Big data mining system based on traditional Chinese medicine diabetes database

Abstract

The invention discloses a big data mining system based on a traditional Chinese medicine diabetes database in the technical field of big data, and the system comprises an acquisition unit and a construction unit, and the construction unit is used for obtaining the age, gender, pathogenesis, illness, traditional Chinese medicine diagnosis and traditional Chinese medicine symptom type of a patient according to the features of the patient, and obtaining the drug type according to a prescription; constructing a matrix of different drug types and the age, gender, pathogenesis, disease symptoms, traditional Chinese medicine diagnosis and traditional Chinese medicine symptom types of the patient, thereby constructing the association between the different drug types and the age, gender, pathogenesis, disease symptoms, traditional Chinese medicine diagnosis and traditional Chinese medicine symptom types of the patient, and establishing a special diabetes disease knowledge graph; an input unit; and the mining unit is used for mining required treatment drugs from the special diabetes disease knowledge map according to the age, the gender, the pathogenesis, the diseases, the traditional Chinese medicine diagnosis and the traditional Chinese medicine symptom type so as to solve the problem that in the prior art, the drug combination in the prescription is difficult to flexibly adjust according to the physique and the symptom of the patient, so that the mined prescription is relatively weak in pertinence.

Description

Technical Field

[0001] The present invention belongs to the technical field of big data, and specifically is a big data mining system based on a traditional Chinese medicine diabetes database. Background Art

[0002] Diabetes is a chronic disease characterized by high blood sugar, caused by absolute or relative insulin deficiency and utilization disorder. my country began to study diabetes as early as the 2nd century AD, which made the diabetes diagnosis and treatment theory system and clinical experience recorded in ancient and modern Chinese medicine literature rich and complete.

[0003] However, the research on traditional Chinese medicine in my country is mainly based on statistical induction methods. This method is suitable for frequency statistics and correlation analysis of information such as symptoms and treatment methods in traditional Chinese medicine medical records. It has certain applicability in revealing characteristics and laws, but it is difficult to organize and optimize prescription data, which hinders the promotion and application of traditional Chinese medicine.

[0004] In the prior art, medical case data and corresponding prescription data are collected and sorted by constructing a diabetes literature information database, and a mining system is constructed. During use, after the patient's symptoms are input into the system, the system outputs the prescriptions of patients with similar symptoms in the database for the physician's reference and use. However, Chinese medicine focuses on syndrome differentiation and treatment. Different prescriptions need to be prescribed according to the constitution and symptoms of different patients during the prescription process. The drug combination in the prescription of this scheme is fixed, and it is difficult to make fine adjustments according to the patient's constitution and symptoms. The patient's symptoms are not well targeted, and the prescription effect is not good, which affects the application of Chinese medicine. Summary of the invention

[0005] The purpose of the present invention is to provide a big data mining system based on a traditional Chinese medicine diabetes database to solve the problem in the prior art that it is difficult to flexibly adjust the drug combination in a prescription according to the patient's constitution and symptoms, thereby making the mined prescription less targeted.

