A matching method between disease and medicine based on decision tree

Abstract

The present invention discloses a method for matching symptoms and drugs based on a decision tree. The present invention can intuitively and vividly reflect the medication rules of drugs. Doctors can quickly find drugs that can be treated according to the symptoms of patients' symptoms and form new drug combinations. This can provide a reference basis for determining treatment plans, recommend reasonable, effective and efficient drugs to doctors, reduce the situation of inappropriate drugs, provide strong decision-making support for physicians in drug research and development, reduce the degree of dependence on doctors' experience, improve doctors' work efficiency, reduce the time for drug research and development, provide new ideas for developing new drugs, and open up new ways for the research and development of traditional formulas of ethnic medicine.

Description

Technical field:

[0001] The present invention relates to the field of biomedical technology, and in particular to a method for matching symptoms and drugs based on a decision tree. Background technology:

[0002] In the traditional Chinese and Mongolian medicine treatment process, doctors need to prescribe medicine for patients based on what they have learned. However, due to the complexity and diversity of symptoms and the efficacy of medicinal materials, the doctor is highly dependent on his or her experience when prescribing drugs. This leads to irrational use of drugs, which will directly affect the treatment effect and even the health of the patient.

[0003] At present, there are many methods to study the relationship between the drugs in the prescription and the diseases they treat, and most of them use the method of decomposing the prescription, such as the drug pair research method, the single drug research method, etc. This kind of method can obviously only change one or two drugs while keeping other drugs unchanged, explore the relationship between the drugs in the prescription and the disease, and study one or two drugs for the diseases treated in the prescription. Obviously, it is not possible to accurately conduct a comprehensive study on the nonlinear mapping relationship between the diseases treated in the prescription and the drugs from an overall perspective, and it is impossible to provide a reference for doctors to prescribe medicines, and it is impossible to efficiently and effectively reduce the situation of irrational drug use and the degree of dependence on doctors' experience. Summary of the invention:

[0004] The purpose of the present invention is to provide a method for matching symptoms and drugs based on a decision tree, which can provide a reference for determining a treatment plan, recommend reasonable, effective and efficient drugs to doctors, reduce the situation of irrational use of drugs, reduce the dependence on doctors' experience, and improve doctors' work efficiency.

[0005] The present invention is implemented by the following technical solutions:

[0006] A method for matching disease and medicine based on a decision tree comprises the following steps:

[0007] S1. Collect several traditional prescriptions, match the symptoms X corresponding to the prescriptions with the drugs Y used, and form a prescription sample set Z;

[0008] S2, using part of the formulas in the formula sample set Z established in S1 as a training formula set, calculating the information entropy of the drugs in the training formula set and the conditional entropy of the drugs under the given condition of the disease symptoms according to the training formula set, and then calculating the information gain of the disease symptoms to the drugs by using the information entropy of the drugs and the conditional entropy of the drugs under the given condition of the disease symptoms;

[0009] S3, constructing a drug decision tree corresponding to each drug according to the information gain of the disease symptoms on the drug calculated in S2;

[0010] S4, taking the symptoms X' of the case to be prescribed as input parameters, and sequentially inputting them into the multiple drug decision trees obtained in S3, and obtaining a conclusion on whether the symptoms X' of the case matches the drug corresponding to the drug decision tree;

[0011] S5. All drugs matching the symptoms X' are combined to obtain drug results Y=(y 1 ,y 2 ,…,y J ).

