A drug vigilance system access system and method based on artificial intelligence
By introducing artificial intelligence and logic verification modules into the pharmacovigilance system, the problem that existing systems cannot accurately generate pharmacovigilance reports is solved, intelligent management and priority judgment of pharmacovigilance information are realized, and the accuracy of reports and users are improved. Experience.
Patent Information
- Application Number
- CN202210251462.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-03-15
- Publication Date
- 2024-12-31
- Estimated Expiration
- 2042-03-15
AI Technical Summary
The existing pharmacovigilance system access system cannot accurately generate pharmacovigilance reports based on the types of drugs that users need to access, and lacks intelligent management and processing of pharmacovigilance information.
A pharmacovigilance system access system based on artificial intelligence is designed, including drug query module, patient-related information acquisition module, verification verification module and pharmacovigilance report generation module. The pharmacovigilance report is obtained and generated through logical verification and manual verification, and according to Priority is determined by the report content.
It realizes effective management of pharmacovigilance reports related to accessing drug types, improves the accuracy and user experience of the report, and ensures the accuracy and orderliness of information through the collaborative work of artificial intelligence and logic verification modules.
Smart Images
Figure CN114639458B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of computer technology, and in particular to an artificial intelligence-based drug vigilance system access system and method. Background Art
[0002] With the rapid development of computer technology, people are using it more and more widely, especially the promotion of artificial intelligence, which makes the systems using computer technology more intelligent, thus bringing great convenience to people; in the application of artificial intelligence technology in the field of drug vigilance, people are more organized in the preservation and processing of drug vigilance information, and the processing method is more convenient, which saves a lot of manpower and material resources to a certain extent.
[0003] However, the current pharmacovigilance system still has major defects, especially the pharmacovigilance system access system, which simply manages and controls the user's account and cannot generate corresponding pharmacovigilance reports more accurately based on the types of drugs that the user needs to access.
[0004] In view of the above situation, we need an artificial intelligence-based drug vigilance system access system and method. Summary of the invention
[0005] The purpose of the present invention is to provide a drug vigilance system access system and method based on artificial intelligence to solve the problems raised in the above background technology.
[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a drug vigilance system access system based on artificial intelligence, comprising:
[0007] A drug query module, wherein the drug query module is used to query the types of drugs to be accessed;
[0008] A pharmacovigilance information acquisition module, which includes a patient-related information acquisition module, a verification module, and a pharmacovigilance report generation module.
[0009] The patient-related information acquisition module is used to acquire various related information of the trial patients of the drug type to be accessed in the drug query module;
[0010] The verification module is used to perform logic verification on various pieces of associated information of the trial patients of the drug type to be accessed obtained in the patient associated information acquisition module, determine whether there is suspicious information in the associated information, and perform manual verification on the suspicious information;
[0011] The pharmacovigilance report generation module is used to summarize the various related information of the trial patients of the drug type to be accessed that have passed the verification and verification in the verification and verification module, obtain the pharmacovigilance report corresponding to each trial patient, and determine the priority of the pharmacovigilance report based on the content of the pharmacovigilance report.
[0012] The present invention realizes the acquisition of drug vigilance reports related to the drug types to be visited through the collaborative cooperation of various modules. This process involves the verification and verification of drug vigilance report information of trial patients, the generation of drug vigilance reports and the priority judgment of drug vigilance reports, thereby realizing the effective management of drug vigilance reports related to the drug types to be visited.
[0013] Furthermore, the patient-related information acquisition module acquires various related information of the trial patients of the drug type to be accessed in the drug query module, including clinical drug adverse reaction reports, patient historical medical records and patient daily behavior.
[0014] Furthermore, the verification module includes a logic verification module and a manual verification module.
[0015] The logic verification module is used to obtain the clinical adverse drug reaction feature set, historical medical record feature set and daily behavior feature set corresponding to each trial drug patient, and perform a logical analysis on the clinical adverse drug reaction feature set corresponding to each trial drug patient and the historical medical record feature set and daily behavior feature set corresponding to the trial drug patient, to determine whether the clinical adverse drug reaction feature set corresponding to the trial drug patient and the corresponding historical medical record feature set and daily behavior feature set conform to a logical relationship.
[0016] If the corresponding clinical adverse drug reaction feature set of the trial drug patient and the corresponding historical medical record feature set and daily behavior feature set have a logical relationship, then the associated information of the trial drug patient is judged to be passed, and there is no suspicious information in the associated information. The associated information of the trial drug patient who has passed the test is added to the verification database.
[0017] If there is no logical relationship between the clinical adverse drug reaction feature set of the trial patient and the corresponding historical medical record feature set and daily behavior feature set, it is determined that the associated information of the trial patient has failed and there is suspicious information in the associated information;
[0018] The manual verification module submits the various related information corresponding to the trial patients whose related information in the logic verification module is suspicious to the management personnel for verification. The management personnel add the various related information corresponding to the trial patients that have passed the verification to the verification database, and do not process the various related information corresponding to the trial patients that have failed the verification.