[0006] In order to achieve the above object, the technical solution of the present invention is as follows: A big data mining system based on a traditional Chinese medicine diabetes database, comprising: Collection unit: The collection unit is used to collect information about patients with a history of diabetes from the hospital database, and obtain the patient's characteristics, the corresponding prescriptions, the corresponding treatment effects, and the patient's description of his or her own condition. At the same time, the patients are divided into two groups, effective and ineffective, according to the treatment effects; Analysis unit: The analysis unit is used to process the patient's description of his or her own condition, obtain the patient's condition description information, and adjust the patient's characteristics according to the patient's condition description information; Construction unit: The construction unit is used to obtain the patient's age, gender, cause, symptom, TCM diagnosis and TCM symptom type according to the patient characteristics of the effective group, obtain the drug type according to the prescription, and construct a matrix of different drug types and the patient's age, gender, cause, symptom, TCM diagnosis and TCM symptom type, so as to construct the association between different drug types and the patient's age, gender, cause, symptom, TCM diagnosis and TCM symptom type, and establish a diabetes-specific knowledge graph; Input unit: The input unit is used to receive the basic population information and historical diagnosis and treatment information of the patient to be checked input by the user; Mining unit: The mining unit is used to obtain the age and gender of the patient to be checked based on basic population information, obtain the cause of disease, symptoms, TCM diagnosis and TCM symptom type of the patient to be checked based on historical diagnosis and treatment information, and mine the required therapeutic drugs from the diabetes disease knowledge map based on age, gender, cause of disease, symptoms, TCM diagnosis and TCM symptom type, and generate a prescription containing all the required drugs; Inspection unit: classify patients according to the patient characteristics of the invalid group, and obtain the invalid prescription corresponding to each category of patients, compare the generated prescriptions with the corresponding invalid prescriptions in turn, and after the same invalid prescription appears, when the patient characteristics of the patient to be checked and the patient characteristics corresponding to the invalid prescription are different, output the prescription, and when the patient characteristics of the patient to be checked and the patient characteristics corresponding to the invalid prescription are the same, do not output the prescription; Symptoms include continuous blood glucose data and clinical symptoms within a unit time; It also includes a dynamic monitoring unit, which is used to collect the user's blood sugar data and the user's physiological index data, generate verification information according to the user's physiological index data, and output the verification information to the user. When the user confirms that the verification information is consistent with his or her own feelings or actual behavior, it is determined whether the blood sugar information at this time can be used as a symptom according to the verification information, and only when the blood sugar information can be used as a symptom, the blood sugar data at this time is used as the symptom to proceed to the next step. If the verification information is inconsistent with his or her own feelings or actual behavior, a prompt message is output to the user; The acquisition unit can also be used to process patient information. The processing steps include: Step 1: Gather patient information into a data set, establish a data space based on the data set, calculate the density of data feature points and the temporal correlation of data features, and identify abnormal data in the data set; Step 2: Based on Calculate the time series length of missing data, where h represents the time series length of missing data, c represents the current dimension of the data, δ represents the temporal correlation of data features, and θ represents the density of data feature points; Step 3: According to Fill in the missing parts of the abnormal data, where J represents the missing content filled in by the abnormal data, and Y represents the time series length of the normal data with the same time series correlation as the abnormal data; Step 4: Perform identification calculations on the padded data again to determine whether it meets the preset non-abnormal data requirements. If not, fill in the abnormal data again until all data meet the preset non-abnormal data requirements. Import the padded data back into the data set to complete the data processing.

[0007] Furthermore, it also includes a verification unit, which is used to receive the expert guide information input by the user, and compare the output prescription with the expert guide information, determine whether the prescription has incompatibility taboos, and output the judgment result to the user.

[0008] Furthermore, the verification unit can also be used to replace the incompatible drugs according to the expert guide information after the prescription has incompatible drugs, and output the replaced prescription to the user.

[0009] Furthermore, it also includes a correction unit, which can also be used to receive expert guide information and feedback information input by the user, judge the effect of the prescription according to the feedback information, and correct the compatibility of the drugs in the prescription according to the judgment result and the expert guide information.

[0010] Furthermore, the acquisition unit can also be used to perform denoising on the data set collected from the patient information.

[0011] Furthermore, the correction unit can also be used to receive expert guide information and basic demographic information, and number the patients to be checked according to the basic demographic information. When the patient shows allergic symptoms in the feedback information, the patient and the drugs in the prescription are marked. When the basic demographic information of the numbered patient is input again, the prescription output this time is compared with the marked drug, and when the output prescription contains the marked drug, the marked drug is replaced according to the expert guide information.

[0012] Furthermore, the correction unit can also be used to update the expert guide information according to the feedback information.