[0012] Specifically, in S1,

[0013] Recipe sample set Z = {(X 1 ,Y 1 ),(X 2 ,Y 2 ),…,(X n ,Y n ),…,(X N ,Y N )},

[0014] Where N is the total number of recipes in the recipe sample set;

[0015] The symptom characteristic corresponding to the nth prescription is recorded as In vector is the symptom x of the i-th disease in the n-th recipe i ; There are two possible situations: when the ith symptom exists in the nth prescription, the drug combination in the nth prescription can treat the symptom. It is recorded as When the ith symptom does not exist in the nth prescription, that is, the drug combination in the nth prescription does not treat the symptom. It is recorded as

[0016] The drug combination corresponding to the nth prescription is recorded as In vector It is the jth drug y in the nth recipe j ; There are two possible situations. When the jth drug is used in the nth recipe, it is recorded as When the jth drug is not used in the nth formulation, it is recorded as

[0017] Specifically, in S2,

[0018] The information gain of disease symptoms on drugs G(yj ,x i ) is calculated as shown in Formula 1:

[0019] G(y j ,x i )=H(y j )-E(y j |x i )(1)

[0020] In formula 1, H(y j ) is the information entropy of the aforementioned drug, E(y j |x i ) is the conditional entropy of the drug under the given condition of the disease symptoms mentioned above. The disease symptom x can be calculated by formula 1. i For drugs j The information gain G(y j ,x i ).

[0021] Specifically, in S2,

[0022] The information entropy of the drug H(y j ) is calculated as shown in Formula 2:

[0023]

[0024] In formula 2, l = 1, 2 represents drug y j There are two cases: used and not used; N(y jl ) indicates that the drug is in y in the training formula set jl The total number of cases; N represents the total number of samples in the training formula set; Indicates that the drug is in y in the training formula set jl The probability of

[0025] The conditional entropy E(y) of the drug under the given condition of the disease symptoms j |x i ) is calculated as shown in Formula 3:

[0026]

[0027] In formula 3, t=1, 2 represents being in and not in the condition symptom x i Two cases; N(x it ) indicates that the symptoms of the disease are in x in the training formula set it The total number of times when Indicates that the symptoms of the disease are in x in the training formula set it The probability of the situation; N( it ∩y jl ) indicates that the symptoms of the disease are in x in the training formula setit Condition and drug in y jl The total number of times when Indicates that in the training formula set, when the symptom is in x it Condition and drug in y jl The probability of the situation.

[0028] Specifically, in S3,

[0029] The method for constructing a drug decision tree corresponding to each drug is as follows:

[0030] The symptoms of various diseases in traditional prescriptions are used as attributes to form the internal nodes of the decision tree; whether the drug is used is used as the classification result to form the leaf nodes.

[0031] Specifically, in S3,

[0032] Select the drug in y jl In this case, the symptom with the largest information gain in the traditional formula is used as the root node; and other symptom in the traditional formula are used as internal nodes.

[0033] Advantages of the present invention:

[0034] The present invention can intuitively and vividly reflect the medication rules of drugs. Doctors can quickly find drugs that can be used for treatment according to the patient's symptoms and form new drug combinations. This can provide a reference for determining treatment plans, recommend reasonable, effective and efficient drugs to doctors, reduce the situation of inappropriate medication, provide strong decision-making support for physicians in drug research and development, reduce the degree of dependence on doctors' experience, improve doctors' work efficiency, reduce the time for drug research and development, provide new ideas for the development of new drugs, and open up new ways for the research and development of traditional formulas of ethnic medicine. Description of the drawings:

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0036] Figure 1 A flowchart for constructing a drug decision tree in the present invention;

[0037] Figure 2 A flowchart of drug selection according to a drug decision tree in the present invention;

[0038] Figure 3 The agarwood decision tree obtained by using the present invention;

[0039] Figure 4 A sandalwood decision tree obtained by using the present invention;

[0040] Figure 5 A decision tree for Choerospondias axillaris obtained by using the present invention;

[0041] Figure 6 A decision tree of Aucklandia lappa obtained by using the present invention;

[0042] Figure 7 A decision tree for white cardamom obtained by using the present invention;

[0043] Figure 8 The Terminalia chebula decision tree obtained by using the present invention;

[0044] Fig. 9 A decision tree of Ferula foetida obtained by using the present invention;

[0045] Fig.10 A decision tree of Heiyunxiang obtained by using the present invention;