[0019] The verification and checking module of the present invention includes a logic verification module and a manual verification module. The logic verification module realizes the extraction of features in various related information of trial drug patients, and analyzes the logical relationship between each feature according to the set corresponding to each extracted feature, thereby making the analysis of various information of trial drug patients more intelligent, and combining the analysis with the big data in the database will make the final analysis result more accurate; for suspicious information in the related information, a manual verification module is set up, considering that when the database-based artificial intelligence analyzes the related information, if the database does not contain relevant data, it may cause the logic verification module to analyze the suspicious information in the related information. The database cannot provide corresponding data basis. This situation may be caused by the lack of content of this part of the data, rather than just due to logical errors in the data. Therefore, setting up a manual verification module is equivalent to a secondary verification of the suspicious data, which can effectively improve the accuracy of the analysis results of the logic verification module on various related information of trial drug patients.
[0020] Furthermore, the method for the logic verification module to obtain the clinical adverse drug reaction feature set, historical medical record feature set and daily behavior feature set corresponding to each drug trial patient includes the following steps:
[0021] S1.1. Obtain the corresponding clinical adverse drug reaction feature set for each trial patient, wherein the clinical adverse drug reaction features include the time, type and adverse manifestation of the adverse reaction.
[0022] When the trial drug patient has no adverse reaction, the adverse reaction feature set corresponding to the trial drug patient is empty;
[0023] S1.2. Obtain the corresponding historical medical record feature set for each drug trial patient, wherein the historical medical record features include the type of historical medical record and the various contraindications corresponding to each historical medical record.
[0024] When a trial drug patient does not have a historical medical record, the historical medical record feature set corresponding to the trial drug patient is empty;
[0025] S1.3. Obtain a set of daily behavioral characteristics corresponding to each drug trial patient, wherein the daily behavioral characteristics include the corresponding food types in the daily diet of the drug trial patient.
[0026] When the logic verification module of the present invention obtains the corresponding clinical drug adverse reaction feature set, historical medical record feature set and daily behavior feature set of each trial patient, and when obtaining the various contents of the clinical drug adverse reaction characteristics, historical medical record characteristics and daily behavior characteristics, it obtains each feature information by extracting keywords.
[0027] Furthermore, the method in which the logic verification module performs logic analysis on the clinical adverse drug reaction feature set corresponding to each trial drug patient and the historical medical record feature set and daily behavior feature set corresponding to the trial drug patient comprises the following steps:
[0028] S2.1. Obtain the clinical adverse drug reaction feature set, historical medical record feature set and daily behavior feature set corresponding to each trial drug patient to form a logical verification data set corresponding to the trial drug patient. The logical verification set corresponding to the nth trial drug patient is recorded as An, wherein An = {Bn, Cn, Dn}.
[0029] Among them, Bn represents the clinical adverse drug reaction feature set corresponding to the nth trial patient.
[0030] Cn represents the historical medical record feature set corresponding to the nth drug trial patient,
[0031] Dn represents the daily behavior feature set corresponding to the nth drug trial patient;
[0032] S2.2, obtain the drug ingredients in the drug category to be accessed, and compare the obtained results with each contraindicated drug corresponding to each historical medical record in Cn corresponding to the nth drug trial patient, and match the comparison results with Bn to obtain the first logical coefficient Q1n corresponding to the nth drug trial patient;
[0033] S2.3, obtain the food types in the daily diet of the drug trial patient in Dn corresponding to the nth drug trial patient, determine whether the obtained food types can produce adverse reactions with the drug components in the drug types to be visited, and count the food types in Dn that can produce adverse reactions with the drug components in the drug types to be visited, and the adverse reactions produced are the same as the adverse manifestations in Bn, compare with the database, obtain the adverse reaction coefficient corresponding to each food type, accumulate and sum the adverse reaction coefficients corresponding to the statistical food types, and then obtain the second logical coefficient Q2n;
[0034] S2.4, obtaining the logical value Qn of the nth trial patient, wherein Qn=r1*Q1n+r2*Q2n, wherein r1 represents the first weighted value, and r2 represents the second weighted value;
[0035] S2.5, compare Qn with the first preset value,
[0036] When Qn is greater than or equal to the first preset value, it is determined that the associated information of the trial patient passes.
[0037] When Qn is less than the first preset value, it is determined that the associated information of the trial patient has failed;
[0038] The method for obtaining r2 comprises the following steps:
[0039] S2.4.1. Obtain the food types in the daily diet of the trial patient in Dn corresponding to the nth trial patient in S2.3;
[0040] S2.4.2. Obtain the food types in Dn that can produce adverse reactions with the drug ingredients in the drug type to be accessed, and the adverse reactions produced are different from the adverse manifestations in Bn, and enter the corresponding adverse reaction coefficients of the obtained food types in the database into a blank set one by one, and obtain the set Among them, m represents the set The number of elements in the set The value of the adverse reaction coefficient corresponding to the yth element in is recorded as
[0041] S2.4.3, obtain the second weighted value r2,
[0042] when When
[0043] when , the r2=0.