[0013] Furthermore, the correction unit can also be used to update the diabetes-specific knowledge graph based on the output prescription and the corresponding feedback information, as well as the patient's basic demographic information and historical diagnosis and treatment information.

[0014] Furthermore, the correction unit can also be used to identify the drugs in the output prescription and delete the duplicate drugs in the prescription.

[0015] The technical principles and beneficial effects of the above scheme are as follows: This solution collects the characteristics of diabetic patients and the corresponding treatment prescriptions, establishes the association between different characteristics and different drugs, and thus establishes a diabetes-specific knowledge graph. During use, the user only needs to input the patient's basic demographic information and historical diagnosis and treatment information into the system, and the system can obtain the best combination of treatment drugs matched according to the patient's characteristics to form a prescription. This is convenient and fast, shortens the speed of medical staff to prescribe, greatly improves the diagnosis and treatment efficiency of medical staff, and promotes the application of traditional Chinese medicine.

[0016] Compared with the existing technology, this solution associates patient characteristics with drugs, so that this solution can adjust the prescription in a refined manner according to the individual differences of patients. This greatly enhances the flexibility of the combination of various drugs in the prescription, allowing this solution to provide more personalized treatment prescriptions and help improve the effectiveness of the prescriptions.

[0017] At the same time, this solution also monitors the patient's blood sugar in real time through a dynamic monitoring unit, and can also filter the blood sugar data according to the user's real-time status. Compared with the existing technology, this solution can effectively exclude blood sugar data such as when eating, and avoid erroneous data or data that does not reflect the patient's true blood sugar status from entering the subsequent diabetes-specific knowledge graph, affecting the accuracy of the subsequent mined prescriptions.

[0018] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a structural schematic diagram of an embodiment of a big data mining system based on a traditional Chinese medicine diabetes database of the present invention; Figure 2 It is a flow chart of an embodiment of a big data mining system based on a TCM diabetes database of the present invention; Figure 3 It is a flow chart of a verification unit of an embodiment of a big data mining system based on a TCM diabetes database of the present invention; Figure 4 This is a data processing flow chart of the acquisition unit of an embodiment of the big data mining system based on the traditional Chinese medicine diabetes database of the present invention.

[0020] The reference numerals in the drawings of the specification include: a collection unit 1, a construction unit 2, a mining unit 3, an input unit 4, a verification unit 5, an inspection unit 6, a correction unit 7, an analysis unit 8, and a dynamic monitoring unit 9. DETAILED DESCRIPTION

[0021] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0022] The following is further described in detail through specific implementation methods: Example 1 As attached Figure 1 -Attached Figure 4 Shown: A big data mining system based on a traditional Chinese medicine diabetes database, comprising: Collection unit 1: Identify and screen patient information in the hospital database, select patient information with a history of diabetes, and obtain patient characteristics, corresponding prescriptions for patients, corresponding treatment effects, and patient descriptions of their own conditions based on the patient information. At the same time, the patients are divided into effective and ineffective groups based on the treatment effects. Collection unit 1 can also be used to process patient information. The processing steps include: Step 1: Gather patient information into a data set, establish a data space based on the data set, and use

[0023] Calculate the density of data feature points, where θ represents the density of data feature points, i represents data, k represents the data dimension in space, a represents the distribution size of data in space, and calculate the temporal correlation of data features to identify abnormal data in the data set; Step 2: Based on Calculate the time series length of missing data, where h represents the time series length of missing data, c represents the current dimension of the data, and δ represents the temporal correlation of data features; Step 3: According to Fill in the missing parts of the abnormal data, where J represents the missing content filled in by the abnormal data, and Y represents the time series length of the normal data with the same time series correlation as the abnormal data; Step 4: Recognize and calculate the filled data again to determine whether it meets the preset non-abnormal data requirements. If not, fill in the abnormal data again until all the data meet the preset non-abnormal data requirements, import the filled data back into the data set, complete the data processing, and improve the data quality through data cleaning, thereby improving the accuracy of the subsequent diabetes disease knowledge graph constructed; at the same time, after the patient information is aggregated into a data set, the data set can be denoised to remove the noise in the data, further improve the data quality, and reduce the subsequent computing costs.

