhizoma Pinelliae crude drug amount optimal dose 21g in the pharmaceutical units preparation.
(2) Radix Scutellariae has the application in the medicine that promotes pepsin effect, gastric acid inhibitory secretory action, gastric acid secretion effect, the effect of inhibition gastric mucus secretion, suppresses the gastrointestinal motility effect in preparation.
The effect of wherein said promotion pepsin is meant and contains suitable Radix Scutellariae crude drug amount 5-25g in the pharmaceutical units preparation, contains suitable Radix Scutellariae crude drug amount optimal dose 12g in the pharmaceutical units preparation.
Wherein said gastric acid inhibitory secretory action is meant and contains suitable Radix Scutellariae crude drug amount 2-6g in the pharmaceutical units preparation, contains suitable Radix Scutellariae crude drug amount optimal dose 6g in the pharmaceutical units preparation.
Wherein said gastric acid secretion effect is meant and contains suitable Radix Scutellariae crude drug amount in the pharmaceutical units preparation greater than 6g.
The effect of wherein said inhibition gastric mucus secretion is meant and contains suitable Radix Scutellariae crude drug amount 9-25g in the pharmaceutical units preparation.
The effect of wherein said inhibition gastrointestinal motility is meant and contains suitable Radix Scutellariae crude drug amount 10-24g in the pharmaceutical units preparation.
(3) Rhizoma Coptidis have in preparation promote the pepsin effect, suppress the gastric mucus secretion effect, antagonism gastric emptying or suppress application in the medicine of gastrointestinal motility effect.
The effect of wherein said promotion pepsin is meant and contains suitable Rhizoma Coptidis crude drug amount 2-15g in the pharmaceutical units preparation, contains suitable Rhizoma Coptidis crude drug amount optimal dose 6g in the pharmaceutical units preparation.
The effect of wherein said inhibition gastric mucus secretion is meant and contains suitable Rhizoma Coptidis crude drug amount 9-15g in the pharmaceutical units preparation.
(4) Radix Codonopsis has the pepsin effect that suppresses in preparation, suppresses gastric emptying, promotes gastric emptying, promotes the application in the medicine of gastrointestinal motility effect.
(5) Rhizoma Zingiberis has the application in the medicine that suppresses the pepsin effect, promotes gastric mucus secretion effect, inhibition gastric emptying or the effect of antagonism gastrointestinal motility in preparation.
The effect of wherein said inhibition pepsin is meant and contains suitable Rhizoma Zingiberis crude drug amount 10-25g in the pharmaceutical units preparation, contains suitable Rhizoma Zingiberis crude drug amount optimal dose 18g in the pharmaceutical units preparation.
(6) Fructus Jujubae has application in the medicine that strengthens pepsin effect, gastric acid secretion effect, antagonism gastric emptying or the effect of antagonism gastrointestinal motility in preparation.
The effect of wherein said enhancing pepsin is meant and contains quite dried Fructus Jujubae crude drug amount 9-25g in the pharmaceutical units preparation
Wherein said antagonism gastric emptying is meant the application in the inhibition of preparation treatment gastrointestinal motility, hyperfunction type gastropathy.
The effect of wherein said antagonism gastrointestinal motility is meant and contains quite dried Fructus Jujubae crude drug amount 9-15g in the pharmaceutical units preparation.
(7) Radix Glycyrrhizae has the application in the medicine that promotes the pepsin effect, promotes bowel movement effect, promotion gastric emptying or antagonism gastric emptying in preparation.
(8) Rhizoma Zingiberis Recens has the application in the medicine that increases the pepsin activity effect, promotes the gastrointestinal motility effect in preparation.
The effect of wherein said increase pepsin activity is meant and contains quite dried Rhizoma Zingiberis Recens crude drug amount 5-20g in the pharmaceutical units preparation, contains suitable Rhizoma Zingiberis Recens crude drug amount optimal dose 10g in the pharmaceutical units preparation.
The effect of wherein said promotion gastrointestinal motility is meant and contains quite dried Rhizoma Zingiberis Recens crude drug amount 5-20g in the pharmaceutical units preparation, contains suitable Rhizoma Zingiberis Recens crude drug amount optimal dose 15g in the pharmaceutical units preparation.
Two, new pharmaceutical composition
(1) suffering is opened the group pharmaceutical composition
A kind of pharmaceutical composition is characterized in that containing suitable Rhizoma Pinelliae crude drug amount 10-21g (less than 21g) in the unit formulation of this pharmaceutical composition, contains suitable Rhizoma Zingiberis crude drug amount 10-18g (less than 18g).
The pharmaceutical composition that contains above-mentioned crude drug amount in the unit formulation has gastric acid inhibitory effect, suppresses the effect of pepsin effect or promotion gastric mucus.
A kind of pharmaceutical composition is characterized in that containing suitable Rhizoma Pinelliae crude drug amount 22-30g (greater than 21g) in the unit formulation of this pharmaceutical composition, contains suitable Rhizoma Zingiberis crude drug amount 19-30g (greater than 18g).
The pharmaceutical composition that contains above-mentioned crude drug amount in the unit formulation has the effect of the gastric acid of promotion, promotes the pepsin effect or suppresses the gastric mucus effect.
(2) the hardship pharmaceutical composition of coming down to a lower group
A kind of pharmaceutical composition is characterized in that containing suitable Radix Scutellariae crude drug amount 2-20g in the unit formulation of this pharmaceutical composition, contains suitable Rhizoma Coptidis crude drug amount 2-15g.
Contain suitable Radix Scutellariae crude drug amount 6g, 9g or 12g in the unit formulation of preferred this pharmaceutical composition, contain suitable Rhizoma Coptidis crude drug amount 6g, 9g.
The pharmaceutical composition that contains above-mentioned Radix Scutellariae and Rhizoma Coptidis crude drug amount in the unit formulation has gastric acid secretion, suppresses the gastric mucus effect.
(3) sweet benefit group pharmaceutical composition
A kind of pharmaceutical composition is characterized in that containing suitable Fructus Jujubae crude drug amount 9-30g in the unit formulation of this pharmaceutical composition, contains suitable Radix Glycyrrhizae crude drug amount 20-27g.
The pharmaceutical composition that contains above-mentioned Fructus Jujubae crude drug amount and Radix Glycyrrhizae crude drug amount in the unit formulation has the gastric acid secretion effect, promotes the pepsin effect.
A kind of pharmaceutical composition is characterized in that containing suitable Fructus Jujubae crude drug amount 9-30g in the unit formulation of this pharmaceutical composition, contains suitable Radix Glycyrrhizae crude drug amount 28-40g.
The pharmaceutical composition that contains above-mentioned Fructus Jujubae crude drug amount and Radix Glycyrrhizae crude drug amount in the unit formulation has the gastric acid secretion effect, suppresses to advance the pepsin effect.
Aforementioned pharmaceutical compositions or purposes can but be not limited to determine flavour of a drug, dose or drug effect that this method comprises the steps: by following BP cerebellar model arithmetic computer
1.BP neural network model parameter setting
At the research of banxia xiexin decoction and class side's compatibility rule thereof, for setting up flavour of a drug dose non-linear the hint obliquely at model relevant, be provided with as follows to the parameter of network with pharmacodynamics:
(1) choosing and pretreatment of sample:
Choosing of sample: from flavour of a drug dose and medicine efficacy relation research banxia xiexin decoction and class side's compatibility rule thereof, so choose wherein 8 herbal medicines (Rhizoma Pinelliae, Rhizoma Zingiberis Recens, Radix Scutellariae, Rhizoma Coptidis, Radix Codonopsis, Radix Glycyrrhizae, Fructus Jujubae, Rhizoma Zingiberis) as input layer.These variablees can be represented the compatibility key element of three compound recipes.Choose gastric emptying, bowel movement, stomachial secretion function pharmacodynamic index as output neuron, set up the output layer neuron number and be 3 neural network model.
The standardization processing of sample: because the index of weighing has nothing in common with each other, each number of components magnitude of original sample has very big difference, and this just need carry out standardization processing to sample.When adopting the neutral net regression forecasting, possible certain node input value excessive then its more new increment is excessive accordingly, thereby make weights excessive, cause other components almost to lose regulating and controlling effect, so need carry out appropriate standardization processing to original sample.The standardization processing of sample should design multiple scheme, avoids under the prerequisite of " saturated phenomenon " as far as possible, makes sample uniform distribution in bigger amplitude range.For the data handling utility normalized function of this project problem, make all inputs, output sample standard in the scope of [1,1].
(2) power and the initial codomain of threshold value determines
As a whole, power (comprising the threshold value item) is along with the carrying out of iteration upgrades, and generally is astringent, but the initial value of power is too big, may cause network to reach capacity soon.For avoiding saturated as far as possible, and it is initial at random equably to consider that the convergence rate of network and the complexity of sample data, less initial codomain are more conducive to weights, and it is better that the initial codomain of power is got (0.05,0.05).In this project initial weight is adopted the method for repeated trials in above-mentioned scope,, determine best initial weight by comparing fitting effect.
(3) network topology structure determines
Network topology structure all plays pivotal role for popularization ability, the computational efficiency of whole network.Owing to sample data is predicted output layer is got a node.The key of determining topology of networks is to determine suitable hidden layer number and hidden layer number of unit.
Determining of hidden layer number: the network of determining to use 1 hidden layer.
Determining of hidden layer number of unit: adopted " trial-and-error method " to determine
(4) iteration stopping standard
During modelling, it is 5000 that maximum loop iteration is set; Cycle-index surpasses 5000 times and does not reach the iteration stopping standard yet, then shows the failure of an experiment.The convergence target allows there is error, less than given range of error, then stops the correction to network weight in the output of network and the difference between the sample.Employing can be accelerated the pace of learning of network to the way of e-learning tolerance.Adopt adaptive method, promptly allowable error is got a little louder when the training beginning, then along with training reduces gradually.Setting convergence error at this problem is 0.001.