[0044] When the logic verification module of the present invention performs a logical analysis on the clinical drug adverse reaction feature set corresponding to each trial drug patient and the historical medical record feature set and daily behavior feature set corresponding to the trial drug patient, the logic verification set corresponding to the nth trial drug patient is recorded as An, in order to summarize the various feature information of the trial drug patient to avoid the mismatch between the analyzed feature information and the trial drug patient during the logical analysis, thereby causing deviations in the analysis results; the first logic coefficient and the second logic coefficient corresponding to the trial drug patient are obtained respectively, because when the associated information of the trial drug patient is logically verified, the features in the historical medical record feature set and the daily behavior feature set will produce adverse reactions when reacting with the drug type to be accessed, so it is necessary to analyze from two aspects, namely corresponding to two analysis results (the first logic coefficient and the second logic coefficient); after determining the food types in Dn that can produce adverse reactions with the drug components in the drug type to be accessed, the adverse reactions produced are also compared with the adverse manifestations in Bn, in order to determine whether the adverse reactions produced by the trial drug patient itself are caused by the food types corresponding to the adverse reactions. If not, it means that the food The type of food has no adverse effect on the trial patient, and thus there is no logical relationship between the food type and the adverse reaction of the trial patient, and thus there is no need to conduct statistics on it; by comparing the database, the adverse reaction coefficient corresponding to each food type is obtained, and the adverse reaction coefficients corresponding to the statistical food types are accumulated and summed up, because different food types have different degrees of reaction with the components of the drug type to be accessed, and the degree of adverse reaction caused to the patient is also different. Therefore, by setting the adverse reaction coefficient, the patient's adverse reaction situation can be quantified and concretized, which is convenient for the subsequent logical verification of the related information of the trial patient, and also convenient for the subsequent priority division of the drug alert report corresponding to the trial patient; when obtaining the logical value of the trial patient, a weighted method is adopted, and r1 and r2 are set, because the first logical coefficient and the second logical coefficient of the trial patient are obtained by different standards, and both reflect the adverse reaction of the trial patient. Therefore, the weighted method can effectively combine the acquisition results corresponding to the two standards, that is, to achieve the unification of the two acquisition standards, so that the final analysis result is more accurate.
[0045] Furthermore, the method of comparing the obtained results in S2.2 with each contraindicated drug corresponding to each historical medical record in Cn corresponding to the nth trial patient includes the following steps:
[0046] S2.2.1. List the contraindicated drugs in Cn that have the same drug ingredients as the drug category to be visited, and form a data pair [p1, p2] with the contraindicated drugs and the drug ingredients in the drug category to be visited that are the same as the contraindicated drugs.
[0047] Among them, p1 represents a contraindicated drug, and p2 represents a drug ingredient in the drug category to be accessed that is the same as the contraindicated drug;
[0048] S2.2.2. Obtain the historical medical record type corresponding to p1 in each data pair, compare with the database, and match the final contraindication value corresponding to p1.
[0049] When the historical medical record type corresponding to p1 is one, compare the database to match the taboo value corresponding to p1 in the historical medical record. This taboo value is the final taboo value corresponding to p1.
[0050] When there are multiple types of historical medical records corresponding to p1, the database is compared to match the taboo values corresponding to p1 in each type of historical medical record, and the maximum value of the obtained taboo values is the final taboo value of p1.
[0051] When the present invention compares the obtained results with each contraindication drug corresponding to each historical medical record in Cn corresponding to the nth trial drug patient, the contraindication drug and the drug component in the drug type to be accessed that is the same as the contraindication drug constitute a data pair. This is because each historical medical record of the trial drug patient may correspond to multiple contraindications, and the trial drug patient may contain multiple historical medical records. Different drug components of the drug type to be accessed may be the same as different contraindications, thereby producing different adverse reactions. Therefore, the use of data pairs can accurately lock the drug component of the drug type to be accessed that produces adverse reactions. and the corresponding contraindications, making the analysis more accurate and detailed; when there are multiple types of historical medical records corresponding to p1, the database is compared to match the contraindication values corresponding to p1 in each historical medical record, and the maximum value of the contraindication values obtained is the final contraindication value of p1. This is because there are multiple historical medical records for trial patients, each of which corresponds to multiple contraindications, and when the same contraindication drug corresponding to different historical medical records has the same drug ingredient as the drug type to be accessed, since the contraindication values corresponding to the contraindication drug in different historical medical records are different, in this case, the data pair corresponds to multiple contraindication values at the same time.
[0052] Furthermore, the method for obtaining the first logical coefficient Q1n corresponding to the nth drug trial patient in S2.2 comprises the following steps:
[0053] S2.2-1. Obtain each data pair in S2.2.1;
[0054] S2.2-2. The adverse performance caused by the drug ingredients in the drug category to be visited in each data pair that are the same as the contraindicated drugs is recorded as e, and e is compared with Bn.
[0055] When e is included in Bn, the corresponding final taboo value in the data pair corresponding to e is obtained.
[0056] When e is not included in Bn, the corresponding final taboo value in the data pair corresponding to e is not processed;
[0057] S2.2-3. Calculate the sum of all final contraindication values obtained in S2.2-2 to obtain the first logical coefficient Q1n corresponding to the nth trial patient.