[0024] Analysis unit 8: Analysis unit 8 is used to process the patient's description of his own condition, by analyzing and understanding the patient's language, matching the patient's main complaint with the professional name, obtaining the patient's symptom description information, and adjusting the patient's characteristics according to the patient's symptom description information; For example: the patient complains about going to the toilet frequently, drinking water all the time, and getting hungry easily, etc. At this time, the above content is analyzed and the symptoms of the patient are "three mores and one less". The obtained patient disease description information is "three mores and one less". If this information does not appear in the patient characteristics, this part of the content will be added to the patient characteristics to ensure the comprehensiveness of the information obtained.

[0025] Construction unit 2: According to the characteristics of the effective group patients, the patient's age, gender, cause, symptoms, TCM diagnosis and TCM symptom type are obtained, where the symptoms include blood sugar data and clinical symptoms within a unit time, and the clinical symptoms include three more and one less, frequent urination, etc., and the drugs corresponding to the prescriptions used for treatment are obtained, and the correlation between various drugs in the prescriptions and the patient's age, gender, cause, symptoms, TCM diagnosis and TCM symptom type is calculated, and a matrix is ​​constructed based on the correlation, and the correlation between various drugs in the prescriptions and the patient's age, gender, cause, symptoms, TCM diagnosis and TCM symptom type is further judged based on the matrix results, and a diabetes-specific knowledge graph is constructed based on various drugs in the prescriptions, the patient's age, gender, cause, symptoms, TCM diagnosis and TCM symptom type and the correlation therein; For example: based on the characteristics of patient A, we know that he is 69 years old, male, obese, has three more and one less, has a pale yellow complexion, and a light red tongue. The corresponding prescription contains raw astragalus, yam, atractylodes, Scrophularia, angelica, red peony root, ligusticum chuanxiong and motherwort. We establish the association between several patient characteristics and the various drugs in the treatment prescription. For example, when the patient has a pale yellow complexion, the treatment prescription needs to include astragalus. Based on the patient characteristics, the drugs in the treatment prescription and the association between the two, we construct a diabetes-specific knowledge graph.

[0026] Input unit 4: accepts basic demographic information and historical medical information of the patient to be checked input by the user.

[0027] Mining unit 3: Obtain the age and gender of the patient to be investigated based on basic demographic information, obtain the cause of the patient's disease, symptoms, TCM diagnosis and TCM symptom type based on historical diagnosis and treatment information, and select the corresponding medicine through the diabetes disease knowledge library to form a prescription.

[0028] For example: input patient B, 50 years old, male, hereditary, prone to hunger due to frequent eating, yellow greasy tongue coating, slippery and stringy pulse, etc., and select the drugs that need to be included in the diabetes treatment prescription for the 50-year-old age group through the diabetes disease knowledge library, the drugs that need to be included in the treatment prescription for male diabetic patients, the drugs that need to be included in the treatment prescription for hereditary, prone to hunger due to frequent eating, yellow greasy tongue coating, slippery and stringy pulse, etc., and summarize all the above drugs to obtain the corresponding prescription.

[0029] Inspection unit 6: Classify patients according to the patient characteristics of the invalid group, and obtain the invalid prescription corresponding to each category of patients, compare the generated prescriptions with the corresponding invalid prescriptions in turn, and after the same invalid prescription appears, when the patient characteristics of the patient to be checked and the patient characteristics corresponding to the invalid prescription are different, output the prescription; when the patient characteristics of the patient to be checked and the patient characteristics corresponding to the invalid prescription are the same, do not output the prescription.