(5) BP algorithm: use adding moment of inertia algorithm and adaptive step algorithm to improve the pace of learning of standard BP network
2. the foundation of the nonlinear mapping model of flavor dose and drug effect BP neutral net
With the experimental data is input, use the BP neutral net, set up the nonlinear mapping model of nonlinear mapping model, banxia xiexin decoction and class side's flavour of a drug dose and bowel movement of nonlinear mapping model, banxia xiexin decoction and the class side's flavour of a drug dose and the gastric emptying of banxia xiexin decoction and class side's flavour of a drug dose and stomachial secretion (gastric acid, pepsin and gastric mucus secretion) respectively.Being set at of each parameter in the network: input layer is 8, and output layer is 3, and hidden layer is 10, and topological structure is 8-10-3, and learning algorithm is selected " the BP algorithms that activity level method and learning rate self adaptation are adjusted two kinds of stragetic innovations ", and the network convergence goal setting is 1E-3.
3. model evaluation
By using the difference of mean square error function comparison object value and predictive value, calculate the error between desired value and predictive value, observe network model to training, the network fitted figure is estimated in network performance.
(1) flavour of a drug dose and bowel movement relational model are estimated
The mean square error of measured value and match value is that flavour of a drug dose and bowel movement mapping model MSE are 9.9978e-004.
Desired value and predictive value comparison flavour of a drug dose and bowel movement relational model are to the match situation of the data learnt and the data prediction ability of not learning is relatively shown, less than 10%, reflect that network match performance is outstanding for the sample relative error of learn; , good for the sample relative error of not learning from other one side reflection flavour of a drug dose and bowel movement relational model network estimated performance less than 20%.
(2) flavour of a drug dose and gastric emptying relational model are estimated
The mean square error flavour of a drug dose of measured value and match value and gastric emptying mapping model MSE are 0.0015.
Desired value and predictive value comparison flavour of a drug dose and gastric emptying relational model are to the match situation of the data learnt and the data prediction ability of not learning is relatively shown, less than 10%, reflect that network match performance is outstanding for the sample relative error of learn; , good for the sample relative error of not learning from other one side reflection network estimated performance less than 20%.
(3) flavour of a drug dose and stomachial secretion relational model are estimated
The mean square error of measured value and match value is that flavour of a drug dose and stomachial secretion mapping model model M SE are 14.6922.
Desired value and predictive value comparison flavour of a drug dose and stomachial secretion relational model are to the match situation of the data learnt and the data prediction ability of not learning is relatively shown, less than 10%, reflect that network match performance is outstanding for the sample error of learn; , good for the sample error of not learning from other one side reflection flavour of a drug dose and stomachial secretion relational model network estimated performance less than 20%.
Three, the new pharmaceutical use of XIEXIN TANG
Banxia xiexin decoction has the application in the middle of the medicine of following effect in preparation: the delayed gastric emptying that the antagonism atropine causes, the intestinal propulsion function that the antagonism atropine causes suppresses, makes VIP in the blood, the decline of SP level, and the SS level raises, reduces gastric antrum smooth muscle cell CAP and CAD expression or directly change CAP mRNA relevant with the expression of CAD mRNA
Glycyrrhizae Decoction for Purging Stomach-Fire has the application in the middle of the medicine of following effect in preparation: the intestinal propulsion function that the antagonism atropine causes suppresses, makes that MTL level, GAS level raise, downward modulation gastric antrum smooth muscle cell CAP and CAD expression or directly to change CAP mRNA relevant with the expression of CAD mRNA.
Shengjian Xiexin Tang has the application in the middle of the medicine of following effect in preparation: the intestinal propulsion function that the antagonism atropine causes suppresses, makes that motilin level, gastrin level, somatostatin level raise, downward modulation gastric antrum smooth muscle cell CAP and CAD expression or directly to change CAP mRNA relevant with the expression of CAD mRNA.
Experimental example 1 stomachial secretion drug action research
Experimental technique
(1) to choose body weight be 200～250gSD rat for animal grouping, is divided into 24 groups at random, and 15 every group, wherein experimental group is 20 groups, 1 group of normal control, and other establishes three classical prescription groups, and wherein the banxia xiexin decoction group participates in modeling, and Radix Glycyrrhizae, Shengjian Xiexin Tang group give over to reserves check.Each experimental group as investigating the factor, as the investigation level, is got 20 levels with the consumption of each flavour of a drug with each flavour of a drug, and by the experimental program that uniform designs table obtains, totally 20 proportion relations see Table 1-1.
Table 1-1 respectively organizes the dosage level (cutting table) of flavour of a drug
| Group | Rhizoma Zingiberis Recens (g) | The Rhizoma Pinelliae (g) | Radix Scutellariae (g) | Rhizoma Coptidis (g) | Radix Codonopsis (g) | Rhizoma Zingiberis (g) | Fructus Jujubae (piece) | Radix Glycyrrhizae (g) |
| 1 2 3 · · · 22 23 24 | 21 27 6 · · · 0 0 12.6 | 10 6 16 · · · 9 8.1 10.7 | 21 18 12 · · · 7.6 6.8 9.0 | 8 8 2 · · · 2.5 2.3 3.0 | 12 30 36 · · · 7.6 6.8 9.0 | 0 24 3 · · · 7.6 6.8 3.0 | 9 21 3 · · · 3.4 3.0 4.0 | 10 2 0 · · · 7.6 9.0 9.0 |
(2) the 1. collection of gastric juice of experiment content and technology: the experimental group rat is by 7 times of administrations of people's consumption, and every day twice, continuous 5 days, matched group distilled water every day was irritated stomach.It is stand-by to collect gastric juice after 5 days.2. the mensuration of gastric acid: adopt sodium hydroxide titre computing method.3. the mensuration of pepsin activity: adopt improvement peace Song Shi method.4. the mensuration of gastric mucus binding capacity: adopt colorimetry (wavelength 615nm) to measure.
Date processing
Each organizes the method that relatively adopts variance analysis between the group of experimental data, the pharmacological experimental data of gained and tripartite different proportionings input uniform Design software, by stepwise regression analysis, reject inapparent variable, introducing is contributed significant variable to pharmacodynamics index, draw the regression equation of gastric acid, pepsin activity, gastric juice mucus binding capacity respectively, and make isogram.
Experimental result
The different proportionings of medicine see Table 1-2 to normal rat stomach secretion influence, and gastric acid, pepsin activity and gastric mucus binding capacity are set up mathematical model and analyzed, and the results are shown in Table 1-3,1-4,1-5.
The different proportionings of table 1-2 medicine are to normal rat stomach secretion influence (x ± s) (cut table)
| Group | Quantity | Gastric acid (Y1) sodium hydroxide neutralization (mmol/l) | Pepsin activity (Y2) μ g tyrosine/ml gastric juice/min | Gastric juice mucus binding capacity (Y3) (mg ALician) |
| 1 2 3 · · · 22 23 24 | 13 13 14 · · · 15 15 15 | 99.2308±25.09 104.7692±21.54 94.7857±21.20 · · · 101.6429±28.24 100.5693±21.13 103.3333±14.90 | 280.6653±116.74 241.9013±50.48 280.1288±83.93 · · · 314.2951±79.9 278.7935±52.33 289.5517±60.97 | 0.6380±0.13 0.5637±0.16 0.5940±0.19 · · · 0.6027±0.14 0.6228±0.11 0.6020±0.15 |
Table 1-3 analysis of variance table
| Soruces of variation | Quadratic sum | Degree of freedom | Mean square and | Significance |
| Regression residual amounts to | 7.9128E+02 4.7406E+01 8.3869E+02 | 10 11 21 | 7.9128E+01 4.3096E+00 | When confidence limit α=0.01, F adds up value=1.8361E+01 F (10,11)=4.5393E+00 residual standard deviation S=7.1002E-02 |
| The coefficient of determination Ra=+0.90108268 that coefficient of multiple correlation R=0.971326833 is adjusted | ||||
Check done to reserve in last 2 records:
| Sequence number | X1 | X2 | X3 | X4 | X5 | X6 | X7 | X8 | Y1 | Predicted value | Error |
| 1 21 | 0.0 2.6 | 8.1 10.7 | 6.8 9.0 | 2.3 3.0 | 6.8 9.0 | 6.8 3.0 | 3.0 4.0 | 9.0 9.0 | 100.5693 103.3333 | 98.7230 95.5568 | -1.8463 -7.7765 |
Table 1-4 analysis of variance table
| Soruces of variation | Quadratic sum | Degree of freedom | Mean square and | Significance |
| Regression residual amounts to | 2.3791E+04 3.0070E+02 2.4091E+04 | 12 9 21 | 1.9826E+03 3.3411E+01 | When confidence limit α=0.01, F adds up value=5.9338E+01 F (12,9)=5.1114E+00 residual standard deviation S=4.6786E-02 |
| The coefficient of determination Ra=+0.97378839 that coefficient of multiple correlation R=0.993739544 is adjusted | ||||
Check done to reserve in last 2 records:
| Sequence number | X1 | X2 | X3 | X4 | X5 | X6 | X7 | X8 | Y2 | Predicted value | Error |
| 1 21 | 0.0 2.6 | 8.1 10.7 | 6.8 9.0 | 2.3 3.0 | 6.8 9.0 | 6.8 3.0 | 3.0 4.0 | 9.0 9.0 | 278.7935 289.5517 | 300.4332 288.0872 | 21.6397 -1.4645 |
Regression equation Y1
| Y＝+6.80274E-01 | F |
| -7.45130E-01*X2 -7.29765E-01*X3 +4.64190E-01*X8 +8.28553E-01*X2*X5 +4.19682E-01*X3*X3 +1.93167E-01*X4*X7 -1.45498E+00*X5*X7 +6.54311E-01*X7*X7 -5.83424E-01*X8*X8 +8.90001E-01*X3*X3*X7 | 6.8592E+01 1.6778E+01 6.7698E+00 2.9349E+01 4.5257E+00 4.2154E+00 6.2588E+01 7.2957E+01 1.0976E+01 2.5647E+01 |
Table 1-5 analysis of variance table
| Soruces of variation | Quadratic sum | Degree of freedom | Mean square and | Significance |
| Regression residual amounts to | 6.3648E-02 7.2521E-03 7.0900E-02 | 10 11 21 | 6.3648E-03 6.5929E-04 | When confidence limit α=0.01, F adds up value=9.6541E+00 F (10,11)=4.5393E+00 residual standard deviation S=1.3782E-01 |
| The coefficient of determination Ra=+0.82099850 that coefficient of multiple correlation R=0.947477404 is adjusted | ||||
Check done to reserve in last 2 records:
| Sequence number | X1 | X2 | X3 | X4 | X5 | X6 | X7 | X8 | Y3 | Predicted value | Error |
| 1 2 | 0.0 12.6 | 8.1 10.7 | 6.8 9.0 | 2.3 3.0 | 6.8 9.0 | 6.8 3.0 | 3.0 4.0 | 9.0 9.0 | 0.6228 0.6020 | 0.5765 0.5676 | -0.0463 -0.0344 |
Conclusion
This experiment is divided into groups with uniform design, be equivalent to from comprehensive test, pick out representative point and implement experiment, the gained data fit by computer, gastric acid, pepsin stomach function regulating mucus amount regression equation have been provided respectively, analysis of variance table shows that equation fits well, also further verified the reliability of regression equation through reserving check, had complicated interaction as can be seen between the flavour of a drug from regression equation.Main Conclusions is as follows: 1. Fructus Jujubae can significantly promote the normal rat gastric acid secretion, the Rhizoma Pinelliae then can significantly suppress its gastric acid secretion, obviously gastric acid inhibitory secretion behind Radix Codonopsis and the Fructus Jujubae compatibility, the Rhizoma Zingiberis Recens Rhizoma Zingiberis does not have influence to gastric acid secretion, and Rhizoma Coptidis is not outstanding to the gastric acid secretion influence.2. Fructus Jujubae significantly increases the pepsin activity of normal rat, Rhizoma Zingiberis Recens and Radix Glycyrrhizae compatibility also can significantly promote pepsin activity later on, Fructus Jujubae and Radix Glycyrrhizae compatibility can significantly reduce pepsin activity later, Rhizoma Zingiberis does not have influence to normal rat stomach proteinase activity, and Rhizoma Coptidis, Radix Codonopsis influence outstanding to pepsin activity.3. the Rhizoma Pinelliae, Rhizoma Coptidis single medicinal material show as the gastric mucus binding capacity that can significantly reduce normal rat in the side, and Rhizoma Zingiberis, Fructus Jujubae and Rhizoma Pinelliae compatibility then can significantly increase the gastric mucus binding capacity later on.4. three XIEXIN TANG all do not have obvious influence to gastric acid secretion, pepsin activity, the gastric juice mucus binding capacity of normal rat.