[0046] Fig.11 A decision tree for the purple borax obtained by using the present invention;

[0047] Fig.12 A Sophora flavescens decision tree obtained by using the present invention;

[0048] Fig.13 The present invention is used to obtain the lilac decision tree. Specific implementation method:

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] Embodiment 1:

[0051] like Figure 1 , Figure 2 A method for matching disease and medicine based on a decision tree is shown, comprising the following steps:

[0052] S1. Collect 32 traditional Mongolian medicine prescriptions for the treatment of Heyi disease, and match the symptoms X corresponding to the prescriptions with the medicines Y used to form a prescription sample set Z = {(X 1 ,Y 1 ),(X 2 ,Y 2 ),…,(X n ,Y n),…,(X 32 ,Y 32 )}, as shown in Table 1;

[0053] Table 1 Recipe sample set

[0054]

[0055]

[0056]

[0057]

[0058] Table 1 involves 47 symptoms and 78 medicines. i (i=1,2,…,47) and drug y j (j=1,2,…,78) are listed in Table 2.

[0059] Table 2 Symptoms and medications

[0060]

[0061] S2. The first 30 recipes in the recipe sample set Z established in S1 are used as the training recipe set, as shown in Table 3; the last 2 are used as the test recipe set.

[0062] Table 3 Training recipe set

[0063]

[0064]

[0065]

[0066]

[0067] According to the training formula set shown in Table 3, the information entropy of the drugs in the training formula set and the conditional entropy of the drugs under the given condition of the disease symptoms are calculated, and then the information gain of the disease symptoms to the drugs is calculated using the information entropy of the drugs and the conditional entropy of the drugs under the given condition of the disease symptoms; specifically:

[0068] The information entropy of the drug H(y j ) is calculated as shown in Formula 2:

[0069]

[0070] In formula 2, l = 1, 2 represents drug y j There are two cases: used and not used; N(y jl ) indicates that the drug is in y in the training formula setjl The total number of cases; N represents the total number of samples in the training formula set; Indicates that the drug is in y in the training formula set jl The probability of

[0071] The conditional entropy E(y) of the drug under the given condition of the disease symptoms j |x i ) is calculated as shown in Formula 3:

[0072]

[0073] In formula 3, t=1, 2 represents being in and not in the condition symptom x i Two cases; N(x it ) indicates that the symptoms of the disease are in x in the training formula set it The total number of times when Indicates that the symptoms of the disease are in x in the training formula set it The probability of the situation; N(x it ∩y jl ) indicates that the symptoms of the disease are in x in the training formula set it Condition and drug in y jl The total number of times when Indicates that in the training formula set, when the symptom is in x it Condition and drug in y jl probability of the situation;

[0074] The information gain of disease symptoms on drugs G(y j ,x i ) is calculated as shown in Formula 1:

[0075] G(y j ,x i )=H(y j )-E(y j |x i )(1)

[0076] In formula 1, H(y j ) is the information entropy of the aforementioned drug, E(y j |x i ) is the conditional entropy of the drug under the given condition of the disease symptoms mentioned above. The disease symptom x can be calculated by formula 1. i For drugs j The information gain G(y j ,x i ).

[0077] S3. Based on the information gain of the symptoms on the medicine calculated in S2, a decision tree is constructed for each Mongolian medicine in the sample set, and a total of 78 Mongolian medicine decision trees are constructed. The specific method is: using the symptoms of each disease in the traditional prescription as attributes to form the internal nodes of the decision tree; using whether the medicine is used as the classification result to form the leaf node; specifically, selecting the medicine in y jl In this case, the symptom with the largest information gain in the traditional formula is used as the root node; and other symptom in the traditional formula are used as internal nodes.

[0078] Take the decision tree of agarwood as an example. 2 express.