[0030] For example: in the invalid group, there is patient C, 50 years old, male, hereditary, blood sugar higher than normal value within 24 hours, yellow greasy tongue coating, and slippery pulse. All patients with the same characteristics as patient C are grouped together, and classified and integrated according to the prescriptions of all patients in the group. When the prescription mined out for patient B is different from the prescription in the group where patient C is located, the mined prescription is output; when the prescription mined out for patient B is the same as the prescription in the group where patient C is located, at this time, patient B and patient C have the same characteristics, so it can be seen that the prescription is an invalid prescription and is not output; if there is patient D, 35 years old, male, obese, eats a lot and gets hungry easily, has yellow greasy tongue coating, and slippery pulse, all patients with the same characteristics as patient D are grouped together, and classified and integrated according to the prescriptions of all patients in the group. When the prescription mined out for patient B is the same as that of patient D, at this time, since patient B and patient D have different characteristics, the mined prescription is output.

[0031] Verification unit 5: is used to receive expert guide information for input, including the compatibility relationship, incompatibility contraindications and replacement relationship of similar drugs of various drugs, and compare the output prescription with the drug incompatibility contraindications in the expert guide information one by one. When incompatibility contraindications appear in the prescription, the judgment result is output to the user to assist the user in adjusting the prescription. At the same time, the drugs with incompatibility contraindications can be replaced according to the replacement relationship of similar drugs, and the replaced prescription can be output to avoid users using prescriptions with incompatibility contraindications, causing damage to the patient's body, thereby improving the safety of the excavated prescription.

[0032] Correction unit 77: used to receive expert guide information and feedback information input by the user, that is, the patient's recovery status after using the output prescription. According to the recovery status, the prescription effect is divided into three levels: upper, medium and lower. When the prescription effect is lower, the drugs in the prescription are replaced according to the expert guide information or appropriate, synergistic drugs are added to the prescription to increase the efficacy. When the same prescription is output next time, the corrected prescription is output synchronously for user reference and selection, thereby reducing the waste of medical resources and the probability of patient damage caused by the repeated use of invalid prescriptions; the correction unit 7 can also be used to number the patient according to the name in the basic population information, and mark the patient and the allergic prescription when allergic symptoms appear in the feedback information. When the same drug of the allergic prescription appears in the patient's next prescription, the drug is replaced with the same drug according to the relationship between similar drugs in the expert guide information to avoid the repeated appearance of the allergic drug in the patient's prescription, thereby improving the effectiveness of the output prescription; the correction unit Element 7 can also be used to update and adjust the expert guide information based on feedback information, that is, after taking a prescription containing the same drugs, similar adverse reactions occur in multiple patients. At this time, the drugs may be incompatible. At this time, the compatibility of these drugs is updated to the expert guide information. When these drugs appear in the next output prescription, the verification unit 5 replaces these drugs and outputs the judgment result and the replaced prescription to the user to avoid the repeated appearance of incompatible drug combinations and cause damage to the bodies of subsequent patients; the correction unit 7 can also be used to update the diabetes specialty knowledge graph based on the prescription, feedback information, basic demographic information and historical diagnosis and treatment information, delete the association between the drugs and patient characteristics under multiple effects, and improve the effectiveness of the mined prescription; the correction unit 7 can also be used to identify the drugs in the prescription before outputting the prescription, remove the drugs that appear repeatedly in the prescription, and avoid excessive use of the drugs in the later medication process, causing adverse reactions in patients.