The drug action research of experimental example 2 gastrointestinal motilities
Experiment content has three parts: 1. different flavour of a drug and dosage align normal mice gastrointestinal motility influence.2. the influence that the mice gastrointestinal motility that different flavour of a drug and dosage cause atropine suppresses.3. the hyperfunction influence of mice gastrointestinal motility that different flavour of a drug and dosage cause neostigmine.
Experiment is a leading indicator with gastric emptying rate, small intestinal charcoal end propelling rate.
Experimental technique
(1) animal grouping
1. different flavour of a drug and dosage align the influence research of normal mice gastrointestinal motility and choose the ICR mice that body weight is 25～30g, are divided into 25 groups at random, and 15 every group, wherein experimental group is 20 groups, and the prescription proportioning is with the stomachial secretion active studies, and 21 groups is normal control.22～24 groups is three classical prescription groups, and wherein the banxia xiexin decoction group participates in modeling, and Radix Glycyrrhizae, Shengjian Xiexin Tang group give over to reserves check.25 groups is the metoclopramide group.
2. the ICR mice that body weight is 25～30g is chosen in the influence research that the mice gastrointestinal motility that different flavour of a drug and dosage cause atropine suppresses, and is divided into 26 groups at random, 15 every group.1～20 group is experimental group, and 21 groups is model group, and 22～24 groups is three classical prescription groups, and wherein the Shengjian Xiexin Tang group participates in modeling, and Radix Glycyrrhizae, banxia xiexin decoction group give over to reserves check, and 25 groups is normal control, 26 groups of positive matched groups (metoclopramide and atropine).
3. the ICR mice that body weight is 25～30g is chosen in the hyperfunction influence research of the mice gastrointestinal motility that different flavour of a drug and dosage cause neostigmine, is divided into 26 groups at random, 15 every group.1～20 group is experimental group, and 21 groups is model group, and 22～24 groups is three classical prescription groups, and wherein the banxia xiexin decoction group participates in modeling, and Radix Glycyrrhizae, Shengjian Xiexin Tang group give over to reserves check, and 25 groups is normal control, 26 groups of positive matched groups (diphenoxylate).
(2) experiment content and technology
The experimental group animal is pressed 9 times of administrations of people's consumption, every day twice, and successive administration 3 days, positive controls is given metoclopramide 10mg/kg, gastric infusion, matched group is given isopyknic distilled water.Normal mouse administration in the 3rd day only gives nutrition by 0.6ml/ after 1 hour and sticks with paste; The atropine group gives atropine 2mg/kg subcutaneous injection behind administration 30min, cause the gastrointestinal motility function to suppress model, irritates nutrition behind the modeling 30min and sticks with paste.Give neostigmine 0.12mg/kg subcutaneous injection behind the neostigmine group administration 30min, cause gastrointestinal motility hyperfunctioning model, irritate nutrition behind the 20min and stick with paste.Behind the 20min mice is taken off cervical vertebra and put to death, measure the gastrointestinal motility function.
1. gastric emptying rate is measured: gastric emptying rate %=1-(full stomach quality-empty stomach quality foodstuff quality) * 100%
2. the intestinal propulsion rate is measured: charcoal end propelling rate %=charcoal end head end is to pylorus distance/small intestinal total length * 100% date processing
Each organizes the method that relatively adopts variance analysis between the group of experimental data.Use uniform Design software and set up regression equation, and make isogram.
Experimental result
The different proportionings of medicine see Table 1-6 to the influence of normal mouse, gastrointestinal motility inhibition mice and hyperfunction mice gastric emptying rate of gastrointestinal motility and intestinal propulsion rate, table 1-7, table 1-8.
The different proportionings of table 1-6 medicine are to the influence of normal mouse gastrointestinal motility (x ± s)
| Group | Quantity | Gastric emptying rate (Y1) | Intestinal propulsion rate (Y2) |
| 1 2 3 · · · 23 24 25 | 14 14 14 · · · 14 13 14 | 0.1014±0.0233 0.0705±0.0267 0.1137±0.0452 · · · 0.1543±0.0463 0.1843±0.0589 0.4506±0.1012 | 0.6648±0.1159 0.6396±0.1152 0.6625±0.1103 · · · 0.6253±0.0788 0.6419±0.0784 0.7524±0.0693 |
The influence that the mice gastrointestinal motility that the different proportionings of table 1-7 medicine cause atropine suppresses (x ± s)
| Group | Quantity | Gastric emptying rate (Y1) | Intestinal propulsion rate (Y2) |
| 1 2 3 · · · 24 25 26 | 14 13 13 · · · 13 15 12 | 0.2398±0.0538 0.1790±0.0572 0.2646±0.0623 · · · 0.3100±0.0917 0.3023±0.0532 0.3678±0.0925 | 0.4959±0.1020 0.4655±0.0891 0.5276±0.0655 · · · 0.5168±0.0820 0.6474±0.0670 0.4735±0.0361 |
The hyperfunction influence of the gastrointestinal motility that the different proportionings of table 1-8 medicine cause neostigmine (x ± s)
| Group | Quantity | Gastric emptying rate (Y1) | Intestinal propulsion rate (Y2) |
| 1 2 3 · · · 24 25 26 | 13 13 13 · · · 16 15 13 | 0.4318±0.2217 0.3647±0.1768 0.3920±0.2046 · · · 0.4566±0.1326 0.3023±0.0532 0.3418±0.0964 | 0.7302±0.2076 0.6920±0.1440 0.7439±0.0844 · · · 0.7120±0.0687 0.6474±0.0670 0.5569±0.1024 |
Conclusion
(1) Radix Scutellariae significantly suppresses the gastric emptying of normal mouse, and Radix Glycyrrhizae can significantly promote the intestinal propulsion function of normal mouse.
(2) then can obviously strengthen gastric emptying and intestinal propulsion function behind the Radix Scutellariae Radix Codonopsis compatibility.Behind the Radix Scutellariae Rhizoma Coptidis compatibility, can promote the gastrointestinal motility of normal mouse,, then obviously suppress gastric emptying again with after the Fructus Jujubae.Behind the Radix Scutellariae Radix Glycyrrhizae Radix Codonopsis compatibility, then obviously suppress the intestinal propulsion function.
(3) Rhizoma Zingiberis does not have obvious influence to the gastrointestinal motility function of normal mouse, and the Rhizoma Zingiberis Recens and the Rhizoma Pinelliae are not outstanding to the gastric emptying influence of normal mouse.
(4) tripartite all do not have obvious influence to the small intestine movement of mice ahead running, and difference is then appearred in the influence of gastric emptying, and promptly Radix Glycyrrhizae and Shengjian Xiexin Tang can suppress the gastric emptying of normal mouse, and banxia xiexin decoction then cannot not influence outstandingly this process.
(5) Fructus Jujubae, Radix Glycyrrhizae, Rhizoma Coptidis have suppressed antagonism to the mice gastric emptying that atropine causes, Fructus Jujubae can also significantly promote the intestinal propulsion function, and the Rhizoma Pinelliae has suppressed antagonism to the intestinal motility function that atropine causes.
(6) behind Radix Glycyrrhizae and the Rhizoma Zingiberis compatibility, the intestinal propulsion that remarkable antagonism atropine causes suppresses, and Rhizoma Zingiberis Recens Rhizoma Pinelliae compatibility then increases the weight of the delayed gastric emptying that atropine causes.
(7) aspect the mice gastric emptying inhibition that causes at atropine, the three parts there are differences, i.e. the delayed gastric emptying that can the antagonism atropine causes of banxia xiexin decoction, and Glycyrrhizae Decoction for Purging Stomach-Fire, Shengjian Xiexin Tang can not this effects of antagonism.The intestinal propulsion function that tripartite all energy antagonism atropine causes suppresses.