[0079] According to formula 2, the information entropy H(y 2 ), there are 30 recipes in the training sample set, so the total number of sample sets N is 30, of which there are 19 recipes using agarwood and 11 recipes not using agarwood. The specific calculation process is as follows:

[0080]

[0081] According to formula 3, calculate the symptoms x i The conditional entropy of agarwood under given conditions can be obtained based on the data of symptoms and drugs in the formula training set: the total number of times each symptom exists or not, and the total number of times each symptom and agarwood are at different values. There are six different value situations: the presence of symptoms is represented by x i1 Indicates: if the symptom does not exist, use x i2 Indicates; symptoms of the disease exist and the use of agarwood is x i1 ∩y 21 Indicates that the symptoms of the disease exist and agarwood is not used. i1 ∩y 22 Indicates that the symptoms of the disease do not exist and the use of agarwood is x i2 ∩y 21 Indicates that the symptoms do not exist and no agarwood is used. i2 ∩y 22 The statistical data are shown in Table 4.

[0082] Table 4 Total number of statistics

[0083]

[0084] In this embodiment, palpitations are taken as an example for explanation. 1 According to the statistical data under the palpitation symptoms in Table 4, the conditional entropy E(y) of agarwood under the condition that the palpitation symptoms are known is calculated using Formula 3. 2 |x 1 ), as follows:

[0085]

[0086]

[0087] Finally, the obtained H(y 2 ) and E(y 2 |x 1 ) into formula 1, and the information gain G(y) of palpitations under the condition of agarwood is obtained. 2 ,x 1 ),

[0088] G(y 2 ,x 1 )=H(y 2 )-E(y 2 |x 1 )=0.948-0.919=0.029

[0089] Similarly, the conditional entropy of agarwood under other symptoms and the information gain of its symptoms can be obtained. Some data are listed in Table 5.

[0090] Table 5 Conditional entropy and information gain

[0091]

[0092]

[0093] According to Figure 1The process shown in the figure selects the symptom with the largest information gain as the root node of the decision tree according to the principle of the ID3 algorithm. From the comparison of the information gain of each symptom in Table 5, it can be seen that for agarwood, asthma has the highest information gain value, so asthma is selected as the first internal node of the decision tree, that is, the root node. According to the two values ​​of asthma, a branch subnode is established. If asthma exists, a right branch subnode is established. Then, the data with asthma in the training set is removed after the asthma symptoms are operated to generate a data subset, and it is judged whether the data subset uses agarwood. Therefore, a leaf node using agarwood is obtained; if asthma does not exist, a left branch subnode is established. Then, the data without asthma in the training set is also removed after the asthma symptoms are operated to generate a data subset. Then it is judged whether the data subset uses agarwood. If agarwood is used, a leaf node using agarwood can be directly generated; if agarwood is not used, it is necessary to further judge whether agarwood is not used at all. If agarwood is not used at all, a leaf node not using agarwood can be directly generated. If agarwood is not used at all, it is necessary to calculate the information gain of symptoms in the data subset again, and select the symptoms with the highest information gain value, that is, select He Yi and Xue Xiangbo as the internal node of the left branch. Next, according to the same method, other internal nodes are recursively constructed with the same symptom selection method until there is no situation that can be divided on the branch, that is, agarwood is used or not used, and then a leaf node is obtained. The agarwood decision tree obtained based on this principle is as follows Figure 3 shown.

[0094] exist Figure 3 In the agarwood decision tree, the internal nodes are symptoms of the disease, and the leaf nodes are whether to use agarwood. The leaf nodes that use agarwood are displayed as agarwood, and the leaf nodes that do not use agarwood are displayed as none. In the agarwood decision tree, we can know that there are many medication rules for agarwood. Agarwood is used when the patient has asthma; agarwood is used when the patient has a conflict between He Yi and blood; agarwood can be used when the patient has epilepsy and does not feel chest tightness; agarwood is used if the patient feels dizzy. Any path from the root node to the leaf node of the agarwood decision tree is a medication rule for agarwood.

[0095] Similarly, according to step 2, the information entropy of other Mongolian medicines in the training set and the conditional entropy and information gain under given disease symptoms are calculated, and then the decision tree for the remaining medicines can be constructed.