[0058] When all final contraindication values obtained in S2.2-2 are empty, the first logical coefficient Q1n corresponding to the nth trial drug patient is equal to 0;
[0059] The method for obtaining r1 comprises the following steps:
[0060] S2.4-1. Obtain all data pairs corresponding to the nth trial patient in S2.2.1;
[0061] S2.4-2, screen the data pairs whose corresponding bad performance e in S2.4-1 is not included in Bn, and enter the final taboo values corresponding to each screened data pair into a blank set one by one, and obtain Among them, m1 represents the set The number of elements in, n represents the nth trial patient, the set The final taboo value corresponding to the y1th element in is recorded as
[0062] S2.4-3, obtain the first weighted value r1,
[0063] when When
[0064] when , the r1=0.
[0065] When the present invention obtains the first logical coefficient Q1n corresponding to the nth trial patient, it is necessary to screen the data pairs (by comparing the adverse manifestations corresponding to the data pairs with Bn) to obtain the corresponding first logical coefficient; when all the final contraindication values obtained in S2.2-2 are empty, it means that the trial patient has no adverse reactions or the adverse reactions produced do not match the adverse manifestations corresponding to the data pairs, and then it is determined that the first logical coefficient Q1n corresponding to the nth trial patient is equal to 0.
[0066] Furthermore, the method for the pharmacovigilance report generation module to obtain the pharmacovigilance report corresponding to each drug trial patient includes the following steps:
[0067] S3.1, obtain the data pairs corresponding to each e contained in Bn in S2.2-2;
[0068] S3.2, extract the contraindicated drugs in each data pair in S3.1 and enter them one by one into the suspected drug section of the pharmacovigilance report corresponding to the nth trial patient;
[0069] S3.3. Extract the adverse manifestations corresponding to each data pair in S3.1 respectively, and enter them one by one into the adverse manifestation part of the pharmacovigilance report corresponding to the nth trial patient;
[0070] S3.4. Obtain the personal information of the nth trial patient and enter it into the personal information section of the pharmacovigilance report corresponding to the nth trial patient.
[0071] The pharmacovigilance report corresponding to the trial patient includes a suspicious drug section, an adverse reaction section and a personal information section.
[0072] When the pharmacovigilance report generating module of the present invention fails to obtain the corresponding data pair, the pharmacovigilance report corresponding to the drug trial patient includes a suspicious drug part and an adverse performance part which are empty.
[0073] Furthermore, when the pharmacovigilance report generation module determines the priority of the pharmacovigilance report based on the content of the pharmacovigilance report,
[0074] Sort the logical values of the drug trial patients from large to small. The drug trial patients with higher serial numbers have higher priority for the pharmacovigilance reports.
[0075] When the logical values of the drug trial patients are the same, the number of data pairs corresponding to the drug trial patients is further compared, and the drug trial patients with more corresponding data pairs have a higher priority for the corresponding pharmacovigilance report;
[0076] If the numbers of data pairs corresponding to the trial patients are the same, the priorities of the drug vigilance reports corresponding to the trial patients with the same numbers of data pairs will be randomly sorted.
[0077] The pharmacovigilance report generation module of the present invention determines the priority of the pharmacovigilance report so that when a user accesses a drug type, the displayed pharmacovigilance reports are orderly, and the impact caused by the corresponding adverse reactions is from high to low, so that the user can quickly obtain the main information of the pharmacovigilance report related to the drug type to be accessed.
[0078] A method for accessing a pharmacovigilance system based on artificial intelligence, the method comprising the following steps:
[0079] S1. Query the type of medicine to be accessed through the medicine query module;
[0080] S2, obtaining various related information of the trial patients of the drug type to be accessed in the drug query module through the patient related information acquisition module;
[0081] S3. In the verification module, logic verification is performed on various related information of the trial patients of the drug type to be accessed obtained in the patient related information acquisition module to determine whether there is suspicious information in the related information, and manual verification is performed on the suspicious information;
[0082] S4. The pharmacovigilance report generation module aggregates the related information of the trial patients of the drug type to be accessed that has passed the verification and check in the verification and check module, obtains the pharmacovigilance report corresponding to each trial patient, and determines the priority of the pharmacovigilance report based on the content of the pharmacovigilance report.
[0083] Compared with the prior art, the beneficial effects achieved by the present invention are: when accessing the drug vigilance system, the present invention can not only automatically obtain various related information of drug trial patients according to artificial intelligence, but also automatically implement logical verification and manual verification of the obtained related information, and automatically generate corresponding drug vigilance reports while ensuring that the related information is correct, and calculate the priority of each drug vigilance report according to the corresponding feature sets in the process of generating the drug vigilance report, showing effective management of drug vigilance information. BRIEF DESCRIPTION OF THE DRAWINGS
[0084] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:
[0085] Figure 1 It is a structural schematic diagram of an artificial intelligence-based pharmacovigilance system access system of the present invention;
[0086] Figure 2 It is a flow chart of a method for a logic verification module in an access system of a drug vigilance system based on artificial intelligence of the present invention to perform a logical analysis on a clinical drug adverse reaction feature set corresponding to each drug trial patient and a historical medical record feature set and a daily behavior feature set corresponding to the drug trial patient;
[0087] Figure 3 It is a flow chart of a method for obtaining a pharmacovigilance report corresponding to each drug trial patient by a pharmacovigilance report generation module in a pharmacovigilance system access system based on artificial intelligence according to the present invention;
[0088] Figure 4 It is a flow chart of an artificial intelligence-based drug vigilance system access method of the present invention. DETAILED DESCRIPTION
[0089] 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.