(8) the hyperfunction effect of mice gastric emptying that neostigmine is caused is Radix Scutellariae and Radix Codonopsis significantly, and both all obviously reduce gastric emptying rate; Rhizoma Zingiberis Recens has obvious antagonism to the intestinal motility hyperfunctioning that neostigmine causes.
(9) it is hyperfunction that Radix Codonopsis, Rhizoma Pinelliae compatibility further increase the weight of the gastric emptying that neostigmine causes, behind Radix Scutellariae, Rhizoma Zingiberis Recens compatibility or Radix Glycyrrhizae, the Rhizoma Coptidis compatibility, further promotes the propulsion functions of small intestinal.
(10) in the Rhizoma Pinelliae, Radix Glycyrrhizae two XIEXIN TANG, Fructus Jujubae does not have obvious influence to the gastrointestinal motility hyperfunctioning that neostigmine causes; In the Shengjian Xiexin Tang, behind Fructus Jujubae and the Rhizoma Zingiberis Recens compatibility, the intestinal propulsion function there is slight facilitation.
The mice gastrointestinal motility that (11) three XIEXIN TANG all can not the antagonism neostigmine cause is hyperfunction.
Experimental example 3 is regulated the basic research of stomachial secretion material
This research has mainly been observed pastille serum to the stripped parietal cell H of rabbit
+-K
+The influence of ATPase gene expression dose.
Experimental technique
(1) divide into groups 20 of male new zealand rabbits of animal, body weight 2 ± 0.1kg is divided into four groups at random, banxia xiexin decoction group (B), B1, B2 and intact animal's group (A).Each side's proportioning is as table 1-11.
Table 1-11
| Group | The Rhizoma Pinelliae | Radix Scutellariae | Rhizoma Coptidis | Radix Codonopsis | Rhizoma Zingiberis | Fructus Jujubae | Radix Glycyrrhizae |
| A B B1 B2 | (g) 0 9.0 9.0 9.0 | (g) 0 7.6 7.6 7.6 | (g) 0 2.5 2.5 2.5 | (g) 0 7.6 7.6 7.6 | (g) 0 7.6 7.6 7.6 | (piece) 0 3.4 16.2 27.0 | (g) 0 7.6 7.6 7.6 |
(2) 1. parietal cell separation and purification of experiment content and major technique: get the rabbit gastric mucosa, obtain parietal cell with the digestion of BMEA Digestive system.2. the isolated rabbit parietal cell is identified: expect blue rejection identification of cell survival rate with platform; Calculate parietal cell purity with HE dyeing; With acridine orange picked-up measuring parietal cell function.3. parietal cell former be commissioned to train foster: the stripped parietal cell of purification is inoculated in pretreated culture bottle of poly-D-lysine or 6 orifice plates, adjusts cell density greatly about 1 * 10
6/ ml-9 * 10
6About/ml.37 ℃, 5%CO
2Incubation.4. the collection of pastille serum: each medication group is pressed 3.5 times of administrations of people's consumption, once a day, and continuous 7 days.Rise the 8th day morning administration after 2 hours femoral artery get blood, after solidifying, 4 ℃, centrifugal 10 minutes of 4000rpm.After gained serum mixed by group, 0.22 μ m filtered, and-20 ℃ of preservations are standby.5. pastille serum is to H
+-K
+The influence of the gene expression of ATPase α subunit: the parietal cell of separation and Culture through reverse transcription, is used H after extracting cell total rna
+-K
+ATPase α subunit primer can amplify the band of 470bp, shows that the parietal cell of separation and Culture can be expressed H
+-K
+The gene of ATPase α subunit.Forward primer 5 '＞ACTCTGGGCTCCACGTCG＜3 ', downstream primer 3 '＞GCGACGTCGAGGTAGGA＜5 '.With the parietal cell overnight incubation that is inoculated on 6 orifice plates, PBS cleans one time, adds B, B1 respectively, B2, A group pastille serum is 2.5%, 5%, 10% to final concentration, and other establishes blank group (complete medium).Be 2 hours action time.H to all samples
+-K
+The expression product of ATPase α Subunit mRNA carries out semi-quantitative analysis.Analyze and respectively to organize pastille serum to the parietal cell H that exsomatizes
+-K
+The influence of ATPase α gene expression.6. banxia xiexin decoction is to rabbit body of stomach mucosa parietal cell H
+-K
+The influence that ATPase α subunit gene is expressed: each treated animal is cut open the belly rapidly after medication was got pastille serum on the 8th day after 7 days and is got stomach, and it is standby to get body of stomach mucosa liquid nitrogen cryopreservation.Take out the mucosa component from the purification parietal cell, behind the extraction cell total rna, through reverse transcription, the PCR reaction reaches carries out semi-quantitative analysis to the RT-PCR amplified production.
Experimental result: see Table 1-12,1-13
Table 1-12 respectively organizes 2.5% pastille serum to isolated rabbit parietal cell H
+-K
+The influence of ATPase α subunit expression (x ± s)
| Group | Quantity | Expression |
| Control group A group B group B1 group B2 group | 6 6 6 6 6 | 0.3569±0.1252 0.8365±0.1499** 0.9853±0.1949** 1.6843±0.0493** △△▲▲ 1.8377±0.2534** △△▲▲ |
* P＜0.01vs matched group group;
△ △P＜0.01vsA group;
▲ ▲P＜0.01vsB group
Table 1-13 banxia xiexin decoctions etc. are to rabbit gastric body mucosa parietal cell H
+-K
+ATPase α Subunit mRNA is expressed to be influenced (x ± s)
| Group | Quantity | Expression |
| A group B group B1 group B2 group | 4 5 5 5 | 0.7323±0.2111 0.8900±0.2209 1.0148±0.3251 1.4774±0.2998** △△▲ |
* P＜0.01vsA group;
△ △P＜0.01vsB group;
▲P＜0.05vsB1 group
The result shows, in the parietal cell of incubated overnight, add normal control group serum after, H
+-K
+ATPase α subunit expression strengthens, and means in the normal serum to have certain material, has directly stimulated the acid function of secreting of parietal cell.Remove the influence of normal serum, the pastille serum of B1, B2 side all can obviously strengthen H
+-K
+The expression of ATPase α subunit, and B side is to the parietal cell H that exsomatizes
+-K
+The expression of ATPase α subunit does not have obvious influence.With integral experiment not quite identical be, B1 side does not have obvious influence to the gastric acid secretion of normal rat and new zealand rabbit, and in isolated experiment, the pastille serum of B1 side then can strengthen the acid function of secreting of the parietal cell of exsomatizing.Banxia xiexin decoction is regulated gastric acid secretion and is raised parietal cell H
+-K
+ATPase α subunit expression is relevant.
Experimental example 4 is regulated the material base research of gastrointestinal motility
Experimental technique
Experiment is transferred peaceful albumen (calponin with 5 kinds of gut hormone levels, rat gastric antrum smooth muscle cell (SMC) contracted length, the interior free calcium level of gastric antrum smooth muscle cell, gastric antrum smooth muscle cell in the rat serum, CAP) and calcium mediation albumen (caldesmon, CAD) and gene expression dose be index, inquire into tripartite relation of regulating gastrointestinal motility and These parameters.
(1) three XIEXIN TANG to the gut hormone level affects and with the relation of stomach motion with rat stomach internal labeling thing blue dextran
[2]Relative residual rate be index, observe gastric emptying rate, the application of radiation immunological method is measured the content of motilin (MTL), gastrin (GAS), vasoactive intestinal peptide (VIP), P material (SP) and somatostatin (SS) in the rat serum simultaneously.
Experiment is divided into four groups: normal control group, banxia xiexin decoction group, Glycyrrhizae Decoction for Purging Stomach-Fire group, Shengjian Xiexin Tang group.
(2) gastric antrum smooth muscle contraction function is observed and is adopted the free technology of single gastric antrum smooth muscle cell to prepare gastric antrum smooth muscle cell suspension, adds pastille serum, measures 50 cells that run into micrometer under inverted microscope, calculates its average length.The cellular contraction reaction is represented with administration group cell average length (micron) and the percent that the cellular control unit average length descends
[3]Experiment divides four groups: normal control group, banxia xiexin decoction group, Glycyrrhizae Decoction for Purging Stomach-Fire group, Shengjian Xiexin Tang group.
(3) the free calcium level is observed and is adopted individual cells isolation technics separation gastric antrum smooth muscle cell in the gastric antrum smooth muscle cell, the good cell of calcium fluorescence indicator Fluo-3/AM load is dynamic observed under Laser Scanning Confocal Microscope, after adding pastille serum, fluorescence intensity changes in the continuous sweep cell
[4]Experiment divides four groups: normal control group, banxia xiexin decoction group, Glycyrrhizae Decoction for Purging Stomach-Fire group, Shengjian Xiexin Tang group.
(4) the observation rat successive administration of gastric antrum smooth muscle cell CAP and CAD and gene expression dose thereof is after 7 days, get gastric antrum portion tissue and make paraffin section, carry out morphological observation with the HE staining, use CAP (SP method) and CAD (ABC method) content in the SABC methods analyst gastric antrum smooth muscle cell
[5]Use RT-PCR method mensuration CAP mRNA and CAD mRNA concentration to observe CAP, CAD gene expression dose.Experiment divides four groups: normal control group, banxia xiexin decoction group, Glycyrrhizae Decoction for Purging Stomach-Fire group, Shengjian Xiexin Tang group.