[0033] For example: after outputting prescription A for patient A, the effect of A is bad, then the correction unit 6 corrects it to B, and next time when facing patient B, when outputting prescription A, prescription B is output synchronously for user reference; at the same time, the patients can be numbered according to their names, etc., patient A is numbered 1, patient B is numbered 2, if patient B has an allergic reaction after taking prescription A, all the drugs in 2 and A are marked, and the next time patient B visits the doctor again, prescription C is output, and the drugs in prescription C and prescription A are compared, and all the drugs in prescription C that are the same as prescription A are replaced with the same drugs; when multiple patients are taking prescriptions containing a, b, c After taking prescription D that is combined with four drugs a, b, c and d, similar nausea symptoms appear. It is possible that there are incompatibility contraindications between the four drugs a, b, c and d. At this time, the expert guide information is updated, and the four drugs a, b, c and d are replaced with the same type when prescription D is output next time to improve the safety of the prescription; the feature of sallow complexion is associated with drug e, and the prescriptions of patients with sallow complexion in multiple features all contain drug e, but the feedback information of these corresponding prescriptions is the following, then the association between the feature of sallow complexion and drug e is deleted to reduce the impact of this set of associations on the effect of the final output prescription.

[0034] The dynamic monitoring unit 9 includes wearable devices such as a wearable blood glucose meter and a smart bracelet, which are used to continuously collect the user's blood glucose data and the user's physiological index data, wherein the user's physiological data mainly includes heart rate, blood oxygen saturation, body temperature, number of steps, activity duration, sleep quality, and sedentary time, etc., and also includes fall detection, etc. At the same time, the wearable blood glucose meter is used to collect the user's blood glucose data (such as collecting the user's blood glucose data every 3-5 minutes within 24 hours). When the user's ongoing activities are judged according to the user's physiological data, such as eating, exercising, going to the toilet, sleeping, or bending over, the dynamic monitoring unit outputs verification information to the user in real time through a receiver or a mobile phone (such as "Are you eating?" ), if the user believes that the verification information is the same as the actual behavior or his own feelings, the information is confirmed. At this time, the blood sugar information at this time is judged according to the verification information whether it can be used as a symptom. For example, the blood sugar information during meals often cannot reflect the patient's true blood sugar. In this case, this part of the blood sugar information will be deleted. If it is bending over or other exercises that have little effect on blood sugar, the blood sugar data at this time can be used as a symptom and the next step can be performed. If the user believes that the two are different, the information is denied. The blood sugar at this time cannot be used as a symptom, and a prompt message such as "Your data is abnormal, please check the collection device" is output to the user. At the same time, this part of the blood sugar data is deleted to avoid the negative impact of erroneous data on subsequent judgment and mining.