[0090] See also Figure 1-Figure 4 The present invention provides a technical solution: an artificial intelligence-based pharmacovigilance system access system, comprising:
[0091] A drug query module, wherein the drug query module is used to query the types of drugs to be accessed;
[0092] A pharmacovigilance information acquisition module, which includes a patient-related information acquisition module, a verification module, and a pharmacovigilance report generation module.
[0093] The patient-related information acquisition module is used to acquire various related information of the trial patients of the drug type to be accessed in the drug query module;
[0094] The verification module is used to perform logic verification on various pieces of associated information of the trial patients of the drug type to be accessed obtained in the patient associated information acquisition module, determine whether there is suspicious information in the associated information, and perform manual verification on the suspicious information;
[0095] The pharmacovigilance report generation module is used to summarize the various related information of the trial patients of the drug type to be accessed that have passed the verification and verification in the verification and verification module, obtain the pharmacovigilance report corresponding to each trial patient, and determine the priority of the pharmacovigilance report based on the content of the pharmacovigilance report.
[0096] The present invention realizes the acquisition of drug vigilance reports related to the drug types to be visited through the collaborative cooperation of various modules. This process involves the verification and verification of drug vigilance report information of trial patients, the generation of drug vigilance reports and the priority judgment of drug vigilance reports, thereby realizing the effective management of drug vigilance reports related to the drug types to be visited.
[0097] The patient-related information acquisition module acquires various related information of the trial patients of the drug type to be accessed in the drug query module, including clinical drug adverse reaction reports, patient historical medical records and patient daily behavior.
[0098] The verification and checking module includes a logic verification module and a manual verification module.
[0099] The logic verification module is used to obtain the clinical adverse drug reaction feature set, historical medical record feature set and daily behavior feature set corresponding to each trial drug patient, and perform a logical analysis on the clinical adverse drug reaction feature set corresponding to each trial drug patient and the historical medical record feature set and daily behavior feature set corresponding to the trial drug patient, to determine whether the clinical adverse drug reaction feature set corresponding to the trial drug patient and the corresponding historical medical record feature set and daily behavior feature set conform to a logical relationship.
[0100] If the corresponding clinical adverse drug reaction feature set of the trial drug patient and the corresponding historical medical record feature set and daily behavior feature set have a logical relationship, then the associated information of the trial drug patient is judged to be passed, and there is no suspicious information in the associated information. The associated information of the trial drug patient who has passed the test is added to the verification database.
[0101] If there is no logical relationship between the clinical adverse drug reaction feature set of the trial patient and the corresponding historical medical record feature set and daily behavior feature set, it is determined that the associated information of the trial patient has failed and there is suspicious information in the associated information;
[0102] The manual verification module submits the various related information corresponding to the trial patients whose related information in the logic verification module is suspicious to the management personnel for verification. The management personnel add the various related information corresponding to the trial patients that have passed the verification to the verification database, and do not process the various related information corresponding to the trial patients that have failed the verification.
[0103] The verification and checking module of the present invention includes a logic verification module and a manual verification module. The logic verification module realizes the extraction of features in various related information of trial drug patients, and analyzes the logical relationship between each feature according to the set corresponding to each extracted feature, thereby making the analysis of various information of trial drug patients more intelligent, and combining the analysis with the big data in the database will make the final analysis result more accurate; for suspicious information in the related information, a manual verification module is set up, considering that when the database-based artificial intelligence analyzes the related information, if the database does not contain relevant data, it may cause the logic verification module to analyze the suspicious information in the related information. The database cannot provide corresponding data basis. This situation may be caused by the lack of content of this part of the data, rather than just due to logical errors in the data. Therefore, setting up a manual verification module is equivalent to a secondary verification of the suspicious data, which can effectively improve the accuracy of the analysis results of the logic verification module on various related information of trial drug patients.
[0104] The method for the logic verification module to obtain the clinical adverse drug reaction feature set, historical medical record feature set and daily behavior feature set corresponding to each drug trial patient includes the following steps:
[0105] S1.1. Obtain the corresponding clinical adverse drug reaction feature set for each trial patient, wherein the clinical adverse drug reaction features include the time, type and adverse manifestation of the adverse reaction.
[0106] When the trial drug patient has no adverse reaction, the adverse reaction feature set corresponding to the trial drug patient is empty;
[0107] S1.2. Obtain the corresponding historical medical record feature set for each drug trial patient, wherein the historical medical record features include the type of historical medical record and the various contraindications corresponding to each historical medical record.
[0108] When a trial drug patient does not have a historical medical record, the historical medical record feature set corresponding to the trial drug patient is empty;
[0109] S1.3. Obtain a set of daily behavioral characteristics corresponding to each drug trial patient, wherein the daily behavioral characteristics include the corresponding food types in the daily diet of the drug trial patient.