Experimental result
See Table 1-14, table 1-15, table 1-16, table 1-17
The table 1-14 Rhizoma Pinelliae, Rhizoma Zingiberis Recens and Glycyrrhizae Decoction for Purging Stomach-Fire are to influence (x ± s) (1) of gut hormone in the rat serum
| Group | Quantity | MTL(pg/ml) | GAS(pg/ml) | VIP(pg/ml) |
| Matched group Rhizoma Pinelliae group Rhizoma Zingiberis Recens group | 15 15 15 15 | 162.59±19.54 179.07±17.46 205.03±21.95** ▲ 205.13±43.11* | 55.98±11.47 58.47±9.62 70.72±11.87* △ 79.74±9.26** ▲ | 445.76±101.86 348.11±102.54* 489.28±154.82 ▲ 457.28±118.96 △ |
The table 1-15 Rhizoma Pinelliae, Rhizoma Zingiberis Recens and Glycyrrhizae Decoction for Purging Stomach-Fire are to influence (x ± s) (2) of gut hormone in the rat serum
| Group | Quantity | SP(pg/ml) | SS(pg/ml) |
| Matched group Rhizoma Pinelliae group Rhizoma Zingiberis Recens group | 15 15 15 15 | 63.02±14.71 47.03±15.15* 53.83±20.76 61.49±14.83 △ | 461.66±77.04 562.32±149.69* 553.85±98.93* 530.52±93.72 |
Compare * p＜0.05, * * p＜0.01 with matched group; * p＜0.05, * * p＜0.01, vs matched group;
Compare with the banxia xiexin decoction group,
△P＜0.05,
▲P＜0.01;
△P＜0.05,
▲P＜0.01, vs Rhizoma Pinelliae group
The influence that the table 1-16 Rhizoma Pinelliae, Rhizoma Zingiberis Recens and Glycyrrhizae Decoction for Purging Stomach-Fire pastille serum (final concentration 2.5%) shrink the gastric antrum smooth muscle cell (x ± s)
| Group | Quantity | Cell length (μ m) | ||
| 1min | 5min | 10min | ||
| Matched group Rhizoma Pinelliae group Rhizoma Zingiberis Recens group | 5 5 5 5 | 78.63±18.11 65.24±14.81** 68.30±19.86* 69.41±15.73* | 77.80±14.68 63.07±16.20** 64.40±14.18** 67.34±16.80* | 77.62±14.14 67.83±15.65* 66.27±13.41* 67.80±16.47* |
Compare * p＜0.05, * * p＜0.01 with the normal control group;
* p＜0.05, * * p＜0.01, vs matched group
The influence that the table 1-17 Rhizoma Pinelliae, Rhizoma Zingiberis Recens and Glycyrrhizae Decoction for Purging Stomach-Fire pastille serum (final concentration 5%) shrink the gastric antrum smooth muscle cell (x ± s)
| Group | Quantity | Cell length (μ m) | ||
| 1min | 5min | 10min | ||
| Matched group Rhizoma Pinelliae group Rhizoma Zingiberis Recens group | 5 5 5 5 | 79.22±19.44 65.85±10.54** 66.49±16.83** 68.52±13.81* | 78.45±14.87 61.21±13.81** 64.30±13.30** 65.22±15.48** | 76.09±13.03 68.80±15.21* 67.42±15.40* 67.95±18.08* |
Compare * p＜0.05, * * p＜0.01 with the normal control group;
* p＜0.05, * * p＜0.01, vs matched group
The influence that the table 1-18 Rhizoma Pinelliae, Rhizoma Zingiberis Recens and Glycyrrhizae Decoction for Purging Stomach-Fire pastille serum (final concentration 10%) shrink the gastric antrum smooth muscle cell (x ± s)
| Group | Quantity | Cell length (μ m) | ||
| 1min | 5min | 10min | ||
| Matched group Rhizoma Pinelliae group Rhizoma Zingiberis Recens group | 5 5 5 5 | 78.04±16.91 65.61±15.60** 69.84±17.20* 70.07±16.34* | 76.48±15.23 62.42±12.20** 67.65±14.61* 67.80±11.46* | 75.83±15.54 68.24±13.21* 66.48±12.80* 71.61±14.43 |
Compare * p＜0.05, * * p＜0.01 with the normal control group;
* p＜0.05, * * p＜0.01, vs matched group
The table 1-19 Rhizoma Pinelliae, Rhizoma Zingiberis Recens and Glycyrrhizae Decoction for Purging Stomach-Fire pastille serum are to the influence of calcium ion peak value in the gastric antrum smooth muscle cell (x ± s)
| Group | Quantity | The calcium ion peak value | |
| Statistical value | Peak value | ||
| Matched group Rhizoma Pinelliae group Rhizoma Zingiberis Recens group | 5 5 5 5 | 261.20±40.49 257.60±46.47 268.80±53.59 265.60±32.68 | 291.80±33.34 402.40±66.98** 341.00±45.08 ▲ 279.60±39.38 ▲△ |
Compare * * p＜0.01 with the normal control group; * p＜0.01, the vs matched group,
Compare with the banxia xiexin decoction group,
▲P＜0.05;
▲P＜0.05, vs Rhizoma Pinelliae group,
Compare with the Shengjian Xiexin Tang group,
△P＜0.05;
△P＜0.05, vs Rhizoma Zingiberis Recens group
The tripartite pastille serum of table 1-20 (final concentration 5%) is to the influence of rat gastric antrum smooth muscle cell Calponin mRNA and Caldesmon mRNA (x ± s)
| Group | Quantity | Calponin mRNA | Caldesmon mRNA |
| Matched group Rhizoma Pinelliae group Rhizoma Zingiberis Recens group Radix Glycyrrhizae group | 4 4 4 4 | 1.072±0.139 0.374±0.166** 0.619±0.142* 0.735±0.094 | 0.951±0.055 0.344±0.067* 0.631±0.130 0.685±0.137 |
Compare * p＜0.05, * * p＜0.01 with the normal control group; * p＜0.05, * * p＜0.01, vs matched group
Conclusion
(1) three XIEXIN TANG can make VIP in the rat serum, SP level descend to the banxia xiexin decoction that concerns that influence reaches and stomach moves of gut hormone level, and the SS level raises.Shengjian Xiexin Tang can make motilin level, gastrin level, somatostatin level raise.Glycyrrhizae Decoction for Purging Stomach-Fire can make MTL level, GAS level raise.
Hormonal readiness in relative residual rate of pigment and the blood in the three XIEXIN TANG group rat stomach is carried out correlation analysis, found that banxia xiexin decoction promotes that gastric emptying may be relevant with VIP level in the reduction rat serum; Shengjian Xiexin Tang may be relevant with the SS level with VIP in its rat serum that raises to the inhibitory action of gastric emptying; Glycyrrhizae Decoction for Purging Stomach-Fire promotes that the trend of gastric emptying may be relevant with the MTL level that raises in the rat serum.
(2) influence of intracellular calcium is compared with matched group, and banxia xiexin decoction pastille serum obviously increases the photoluminescence peak of calcium ion in the gastric antrum smooth muscle cell, and the influence of the photoluminescence peak of calcium ion is not obvious in other two sides pastille serum pair cells.Relatively, intracellular Ca2+ photoluminescence peak size order is: banxia xiexin decoction group＞Shengjian Xiexin Tang group＞Glycyrrhizae Decoction for Purging Stomach-Fire group between tripartite group.
(3) CAP, CAD and gene expression research CAP thereof and CAD are two regulation and control Profilin of smooth muscle contraction.This experiment shows that CAP has different regulating actions with CAD to gastrointestinal smooth muscle, may mediate the regulating action of three XIEXIN TANG to gastrointestinal movement.Three XIEXIN TANG group rat gastric antrum smooth muscle cell CAP and CAD expression all have down regulation trend.
Showed by immune group result: banxia xiexin decoction group CAP down-regulated expression is the most obvious, and the CAP expression is Shengjian Xiexin Tang group＞Glycyrrhizae Decoction for Purging Stomach-Fire group＞banxia xiexin decoction group.The downward modulation of three XIEXIN TANG group CAD expressions is all comparatively remarkable.The CAD expression is Glycyrrhizae Decoction for Purging Stomach-Fire group＞Shengjian Xiexin Tang group＞banxia xiexin decoction group.
RT-PCR result shows: compare with matched group, three XIEXIN TANG group CAP mRNA and CAD mRNA expression be downward modulation to some extent also, and downward modulation result and SABC result are roughly the same.
Above result is similar to integral experiment, points out the contraction of three XIEXIN TANG pastille serum to the gastric antrum smooth muscle cell, and is relevant with its expression that directly changes CAP mRNA and CAD mRNA.
The research of experimental example 5 BP cerebellar model arithmetic computers screening compound recipe
The design of BP neural net model establishing
The selection of sample, the establishment of topological structure, algorithm summary execution are really being decided the performance of model with stable.At the research of banxia xiexin decoction and class side's compatibility rule thereof, for setting up the flavour of a drug dose nonlinear mapping model relevant, be provided with as follows to the parameter of network with pharmacodynamics.
(1) sample chooses and pretreatment
Sample is the interface of object of study and network model.For this problem, sample data will correctly reflect the different compatibilities of compound recipe as much as possible, takes the performance of network itself simultaneously again into account.
This project subject matter of choosing of sample is from flavour of a drug dose and medicine efficacy relation research banxia xiexin decoction and class side's compatibility rule thereof, so choose wherein 8 herbal medicines (Rhizoma Pinelliae, Rhizoma Zingiberis Recens, Radix Scutellariae, Rhizoma Coptidis, Radix Codonopsis, Radix Glycyrrhizae, Fructus Jujubae, Rhizoma Zingiberis) as input layer.These variablees can be represented the compatibility key element of three compound recipes.Choose gastric emptying, bowel movement, stomachial secretion function pharmacodynamic index as output neuron, set up the output layer neuron number and be 3 neural network model.
The standardization processing of sample is because the index of weighing has nothing in common with each other, and each number of components magnitude of original sample has very big difference, and this just need carry out standardization processing to sample.When adopting the neutral net regression forecasting, possible certain node input value is excessive, then its more new increment is excessive accordingly, thereby make weights excessive, cause other components almost to lose regulating and controlling effect, so need carry out appropriate standardization processing to original sample.The standardization processing of sample should design multiple scheme, under the prerequisite of as far as possible avoiding " saturated phenomenon ", makes sample uniform distribution in bigger amplitude range.For the data handling utility normalized function of this problem, make all inputs, output sample standard in the scope of [1,1].
(2) power and the initial codomain of threshold value determines
As a whole, power (comprising the threshold value item) is along with the carrying out of iteration upgrades, and generally is astringent, but the initial value of power is too big, may cause network to reach capacity soon.For avoiding saturated as far as possible, and it is initial at random equably to consider that the convergence rate of network and the complexity of sample data, less initial codomain are more conducive to weights, and it is better that the initial codomain of power is got [0.05,0.05].In this project initial weight is adopted the method for repeated trials in above-mentioned scope,, determine best initial weight by comparing fitting effect.