[0096] S4, taking the symptoms X' of the case to be prescribed as input parameters, and sequentially inputting them into the multiple drug decision trees obtained in S3, and obtaining a conclusion on whether the symptoms X' of the case matches the drug corresponding to the drug decision tree;

[0097] S5. All drugs matching the symptoms X' are combined to obtain the drug result Y=(y 1 ,y2 ,…,y J ).

[0098] Experimental Example 1:

[0099] The symptoms corresponding to the two prescriptions in the test prescription set in Example 1 were used as input parameters, and the method of the present invention was used to obtain drugs that matched them, thereby verifying the consistency of the drug compatibility in the traditional prescription with the drug compatibility given by the method of the present invention. The symptoms and drug compatibility are listed in Table 6.

[0100] Table 6 Test recipe set

[0101]

[0102] The symptoms of the disease number 1 in Table 6 are sequentially entered into each drug decision tree, and the drug for treating the symptom is selected to form a new drug combination for treating the symptom. Similarly, the symptoms of the disease number 2 in Table 6 are selected using each drug decision tree to treat the disease. The Mongolian medicine combinations obtained are listed in Table 7.

[0103] Table 7 Mongolian medicine compatibility obtained by decision tree

[0104]

[0105] It can be seen from the two groups of drug compatibility in Tables 6 and 7 that under the symptoms of the disease No. 1, the drug compatibility obtained by the decision tree is based on the traditional formula with the addition of sandalwood; under the symptoms of the disease No. 2, the drug compatibility obtained by the decision tree is based on the traditional formula, but five drugs, namely, jujube, white cardamom, asafoetida, purple alum and black cloud osmanthus, are eliminated, and four drugs, namely, costus root, terminalia chebula, clove and sophora flavescens, are added.

[0106] For the same symptoms, there are slight differences between the traditional formula and the drug compatibility obtained by the present invention, so the following analysis is conducted:

[0107] First, the method of Example 1 was used to construct a decision tree for differential drugs, as follows: Figure 4 — Fig.13 .

[0108] Depend on Figure 4 The sandalwood decision tree shown in the figure shows that in the absence of main pulse Heyi and asthma, there are symptoms of heart pain, which can be treated with sandalwood. For the symptoms of the disease No. 1 in the test formula set, which meets the medication rules found by the sandalwood decision tree, therefore, adding sandalwood to the traditional Mongolian medicine combination in Table 6 can better treat heart pain.

[0109] For the symptoms of No. 2 in the test formula set, Figure 6 , Figure 8 , Fig.12 and Fig.13 It can be seen that costus root and terminalia chebula can be used to treat palpitations, sophora flavescens can be used to treat renal palpitations, and cloves can be used to treat renal palpitations and palpitations. Therefore, costus root, terminalia chebula, sophora flavescens and cloves in the drug combination obtained by the present invention have a certain therapeutic effect on the symptoms. Figure 5 , Figure 7 , Figure 7 , Fig.10 and Fig.11 It can be seen that Guangzao has not been used to treat cold-type Hei Yi and Xin He Yi; Bai Dou Kou has not been used to treat palpitations and Xin He Yi, Asafoetida has been used to treat insomnia and Xin He Yi at the same time, and Asafoetida has not been used when only Xin He Yi occurs; The effect of Heilongjiang Codonopsis pilosula on these symptoms is unknown; Ziafarina has not been used to treat palpitations, therefore, Guangzao, Bai Dou Kou, Asafoetida, Ziafarina and Heilongjiang Codonopsis pilosula were excluded from the drug combination.

[0110] In summary, it can be seen that it is reliable to use the present invention to select medicines for disease symptoms.