[0110] When the logic verification module of the present invention obtains the corresponding clinical drug adverse reaction feature set, historical medical record feature set and daily behavior feature set of each trial patient, and when obtaining the various contents of the clinical drug adverse reaction characteristics, historical medical record characteristics and daily behavior characteristics, it obtains each feature information by extracting keywords.
[0111] The method for the logic verification module to perform logic analysis on the clinical adverse drug reaction feature set corresponding to each trial drug patient and the historical medical record feature set and daily behavior feature set corresponding to the trial drug patient comprises the following steps:
[0112] S2.1. Obtain the clinical adverse drug reaction feature set, historical medical record feature set and daily behavior feature set corresponding to each trial drug patient to form a logical verification data set corresponding to the trial drug patient. The logical verification set corresponding to the nth trial drug patient is recorded as An, wherein An = {Bn, Cn, Dn}.
[0113] Among them, Bn represents the clinical adverse drug reaction feature set corresponding to the nth trial patient.
[0114] Cn represents the historical medical record feature set corresponding to the nth drug trial patient,
[0115] Dn represents the daily behavior feature set corresponding to the nth drug trial patient;
[0116] S2.2, obtain the drug ingredients in the drug category to be accessed, and compare the obtained results with each contraindicated drug corresponding to each historical medical record in Cn corresponding to the nth drug trial patient, and match the comparison results with Bn to obtain the first logical coefficient Q1n corresponding to the nth drug trial patient;
[0117] S2.3, obtain the food types in the daily diet of the drug trial patient in Dn corresponding to the nth drug trial patient, determine whether the obtained food types can produce adverse reactions with the drug components in the drug types to be visited, and count the food types in Dn that can produce adverse reactions with the drug components in the drug types to be visited, and the adverse reactions produced are the same as the adverse manifestations in Bn, compare with the database, obtain the adverse reaction coefficient corresponding to each food type, accumulate and sum the adverse reaction coefficients corresponding to the statistical food types, and then obtain the second logical coefficient Q2n;
[0118] S2.4, obtaining the logical value Qn of the nth trial patient, wherein Qn=r1*Q1n+r2*Q2n, wherein r1 represents the first weighted value, and r2 represents the second weighted value;
[0119] S2.5, compare Qn with the first preset value,
[0120] When Qn is greater than or equal to the first preset value, it is determined that the associated information of the trial patient passes.
[0121] When Qn is less than the first preset value, it is determined that the associated information of the trial patient has failed;
[0122] The method for obtaining r2 comprises the following steps:
[0123] S2.4.1. Obtain the food types in the daily diet of the trial patient in Dn corresponding to the nth trial patient in S2.3;
[0124] S2.4.2. Obtain the food types in Dn that can produce adverse reactions with the drug ingredients in the drug type to be accessed, and the adverse reactions produced are different from the adverse manifestations in Bn, and enter the corresponding adverse reaction coefficients of the obtained food types in the database into a blank set one by one, and obtain the set Among them, m represents the set The number of elements in the set The value of the adverse reaction coefficient corresponding to the yth element in is recorded as
[0125] S2.4.3, obtain the second weighted value r2,
[0126] when When
[0127] when , the r2=0.
[0128] In this embodiment, if the first logical coefficient corresponding to the first trial patient is 3.5, D1 = {k1, k2, k3, k4}, B1 = {h1, h2}, r1 = 0.6,
[0129] If k1, k2, k3, and k4 all represent the corresponding food types in the daily diet of the first trial patient, and h1 and h2 are the adverse manifestations in the clinical adverse drug reaction feature set of the first trial patient,
[0130] And the adverse reaction coefficient corresponding to k1 in the database is 0.8,
[0131] The adverse reaction coefficient corresponding to k2 in the database is 0.6.
[0132] The adverse reaction coefficient corresponding to k3 in the database is 1.2.
[0133] The adverse reaction coefficient corresponding to k4 in the database is 1.0.
[0134] When k1 and drug component j1 in the drug category to be visited produce adverse performance h2, k2 and drug component j2 in the drug category to be visited produce adverse performance h3, k3 does not produce adverse performance with all drug components in the drug category to be visited, and k4 and drug component j3 in the drug category to be visited produce adverse performance h1,
[0135] Since h2 is contained in B1, h1 is contained in B1, and h3 is not contained in B1,
[0136] Therefore, the second logical coefficient corresponding to the first trial patient is Q21=0.8+1.0=1.8;
[0137] Since k2 and the drug component j2 in the drug category to be accessed produce an adverse reaction h3, and the adverse reaction coefficient corresponding to k3 in the database is 1.2, then
[0138] Furthermore, the logical value of the first trial patient is Q1=3.5*0.6+1.8*0.6=2.1+1.08=3.18.