[0035] The above is only an embodiment of the present invention, and the common knowledge such as the known specific structure and / or characteristics in the scheme is not described in detail here. It should be pointed out that for those skilled in the art, several deformations and improvements can be made without departing from the structure of the present invention, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A big data mining system based on TCM diabetes database, characterized by: include: Collection unit: The collection unit is used to collect information about patients with a history of diabetes from the hospital database, and obtain the patient's characteristics, the corresponding prescriptions, the corresponding treatment effects, and the patient's description of his or her own condition. At the same time, the patients are divided into two groups, effective and ineffective, according to the treatment effects; Analysis unit: The analysis unit is used to process the patient's description of his or her own condition, obtain the patient's condition description information, and adjust the patient's characteristics according to the patient's condition description information; Construction unit: The construction unit is used to obtain the patient's age, gender, cause, symptom, TCM diagnosis and TCM symptom type according to the patient characteristics of the effective group, obtain the drug type according to the prescription, and construct a matrix of different drug types and the patient's age, gender, cause, symptom, TCM diagnosis and TCM symptom type, so as to construct the association between different drug types and the patient's age, gender, cause, symptom, TCM diagnosis and TCM symptom type, and establish a diabetes-specific knowledge graph; Input unit: The input unit is used to receive the basic population information and historical diagnosis and treatment information of the patient to be checked input by the user; Mining unit: The mining unit is used to obtain the age and gender of the patient to be checked based on basic population information, obtain the cause of disease, symptoms, TCM diagnosis and TCM symptom type of the patient to be checked based on historical diagnosis and treatment information, and mine the required therapeutic drugs from the diabetes disease knowledge map based on age, gender, cause of disease, symptoms, TCM diagnosis and TCM symptom type, and generate a prescription containing all the required drugs; Inspection unit: classify patients according to the patient characteristics of the invalid group, and obtain the invalid prescription corresponding to each category of patients, compare the generated prescriptions with the corresponding invalid prescriptions in turn, and after the same invalid prescription appears, when the patient characteristics of the patient to be checked and the patient characteristics corresponding to the invalid prescription are different, output the prescription, and when the patient characteristics of the patient to be checked and the patient characteristics corresponding to the invalid prescription are the same, do not output the prescription; Symptoms include blood sugar data and clinical symptoms per unit time; It also includes a dynamic monitoring unit, which is used to collect the user's blood sugar data and the user's physiological index data, generate verification information according to the user's physiological index data, and output the verification information to the user. When the user confirms that the verification information is consistent with his or her own feelings or actual behavior, it is determined whether the blood sugar information at this time can be used as a symptom according to the verification information, and only when the blood sugar information can be used as a symptom, the blood sugar data at this time is used as the symptom to proceed to the next step. If the verification information is inconsistent with his or her own feelings or actual behavior, a prompt message is output to the user; The acquisition unit can also be used to process patient information. The processing steps include: Step 1: Gather patient information into a data set, establish a data space based on the data set, calculate the density of data feature points and the temporal correlation of data features, and identify abnormal data in the data set; Step 2: Based on Calculate the time series length of missing data, where h represents the time series length of missing data, c represents the current dimension of the data, δ represents the temporal correlation of data features, and θ represents the density of data feature points; Step 3: According to Fill in the missing parts of the abnormal data, where J represents the missing content filled in by the abnormal data, and Y represents the time series length of the normal data with the same time series correlation as the abnormal data; Step 4: Perform identification calculations on the padded data again to determine whether it meets the preset non-abnormal data requirements. If not, fill in the abnormal data again until all data meet the preset non-abnormal data requirements. Import the padded data back into the data set to complete the data processing.

2. The big data mining system based on the TCM diabetes database according to claim 1 is characterized by: It also includes a verification unit, which is used to receive the expert guide information input by the user, compare the output prescription with the expert guide information, determine whether the prescription has incompatibility taboos, and output the judgment result to the user.

3. The big data mining system based on the TCM diabetes database according to claim 2 is characterized by: The verification unit can also be used to replace the incompatible drugs according to the expert guide information after the prescription has incompatible drugs, and output the replaced prescription to the user.

4. The big data mining system based on the TCM diabetes database according to claim 3 is characterized by: It also includes a correction unit, which can be used to receive expert guide information and feedback information input by users, judge the effect of the prescription according to the feedback information, and correct the compatibility of drugs in the prescription according to the judgment result and the expert guide information.

5. The big data mining system based on the TCM diabetes database according to claim 4 is characterized by: The acquisition unit can also be used to perform denoising on the data set collected from the patient information.

6. The big data mining system based on the TCM diabetes database according to claim 5 is characterized by: The correction unit can also be used to receive expert guide information and basic demographic information, and number the patients to be checked according to the basic demographic information. When the patient shows allergic symptoms in the feedback information, the patient and the drugs in the prescription are marked. When the basic demographic information of the numbered patient is input again, the prescription output this time is compared with the marked drug, and when the output prescription contains the marked drug, the marked drug is replaced according to the expert guide information.

7. The big data mining system based on the TCM diabetes database according to claim 6 is characterized by: The correction unit can also be used to update the expert guide information according to the feedback information.

8. The big data mining system based on the TCM diabetes database according to claim 7 is characterized by: The correction unit can also be used to update the diabetes-specific knowledge graph based on the output prescription and corresponding feedback information, as well as the patient's basic demographic information and historical diagnosis and treatment information.

9. The big data mining system based on the TCM diabetes database according to claim 8, characterized in that: The correction unit can also be used to identify the drugs in the output prescription and delete the duplicate drugs in the prescription.

Images