(3) network topology structure determines
Network topology structure all plays pivotal role for popularization ability, the computational efficiency of whole network.Owing to sample data is predicted output layer is got a node.The key of determining topology of networks is to determine suitable hidden layer number and hidden layer number of unit.
1. definite Kolmogorov theorem of hidden layer number: given arbitrary continuous function f:U
n→ R
m, f (x)=Y, U closes unit interval [0 here, 1], f can accurately realize that with three layers of feedforward network the ground floor of inferior network (being input layer) has n processing unit, there is 2n+1 processing unit in the intermediate layer, and the 3rd layer (being output layer) has m processing unit.This theorem guarantees that arbitrary continuous function can be realized by three layers of feedforward network.
BP theorem: given any ε＞0
With any L
2Function f: [0,1]
n→ R
m, there are three layers of BP network, it can approach f in the ε square error precision arbitrarily.Three layers of BP network of this theorem explanation can be realized L
2Function.
Above-mentioned two theorems illustrate that arbitrary continuous function can be by three layers of feedforward network realization.Simultaneously and since in the BP algorithm error by output layer to the input layer back propagation, hidden layer progression is many more, reverse propagated error is just unreliable more when the input layer.In addition, the many inevitable amounts of calculation of hidden layer progression are big, influence training speed.Therefore, at the problem of required solution in this research, determine to use the network of 1 hidden layer.
2. the node number of the input layer of definite network of hidden layer number of unit and output layer is according to the concrete condition of the real system of training function or institute's identification and fixed.Under the prerequisite that can guarantee training precision, the node number of network hidden layer necessarily can not be too many, as long as guarantee the number of hidden nodes greater than the input layer number, otherwise can influence training speed.The hidden layer unit number fixes under the foundation that satisfies above-mentioned theory really in this study model, has adopted " trial-and-error method " to determine.
(4) iteration stopping standard
During modelling, it is 5000 that maximum loop iteration is set; Cycle-index surpasses 5000 times and does not reach the iteration stopping standard yet, then shows the failure of an experiment.The convergence target allows there is error, less than given range of error, then stops the correction to network weight in the output of network and the difference between the sample.Employing can be accelerated the pace of learning of network to the way of e-learning tolerance.Adopt adaptive method, promptly allowable error is got a little louder when the training beginning, then along with training reduces gradually.Setting convergence error at this problem is 0.001
The BP neural network model is estimated
(1) evaluation criterion
Calculate the poor of predictive value and actual value, call the postreg function, calculate the relative coefficient r between predictive value and the actual value, calculate the mean square error mse between predictive value and actual value, converge to the required study number of times of target epochs, estimate the performance of more above-mentioned several network modeies with these indexs.
(2) flavour of a drug dose and bowel movement relational model are estimated
1. the mean square error flavour of a drug dose of measured value and match value and bowel movement mapping model MSE=9.9978e-004.
2. desired value and predictive value comparison flavour of a drug dose and bowel movement relational model are to the match situation of the data learnt and the data prediction ability of not learning is relatively seen Table 2-1.
3. Fig. 4 is seen in network model's training flavour of a drug dose and the training of bowel movement relational model.By training process figure as can be seen, through time iteration surplus 180, network convergence is to desired value 0.001.
4. model match situation flavour of a drug dose and bowel movement relational model match situation are seen Fig. 5.From the match situation map as can be seen, predictive value and the match of practical measurement value are better.
Table 2-1 flavour of a drug dose and bowel movement relational model desired value, predictive value and forecast error are relatively
| Group number | Desired value | Predictive value | Forecast error | ||||||
| Suppress model group | Normal group | Hyperfunction model group | Suppress model group | Normal group | Hyperfunction model group | Suppress model group | Normal group | Hyperfunction model group | |
| 1 2 3 · · · 22 23 24 | 0.6648 0.6396 0.6625 · · · 0.6319 0.6253 0.6419 | 0.4959 0.4655 0.5276 · · · 0.6344 0.6379 0.6345 | 0.7302 0.6920 0.7439 · · · 0.7362 0.7463 0.7120 | 0.6752 0.5918 0.6563 · · · 0.6638 0.6506 0.6838 | 0.4789 0.4392 0.5291 · · · 0.7048 0.7330 0.6336 | 0.7323 0.6927 0.7369 · · · 0.8724 0.8926 0.824 | 0.0104 -0.0478 -0.0062 · · · 0.0319 0.0253 0.0419 | -0.0170 -0.0263 0.0015 · · · 0.0704 0.0951 0.0009 | 0.0021 0.0007 -0.0070 · · · 0.1362 0.1463 0.1120 |
From table 2-1 as can be seen, the sample relative error of learning is less than 10%, and reflection network match performance is outstanding; The sample relative error of not learning is less than 20%, and is good from other one side reflection flavour of a drug dose and bowel movement relational model network estimated performance.
Flavour of a drug dose and gastric emptying relational model are estimated, and estimate similar to evaluation of stomachial secretion relational model and flavour of a drug dose to the bowel movement relational model to the flavour of a drug dose.
The model application result
With the banxia xiexin decoction is example, and flavour of a drug, dose and medicine efficacy relation result of study such as figure below (Fig. 6～Figure 29)
Description of drawings
Fig. 1 gastric acid regression equation fits comparison diagram
The regression equation of Fig. 2 pepsin activity fits comparison diagram
The regression equation of Fig. 3 gastric juice mucus binding capacity fits comparison diagram
Fig. 4 flavour of a drug dose and the training of bowel movement relational model
Fig. 5 flavour of a drug dose and bowel movement relational model match situation
The relation of Fig. 6 Rhizoma Pinelliae dosage and stomachial secretion
The relation of Fig. 7 Rhizoma Pinelliae dosage and gastric emptying
The relation of Fig. 8 Rhizoma Pinelliae dosage and bowel movement
The relation of Fig. 9 Radix Scutellariae dosage and stomachial secretion
The relation of Figure 10 Radix Scutellariae dosage and gastric emptying
The relation of Figure 11 Radix Scutellariae dosage and bowel movement
The relation of Figure 12 Rhizoma Coptidis dosage and stomachial secretion
The relation of Figure 13 Rhizoma Coptidis dosage and gastric emptying
The relation of Figure 14 Rhizoma Coptidis dosage and bowel movement
The relation of Figure 15 Radix Codonopsis dosage and stomachial secretion
The relation of Figure 16 Radix Codonopsis dosage and gastric emptying
The relation of Figure 17 Radix Codonopsis dosage and bowel movement
The relation of Figure 18 Rhizoma Zingiberis dosage and stomachial secretion
The relation of Figure 19 Rhizoma Zingiberis dosage and gastric emptying
The relation of Figure 20 Rhizoma Zingiberis dosage and bowel movement
The relation of Figure 21 Fructus Jujubae dosage and stomachial secretion
The relation of Figure 22 Fructus Jujubae dosage and gastric emptying
The relation of Figure 23 Fructus Jujubae dosage and bowel movement
The relation of Figure 24 Radix Glycyrrhizae dosage and stomachial secretion
The pass of Figure 25 Radix Glycyrrhizae dosage and gastric emptying
The relation of Figure 26 Radix Glycyrrhizae dosage and bowel movement
Figure 27 Rhizoma Zingiberis Recens dosage and stomachial secretion relation
The relation of Figure 28 Rhizoma Zingiberis Recens dosage and gastric emptying
The relation of Figure 29 Rhizoma Zingiberis Recens dosage and bowel movement
The following example all can be realized above-mentioned experiment effect.
The specific embodiment
Embodiment 1:
Get the crude drug Rhizoma Pinelliae, Rhizoma Zingiberis, make tablet according to a conventional method, every contains suitable Rhizoma Pinelliae crude drug amount 18g, contains suitable Rhizoma Zingiberis crude drug amount 12g.
Embodiment 2:
Get the crude drug Rhizoma Pinelliae, Rhizoma Zingiberis, make capsule according to a conventional method, every contains suitable Rhizoma Pinelliae crude drug amount 11g, contains suitable Rhizoma Zingiberis crude drug amount 17g.
Embodiment 3:
Get the crude drug Rhizoma Pinelliae, Rhizoma Zingiberis, make tablet according to a conventional method, every contains suitable Rhizoma Pinelliae crude drug amount 23g, contains suitable Rhizoma Zingiberis crude drug amount 28g.
Embodiment 4:
Get the crude drug Rhizoma Pinelliae, Rhizoma Zingiberis, make tablet according to a conventional method, every contains suitable Rhizoma Pinelliae crude drug amount 29g, contains suitable Rhizoma Zingiberis crude drug amount 24g.
Embodiment 5:
Get crude drug Radix Scutellariae, Rhizoma Coptidis, make tablet according to a conventional method, every contains suitable Radix Scutellariae crude drug amount 9g, contains suitable Rhizoma Coptidis crude drug amount 6g.
Embodiment 6:
Get crude drug Radix Scutellariae, Rhizoma Coptidis, make tablet according to a conventional method, every contains suitable Radix Scutellariae crude drug amount 18g, contains suitable Rhizoma Coptidis crude drug amount 4g.
Embodiment 7:
Get crude drug Fructus Jujubae, Radix Glycyrrhizae and make capsule according to a conventional method, every contains suitable Fructus Jujubae crude drug amount 11g, contains suitable Radix Glycyrrhizae crude drug amount 26g.
Embodiment 8:
Get crude drug Fructus Jujubae, Radix Glycyrrhizae and make capsule according to a conventional method, every contains suitable Fructus Jujubae crude drug amount 28g, contains suitable Radix Glycyrrhizae crude drug amount 21g.
Embodiment 9:
Get crude drug Fructus Jujubae, Radix Glycyrrhizae and make capsule according to a conventional method, every contains suitable Fructus Jujubae crude drug amount 11g, contains suitable Radix Glycyrrhizae crude drug amount 30g.
Embodiment 10:
Get crude drug Fructus Jujubae, Radix Glycyrrhizae and make capsule according to a conventional method, every contains suitable Fructus Jujubae crude drug amount 28g, contains suitable Radix Glycyrrhizae crude drug amount 38g.