[0139] When the logic verification module of the present invention performs a logical analysis on the clinical drug adverse reaction feature set corresponding to each trial drug patient and the historical medical record feature set and daily behavior feature set corresponding to the trial drug patient, the logic verification set corresponding to the nth trial drug patient is recorded as An, in order to summarize the various feature information of the trial drug patient to avoid the mismatch between the analyzed feature information and the trial drug patient during the logical analysis, thereby causing deviations in the analysis results; the first logic coefficient and the second logic coefficient corresponding to the trial drug patient are obtained respectively, because when the associated information of the trial drug patient is logically verified, the features in the historical medical record feature set and the daily behavior feature set will produce adverse reactions when reacting with the drug type to be accessed, so it is necessary to analyze from two aspects, namely corresponding to two analysis results (the first logic coefficient and the second logic coefficient); after determining the food types in Dn that can produce adverse reactions with the drug components in the drug type to be accessed, the adverse reactions produced are also compared with the adverse manifestations in Bn, in order to determine whether the adverse reactions produced by the trial drug patient itself are caused by the food types corresponding to the adverse reactions. If not, it means that the food The type of food has no adverse effect on the trial patient, and thus there is no logical relationship between the food type and the adverse reaction of the trial patient, and thus there is no need to conduct statistics on it; by comparing the database, the adverse reaction coefficient corresponding to each food type is obtained, and the adverse reaction coefficients corresponding to the statistical food types are accumulated and summed up, because different food types have different degrees of reaction with the components of the drug type to be accessed, and the degree of adverse reaction caused to the patient is also different. Therefore, by setting the adverse reaction coefficient, the patient's adverse reaction situation can be quantified and concretized, which is convenient for the subsequent logical verification of the related information of the trial patient, and also convenient for the subsequent priority division of the drug alert report corresponding to the trial patient; when obtaining the logical value of the trial patient, a weighted method is adopted, and r1 and r2 are set, because the first logical coefficient and the second logical coefficient of the trial patient are obtained by different standards, and both reflect the adverse reaction of the trial patient. Therefore, the weighted method can effectively combine the acquisition results corresponding to the two standards, that is, to achieve the unification of the two acquisition standards, so that the final analysis result is more accurate.
[0140] The method of comparing the obtained results in S2.2 with each contraindicated drug corresponding to each historical medical record in Cn corresponding to the nth trial patient comprises the following steps:
[0141] S2.2.1. List the contraindicated drugs in Cn that have the same drug ingredients as the drug category to be visited, and form a data pair [p1, p2] with the contraindicated drugs and the drug ingredients in the drug category to be visited that are the same as the contraindicated drugs.
[0142] Among them, p1 represents a contraindicated drug, and p2 represents a drug ingredient in the drug category to be accessed that is the same as the contraindicated drug;
[0143] S2.2.2. Obtain the historical medical record type corresponding to p1 in each data pair, compare it with the database, and match the final contraindication value corresponding to p1.
[0144] When the historical medical record type corresponding to p1 is one, compare the database to match the taboo value corresponding to p1 in the historical medical record. This taboo value is the final taboo value corresponding to p1.
[0145] When there are multiple types of historical medical records corresponding to p1, the database is compared to match the taboo values corresponding to p1 in each type of historical medical record, and the maximum value of the obtained taboo values is the final taboo value of p1.
[0146] When the present invention compares the obtained results with each contraindication drug corresponding to each historical medical record in Cn corresponding to the nth trial drug patient, the contraindication drug and the drug component in the drug type to be accessed that is the same as the contraindication drug constitute a data pair. This is because each historical medical record of the trial drug patient may correspond to multiple contraindications, and the trial drug patient may contain multiple historical medical records. Different drug components of the drug type to be accessed may be the same as different contraindications, thereby producing different adverse reactions. Therefore, the use of data pairs can accurately lock the drug component of the drug type to be accessed that produces adverse reactions. and the corresponding contraindications, making the analysis more accurate and detailed; when there are multiple types of historical medical records corresponding to p1, the database is compared to match the contraindication values corresponding to p1 in each historical medical record, and the maximum value of the contraindication values obtained is the final contraindication value of p1. This is because there are multiple historical medical records for trial patients, each of which corresponds to multiple contraindications, and when the same contraindication drug corresponding to different historical medical records has the same drug ingredient as the drug type to be accessed, since the contraindication values corresponding to the contraindication drug in different historical medical records are different, in this case, the data pair corresponds to multiple contraindication values at the same time.
[0147] The method for obtaining the first logical coefficient Q1n corresponding to the nth drug trial patient in S2.2 comprises the following steps:
[0148] S2.2-1. Obtain each data pair in S2.2.1;
[0149] S2.2-2. The adverse performance caused by the drug ingredients in the drug category to be visited in each data pair that are the same as the contraindicated drugs is recorded as e, and e is compared with Bn.
[0150] When e is included in Bn, the corresponding final taboo value in the data pair corresponding to e is obtained.
[0151] When e is not included in Bn, the corresponding final taboo value in the data pair corresponding to e is not processed;
[0152] S2.2-3. Calculate the sum of all final contraindication values obtained in S2.2-2 to obtain the first logical coefficient Q1n corresponding to the nth trial patient.