[0111] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A matching method between disease and medicine based on decision tree, It is characterized in that The following steps are involved: S1. Collect several traditional prescriptions, match the symptoms X corresponding to the prescriptions with the drugs Y used, and form a prescription sample set Z; S2, using part of the formulas in the formula sample set Z established in S1 as a training formula set, calculating the information entropy of the drugs in the training formula set and the conditional entropy of the drugs under the given condition of the disease symptoms according to the training formula set, and then calculating the information gain of the disease symptoms to the drugs by using the information entropy of the drugs and the conditional entropy of the drugs under the given condition of the disease symptoms; S3, constructing a drug decision tree corresponding to each drug according to the information gain of the disease symptoms on the drug calculated in S2; S4, taking the symptoms X' of the case to be prescribed as input parameters, and sequentially inputting them into the multiple drug decision trees obtained in S3, and obtaining a conclusion on whether the symptoms X' of the case matches the drug corresponding to the drug decision tree; S5. All drugs matching the symptoms X' are combined to obtain drug results Y=(y 1 ,y 2 ,…,y J ).

2. A method for matching symptoms and drugs based on a decision tree according to claim 1, It is characterized in that In S1, Recipe sample set Z = {(X 1 ,Y 1 ),(X 2 ,Y 2 ),…,(X n ,Y n ),…,(X N ,Y N )}, Where N is the total number of recipes in the recipe sample set; The symptom characteristic corresponding to the nth prescription is recorded as In vector is the symptom x of the i-th disease in the n-th recipe i ; There are two possible situations: when the ith symptom exists in the nth prescription, the drug combination in the nth prescription can treat the symptom. It is recorded as When the ith symptom does not exist in the nth prescription, that is, the drug combination in the nth prescription does not treat the symptom. It is recorded as The drug combination corresponding to the nth prescription is recorded as In vector It is the jth drug y in the nth recipe j ; There are two possible situations. When the jth drug is used in the nth recipe, it is recorded as When the jth drug is not used in the nth formulation, it is recorded as 3. A method for matching symptoms and drugs based on a decision tree according to claim 2, It is characterized in that In S2, The information gain of disease symptoms on drugs G(y j ,x i ) is calculated as shown in Formula 1: G(y j , x i ) = H(y j ) - E(y j | x i )(1) In formula 1, H(y j ) is the information entropy of the aforementioned drug, E(y j |x i ) is the conditional entropy of the drug under the given condition of the disease symptoms mentioned above. The disease symptom x can be calculated by formula 1. i For drugs j The information gain G(y j ,x i ).

4. A method for matching symptoms and drugs based on a decision tree according to claim 3, It is characterized in that In S2, The information entropy of the drug H(y j ) is calculated as shown in Formula 2: In formula 2, l = 1, 2 represents drug y j There are two cases: used and not used; N(y jl ) indicates that the drug is in y in the training formula set jl The total number of cases; N represents the total number of samples in the training formula set; Indicates that the drug is in y in the training formula set jl The probability of The conditional entropy E(y j | x i ) of the drug under the given conditions of the disease symptoms is calculated by Equation 3 as follows: In formula 3, t=1, 2 represents being in and not in the condition symptom x i Two cases; N(x it ) indicates that the symptoms of the disease are in x in the training formula set it The total number of times when Indicates that the symptoms of the disease are in x in the training formula set it The probability of the situation; N(x it ∩y jl ) indicates that the symptoms of the disease are in x in the training formula set it Condition and drug in y jl The total number of times when Indicates that in the training formula set, when the symptom is in x it Condition and drug in y jl The probability of the situation.

5. A method for matching symptoms and drugs based on a decision tree according to claim 2, It is characterized in that In S3, The method for constructing a drug decision tree corresponding to each drug is as follows: The symptoms of various diseases in traditional prescriptions are used as attributes to form the internal nodes of the decision tree; whether the drug is used is used as the classification result to form the leaf nodes.

6. A method for matching symptoms and drugs based on a decision tree according to claim 5, It is characterized in that In S3, When the selected drug is in y jl the disease symptom with the largest information gain in the traditional formula is used as the root node; Other disease symptoms in traditional formulas are used as internal nodes.

Images