Embodiment 11: utilize BP neural network model pharmaceutical compositions
1.P neural network model parameter setting
At the research of banxia xiexin decoction and class side's compatibility rule thereof, for setting up flavour of a drug dose non-linear the hint obliquely at model relevant, be provided with as follows to the parameter of network with pharmacodynamics:
(1) choosing and pretreatment of sample:
Choosing of sample: from flavour of a drug dose and medicine efficacy relation research banxia xiexin decoction and class side's compatibility rule thereof, so choose wherein 8 herbal medicines (Rhizoma Pinelliae, Rhizoma Zingiberis Recens, Radix Scutellariae, Rhizoma Coptidis, Radix Codonopsis, Radix Glycyrrhizae, Fructus Jujubae, Rhizoma Zingiberis) as input layer.These variablees can be represented the compatibility key element of three compound recipes.Choose gastric emptying, bowel movement, stomachial secretion function pharmacodynamic index as output neuron, set up the output layer neuron number and be 3 neural network model.
The standardization processing of sample: because the index of weighing has nothing in common with each other, each number of components magnitude of original sample has very big difference, and this just need carry out standardization processing to sample.When adopting the neutral net regression forecasting, possible certain node input value excessive then its more new increment is excessive accordingly, thereby make weights excessive, cause other components almost to lose regulating and controlling effect, so need carry out appropriate standardization processing to original sample.The standardization processing of sample should design multiple scheme, avoids under the prerequisite of " saturated phenomenon " as far as possible, makes sample uniform distribution in bigger amplitude range.For the data handling utility normalized function of this project problem, make all inputs, output sample standard in the scope of [1,1].
(2) power and the initial codomain of threshold value determines
As a whole, power (comprising the threshold value item) is along with the carrying out of iteration upgrades, and generally is astringent, but the initial value of power is too big, may cause network to reach capacity soon.For avoiding saturated as far as possible, consider the convergence rate of network and the complexity of sample data, it is initial at random equably that little initial codomain is more conducive to weights, initial codomain get (0.05,0.05) better.In this project initial weight is adopted the method for repeated trials in above-mentioned scope, cross relatively fitting effect, fixed best initial weight.
(3) network topology structure determines
Network topology structure all plays pivotal role for popularization ability, the computational efficiency of whole network.Owing to sample data is predicted output layer is got a node.The key of determining topology of networks is to determine suitable hidden layer number and hidden layer number of unit.
Determining of hidden layer number: the network of determining to use 1 hidden layer.
Determining of hidden layer number of unit: adopted " trial-and-error method " to determine
(4) iteration stopping standard
During modelling, putting maximum loop iteration is 5000; Cycle-index surpasses 5000 times and does not reach the iteration stopping standard yet, shows the failure of an experiment.The convergence target allows there is error, and the output of network and the difference between the sample stop the correction to network weight less than given range of error.Employing can be accelerated the pace of learning of network to the way of e-learning tolerance.Adopt adaptive method, promptly allowable error is got a little louder when the training beginning, and the back is along with training reduces gradually.Setting convergence error at this problem is 0.001.
(5) BP algorithm: use adding moment of inertia algorithm and adaptive step algorithm to improve the pace of learning of standard BP network
2. the foundation of the nonlinear mapping model of flavour of a drug dose and drug effect BP neutral net
With the experimental data is input, use the BP neutral net, set up the nonlinear mapping model of nonlinear mapping model, banxia xiexin decoction and class side's flavour of a drug dose and bowel movement of nonlinear mapping model, banxia xiexin decoction and the class side's flavour of a drug dose and the gastric emptying of banxia xiexin decoction and class side's flavour of a drug dose and stomachial secretion (gastric acid, pepsin and gastric mucus secretion) respectively.Being set at of each parameter in the network: input layer is 8, and output layer is 3, and hidden layer is 10, and topological structure is 8-10-3, and learning algorithm is selected " the BP algorithms that activity level method and learning rate self adaptation are adjusted two kinds of stragetic innovations ", and the network convergence goal setting is 1E-3.
3. model evaluation
By using the difference of mean square error function comparison object value and predictive value, calculate the error between desired value and predictive value, examine network model to training, the network fitted figure is estimated for network performance.
(1) flavour of a drug dose and bowel movement relational model are estimated
The mean square error of measured value and match value is that flavour of a drug dose and bowel movement mapping model MSE are 9.9978e-004.
Desired value and predictive value comparison flavour of a drug dose and bowel movement relational model are to the match situation of the data learnt and the data prediction ability of not learning is relatively shown, less than 10%, reflect that network match performance is outstanding for the sample relative error of learn; , good for the sample relative error of not learning from other one side reflection flavour of a drug dose and bowel movement relational model network estimated performance less than 20%.
(2) flavour of a drug dose and gastric emptying relational model are estimated
The mean square error flavour of a drug dose of measured value and match value and gastric emptying mapping model MSE are 0.0015.
Desired value and predictive value comparison flavour of a drug dose and gastric emptying relational model are to the match situation of the data learnt and the data prediction ability of not learning is relatively shown, less than 10%, reflect that network match performance is outstanding for the sample relative error of learn; , good for the sample relative error of not learning from other one side reflection network estimated performance less than 20%.
(3) flavour of a drug dose and stomachial secretion relational model are estimated
The mean square error of measured value and match value is that flavour of a drug dose and stomachial secretion mapping model model M SE are 14.6922.
Desired value and predictive value comparison flavour of a drug dose and stomachial secretion relational model are to the match situation of the data learnt and the data prediction ability of not learning is relatively shown, less than 10%, reflect that network match performance is outstanding for the sample error of learn; , good for the sample error of not learning from other one side reflection flavour of a drug dose and stomachial secretion relational model network estimated performance less than 20%.
Draw flavour of a drug and the dose proportion relation of embodiment 1-embodiment 10 according to predicting the outcome, get crude drug respectively and be prepared into described tablet of embodiment 1-10 or capsule.
Claims (11)
1, a kind of pharmaceutical composition is characterized in that containing suitable Rhizoma Pinelliae crude drug amount 10-21g in the unit formulation of this pharmaceutical composition, contains suitable Rhizoma Zingiberis crude drug amount 10-18g.
2, pharmaceutical composition as claimed in claim 1 has the application in the middle of the medicine of following effect in preparation: gastric acid inhibitory effect, suppress the pepsin effect or promote the effect of gastric mucus.
3, a kind of pharmaceutical composition is characterized in that containing suitable Rhizoma Pinelliae crude drug amount 22-30g in the unit formulation of this pharmaceutical composition, contains suitable Rhizoma Zingiberis crude drug amount 19-30g.
4, pharmaceutical composition as claimed in claim 3 has the application in the middle of the medicine of following effect in preparation: have the effect of the gastric acid of promotion, promote the pepsin effect or suppress the gastric mucus effect.
5, a kind of pharmaceutical composition is characterized in that containing suitable Radix Scutellariae crude drug amount 2-20g in the unit formulation of this pharmaceutical composition, contains suitable Rhizoma Coptidis crude drug amount 2-15g.
6, pharmaceutical composition as claimed in claim 5 has the application in the middle of the medicine of following effect in preparation: have gastric acid secretion, suppress the gastric mucus effect.
7, a kind of pharmaceutical composition is characterized in that containing suitable Fructus Jujubae crude drug amount 9-30g in the unit formulation of this pharmaceutical composition, contains suitable Radix Glycyrrhizae crude drug amount 20-27g.
8, pharmaceutical composition as claimed in claim 7 has the application in the middle of the medicine of following effect in preparation: have the gastric acid secretion effect, promote the pepsin effect.
9, a kind of pharmaceutical composition is characterized in that containing suitable Fructus Jujubae crude drug amount 9-30g in the unit formulation of this pharmaceutical composition, contains suitable Radix Glycyrrhizae crude drug amount 28-40g.
10, as claim 1,3,5,7 or described preparation of drug combination method, it is characterized in that this method is:
1.BP neural network model parameter setting
At the research of banxia xiexin decoction and class side's compatibility rule thereof, for setting up flavour of a drug dose non-linear the hint obliquely at model relevant, be provided with as follows to the parameter of network with pharmacodynamics:
(1) choosing and pretreatment of sample:
Choosing of sample: from flavour of a drug dose and medicine efficacy relation research banxia xiexin decoction and class side's compatibility rule thereof, so choose wherein 8 herbal medicines (Rhizoma Pinelliae, Rhizoma Zingiberis Recens, Radix Scutellariae, Rhizoma Coptidis, Radix Codonopsis, Radix Glycyrrhizae, Fructus Jujubae, Rhizoma Zingiberis) as input layer, these variablees can be represented the compatibility key element of three compound recipes, choose gastric emptying, bowel movement, stomachial secretion function pharmacodynamic index as output neuron, set up the output layer neuron number and be 3 neural network model;
The standardization processing of sample: because the index of weighing has nothing in common with each other, each number of components magnitude of original sample has very big difference, this just need carry out standardization processing to sample, when adopting the neutral net regression forecasting, possible certain node input value excessive then its more new increment is excessive accordingly, thereby make weights excessive, cause other components almost to lose regulating and controlling effect, so need carry out the standardization processing of appropriateness to original sample, the standardization processing of sample should design multiple scheme, avoid under the prerequisite of " saturated phenomenon " as far as possible, make sample uniform distribution in bigger amplitude range, data handling utility normalized function for this project problem makes all inputs, the output sample standard is in the scope of [1,1];
(2) power and the initial codomain of threshold value determines
As a whole, power (comprising the threshold value item) is along with the carrying out of iteration upgrades, and generally be astringent, but the initial value of power is too big, may cause network to reach capacity soon, for avoiding saturated as far as possible, and consider the convergence rate of network and the complexity of sample data, it is initial at random equably that less initial codomain is more conducive to weights, it is better that the initial codomain of power is got (0.05,0.05), in this project initial weight adopted the method for repeated trials in above-mentioned scope, by comparing fitting effect, determine best initial weight;
(3) network topology structure determines
Network topology structure all plays pivotal role for popularization ability, the computational efficiency of whole network, and in sample data is predicted, output layer is got a node, and the key of deciding topology of networks is to determine suitable hidden layer number and hidden layer number of unit;
Determining of hidden layer number: the network of determining to use 1 hidden layer;
Determining of hidden layer number of unit: adopted " trial-and-error method " to determine;
(4) iteration stopping standard
During modelling, it is 5000 that maximum loop iteration is set; Cycle-index surpasses 5000 times and does not reach the iteration stopping standard yet, then show the failure of an experiment, hold back target and allow there is error, less than given range of error, then stop correction network weight in the output of network and the difference between the sample, with way to the e-learning tolerance, can accelerate the pace of learning of network, use adaptive method, promptly allowable error is got a little bigger when the training beginning, along with training reduces gradually, it is 0.001 that this problem is set convergence error then;
(5) BP algorithm: use adding moment of inertia algorithm and adaptive step algorithm to improve the pace of learning of standard BP network
2. the foundation of the nonlinear mapping model of flavour of a drug dose and drug effect BP neutral net
With the experimental data is input, use the BP neutral net, banxia xiexin decoction and class side flavour of a drug dose and stomachial secretion (gastric acid have been set up respectively, pepsin and gastric mucus secretion) the nonlinear mapping model, the nonlinear mapping model of banxia xiexin decoction and class side's flavour of a drug dose and gastric emptying, the nonlinear mapping model of banxia xiexin decoction and class side's flavour of a drug dose and bowel movement, being set at of each parameter in the network: input layer is 8, output layer is 3, hidden layer is 10, and topological structure is 8-10-3, learning algorithm is selected " the BP algorithm that activity level method and learning rate self adaptation are adjusted two kinds of stragetic innovations ", and the network convergence goal setting is 1E-3;
3. model evaluation
By using the difference of mean square error function comparison object value and predictive value, calculate the error between desired value and predictive value, observe network model to training, the network fitted figure is estimated for network performance;
(1) flavour of a drug dose and bowel movement relational model are estimated
The mean square error of measured value and match value is that dose and bowel movement mapping model MSE are 9.9978e-004;
Desired value and predictive value comparison flavour of a drug dose and bowel movement relational model are to the match situation of the data learnt and the data prediction ability of not learning is relatively shown, less than 10%, reflect that network match performance is outstanding for the sample relative error of learn; , good for the sample relative error of not learning from other one side reflection flavour of a drug dose and bowel movement relational model network estimated performance less than 20%;
(2) flavour of a drug dose and gastric emptying relational model are estimated
The mean square error flavour of a drug dose of measured value and match value and gastric emptying mapping model MSE are 0.0015; Desired value and predictive value comparison flavour of a drug dose and gastric emptying relational model are to the match situation of the data learnt and the data prediction ability of not learning is relatively shown, less than 10%, reflect that network match performance is outstanding for the sample relative error of learn; , good for the sample relative error of not learning from other one side reflection network estimated performance less than 20%;
(3) flavour of a drug dose and stomachial secretion relational model are estimated
The mean square error of measured value and match value is that flavour of a drug dose and stomachial secretion mapping model model M SE are 14.6922;
Desired value and predictive value comparison flavour of a drug dose and stomachial secretion relational model are to the match situation of the data learnt and the data prediction ability of not learning is relatively shown, less than 10%, reflect that network match performance is outstanding for the sample error of learn; , good for the sample error of not learning from other one side reflection flavour of a drug dose and stomachial secretion relational model network estimated performance less than 20%;
Draw the flavour of a drug and the dose proportion relation of pharmaceutical composition according to predicting the outcome, get crude drug respectively and be prepared into drug combination preparation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200610083340 CN1876172A (en) | 2005-06-02 | 2006-06-02 | BP neural net model method for preparing pharmaceutical composition |
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200510073524 | 2005-06-02 | ||
| CN200510073524.4 | 2005-06-02 | ||
| CN 200610083340 CN1876172A (en) | 2005-06-02 | 2006-06-02 | BP neural net model method for preparing pharmaceutical composition |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1876172A true CN1876172A (en) | 2006-12-13 |
Family
ID=37508842
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200610083340 Pending CN1876172A (en) | 2005-06-02 | 2006-06-02 | BP neural net model method for preparing pharmaceutical composition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1876172A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102355524A (en) * | 2011-07-05 | 2012-02-15 | 浙江迦南科技股份有限公司 | Remote pharmaceutical formula automation acquiring method based on mobile phone |
| CN102798704A (en) * | 2012-09-06 | 2012-11-28 | 杨晔宏 | Method of analyzing, forecasting and optimizing drug functions by utilizing component quantification and visualization |
| CN105144177A (en) * | 2013-04-24 | 2015-12-09 | 费森尤斯卡比德国有限公司 | Method of operating a control device for controlling an infusion device |
| WO2018103131A1 (en) * | 2016-12-08 | 2018-06-14 | 深圳先进技术研究院 | Method and system for determining formula proportions of biological tissue simulation material |
| CN110277174A (en) * | 2019-06-14 | 2019-09-24 | 上海海洋大学 | A kind of prediction technique of anticancer drug synergistic effect neural network based |
| WO2020004391A1 (en) * | 2018-06-29 | 2020-01-02 | 小林製薬株式会社 | Herbal medicine prescription information processing system, information processing method, and program |
| CN113509386A (en) * | 2021-09-10 | 2021-10-19 | 江西中医药大学 | Decoction monitoring method and device, electronic equipment and storage medium |
| CN114869764A (en) * | 2022-05-10 | 2022-08-09 | 浙江厚达智能科技股份有限公司 | Method for decocting individual prescription of Chinese medicine |
| CN116711016A (en) * | 2020-08-24 | 2023-09-05 | 肽逻辑公司 | Artificial intelligence engine for generating candidate drugs using experimental validation and peptide drug optimization |
-
2006
- 2006-06-02 CN CN 200610083340 patent/CN1876172A/en active Pending
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102355524A (en) * | 2011-07-05 | 2012-02-15 | 浙江迦南科技股份有限公司 | Remote pharmaceutical formula automation acquiring method based on mobile phone |
| CN102798704A (en) * | 2012-09-06 | 2012-11-28 | 杨晔宏 | Method of analyzing, forecasting and optimizing drug functions by utilizing component quantification and visualization |
| CN102798704B (en) * | 2012-09-06 | 2014-07-02 | 杨晔宏 | Method of analyzing, forecasting and optimizing drug functions by utilizing component quantification and visualization |
| CN105144177A (en) * | 2013-04-24 | 2015-12-09 | 费森尤斯卡比德国有限公司 | Method of operating a control device for controlling an infusion device |
| WO2018103131A1 (en) * | 2016-12-08 | 2018-06-14 | 深圳先进技术研究院 | Method and system for determining formula proportions of biological tissue simulation material |
| WO2020004391A1 (en) * | 2018-06-29 | 2020-01-02 | 小林製薬株式会社 | Herbal medicine prescription information processing system, information processing method, and program |
| TWI703580B (en) * | 2018-06-29 | 2020-09-01 | 日商小林製藥股份有限公司 | Information processing system, information processing method and program for Chinese medicine prescription |
| CN110277174A (en) * | 2019-06-14 | 2019-09-24 | 上海海洋大学 | A kind of prediction technique of anticancer drug synergistic effect neural network based |
| CN110277174B (en) * | 2019-06-14 | 2023-10-13 | 上海海洋大学 | Neural network-based prediction method for anticancer drug synergistic effect |
| CN116711016A (en) * | 2020-08-24 | 2023-09-05 | 肽逻辑公司 | Artificial intelligence engine for generating candidate drugs using experimental validation and peptide drug optimization |
| CN113509386A (en) * | 2021-09-10 | 2021-10-19 | 江西中医药大学 | Decoction monitoring method and device, electronic equipment and storage medium |
| CN114869764A (en) * | 2022-05-10 | 2022-08-09 | 浙江厚达智能科技股份有限公司 | Method for decocting individual prescription of Chinese medicine |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1876172A (en) | BP neural net model method for preparing pharmaceutical composition | |
| Walker et al. | Enhancing yeast alcoholic fermentations | |
| CN1862551A (en) | A treatment effect prediction system, a treatment effect prediction method, and a computer program product thereof | |
| CN1780560A (en) | Method for producing ethanol using raw starch | |
| CN1313030C (en) | Novel nutraceutical compositions comprising biotin | |
| CN1977042A (en) | Non-recombinant saccharomyces strains that grow on xylose | |
| CN101057635A (en) | Low-salinity autumn and winter cultivation ecological bait for litopenaeus vannamei and its preparation method | |
| CN1298862C (en) | Glycoprotein and process for producing same | |
| CN1134727A (en) | Arabinoxylan degrading enzymes | |
| CN1195843C (en) | Solid culture method of antrodia camphorate, its cultured solid substance, and its product and application | |
| CN1560228A (en) | Method of whole manual culture for chinese caterpillar fungus | |
| Wang et al. | Optimization of innovative ethanol production from wheat by response surface methodology | |
| de Winde | Functional genetics of industrial yeasts; of ancient skills and modern applications | |
| CN101035433A (en) | Fermented milk or vegetable proteins comprising receptor ligand and uses thereof | |
| CN1699297A (en) | Process for preparing organic liquid fertilizer and organic solid fertilizer | |
| CN1850122A (en) | Method for preparing Chinese herbs or natural product to make extract component stable | |
| CN1662644A (en) | Method of producing human beta cell lines | |
| CN1124346C (en) | Novel yeast genes | |
| CN1680240A (en) | Composition containing resveratrol and soybean isoflavone and its preparation and use | |
| CN1557352A (en) | Novel usage of chicory aqueous extract | |
| CN1208448C (en) | Angelica dehurica black beer | |
| CN101062947A (en) | Ciliary neurotrophy factor polypeptide allosteric and application thereof | |
| CN1296080C (en) | Liver clearing blood pressure lowering capsule nd its preparation method | |
| CN1816353A (en) | Medicinal composition for preventing or treating Th1 type immunological disease | |
| CN1200097C (en) | Recombinant human epidermal growth factor and its preparing process and medicinal composition |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Open date: 20061213 |