[0153] When all final contraindication values obtained in S2.2-2 are empty, the first logical coefficient Q1n corresponding to the nth trial drug patient is equal to 0;
[0154] The method for obtaining r1 comprises the following steps:
[0155] S2.4-1. Obtain all data pairs corresponding to the nth trial patient in S2.2.1;
[0156] S2.4-2, screen the data pairs whose corresponding bad performance e in S2.4-1 is not included in Bn, and enter the final taboo values corresponding to each screened data pair into a blank set one by one, and obtain Among them, m1 represents the set The number of elements in, n represents the nth trial patient, the set The final taboo value corresponding to the y1th element in is recorded as
[0157] S2.4-3, obtain the first weighted value r1,
[0158] when When
[0159] when , the r1=0.
[0160] When the present invention obtains the first logical coefficient Q1n corresponding to the nth trial patient, it is necessary to screen the data pairs (by comparing the adverse manifestations corresponding to the data pairs with Bn) to obtain the corresponding first logical coefficient; when all the final contraindication values obtained in S2.2-2 are empty, it means that the trial patient has no adverse reactions or the adverse reactions produced do not match the adverse manifestations corresponding to the data pairs, and then it is determined that the first logical coefficient Q1n corresponding to the nth trial patient is equal to 0.
[0161] The method for the pharmacovigilance report generation module to obtain the pharmacovigilance report corresponding to each drug trial patient comprises the following steps:
[0162] S3.1, obtain the data pairs corresponding to each e contained in Bn in S2.2-2;
[0163] S3.2, extract the contraindicated drugs in each data pair in S3.1 and enter them one by one into the suspected drug section of the pharmacovigilance report corresponding to the nth trial patient;
[0164] S3.3. Extract the adverse manifestations corresponding to each data pair in S3.1 respectively, and enter them one by one into the adverse manifestation part of the pharmacovigilance report corresponding to the nth trial patient;
[0165] S3.4. Obtain the personal information of the nth trial patient and enter it into the personal information section of the pharmacovigilance report corresponding to the nth trial patient.
[0166] The pharmacovigilance report corresponding to the trial patient includes a suspicious drug section, an adverse reaction section and a personal information section.
[0167] When the pharmacovigilance report generating module of the present invention fails to obtain the corresponding data pair, the pharmacovigilance report corresponding to the drug trial patient includes a suspicious drug part and an adverse performance part which are empty.
[0168] When the pharmacovigilance report generation module determines the priority of the pharmacovigilance report based on the content of the pharmacovigilance report,
[0169] Sort the logical values of the drug trial patients from large to small. The drug trial patients with higher serial numbers have higher priority for the pharmacovigilance reports.
[0170] When the logical values of the drug trial patients are the same, the number of data pairs corresponding to the drug trial patients is further compared, and the drug trial patients with more corresponding data pairs have a higher priority for the corresponding pharmacovigilance report;
[0171] If the numbers of data pairs corresponding to the trial patients are the same, the priorities of the drug vigilance reports corresponding to the trial patients with the same numbers of data pairs will be randomly sorted.
[0172] The pharmacovigilance report generation module of the present invention determines the priority of the pharmacovigilance report so that when a user accesses a drug type, the displayed pharmacovigilance reports are orderly, and the impact caused by the corresponding adverse reactions is from high to low, so that the user can quickly obtain the main information of the pharmacovigilance report related to the drug type to be accessed.
[0173] A method for accessing a pharmacovigilance system based on artificial intelligence, the method comprising the following steps:
[0174] S1. Query the type of medicine to be accessed through the medicine query module;
[0175] S2, obtaining various related information of the trial patients of the drug type to be accessed in the drug query module through the patient related information acquisition module;
[0176] S3. In the verification module, logic verification is performed on various related information of the trial patients of the drug type to be accessed obtained in the patient related information acquisition module to determine whether there is suspicious information in the related information, and manual verification is performed on the suspicious information;
[0177] S4. The pharmacovigilance report generation module aggregates the related information of the trial patients of the drug types to be accessed that have passed the verification and check in the verification and check module, obtains the pharmacovigilance report corresponding to each trial patient, and determines the priority of the pharmacovigilance report based on the content of the pharmacovigilance report.
[0178] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.
[0179] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. An artificial intelligence-based pharmacovigilance system access system, characterized in that: include: A drug query module, wherein the drug query module is used to query the types of drugs to be accessed; A pharmacovigilance information acquisition module, which includes a patient-related information acquisition module, a verification module, and a pharmacovigilance report generation module. The patient-related information acquisition module is used to acquire various related information of the trial patients of the drug type to be accessed in the drug query module; The verification module is used to perform logic verification on various pieces of associated information of the trial patients of the drug type to be accessed obtained in the patient associated information acquisition module, determine whether there is suspicious information in the associated information, and perform manual verification on the suspicious information; The pharmacovigilance report generation module is used to aggregate the relevant information of the trial drug patients of the drug type to be accessed that has passed the verification and verification in the verification and verification module, obtain the pharmacovigilance report corresponding to each trial drug patient, and determine the priority of the pharmacovigilance report according to the content of the pharmacovigilance report; The patient-related information acquisition module acquires various related information of the trial patients of the drug types to be accessed in the drug query module, including clinical drug adverse reaction reports, patient historical medical records and patient daily behavior; The verification and checking module includes a logic verification module and a manual verification module. The logic verification module is used to obtain the clinical adverse drug reaction feature set, historical medical record feature set and daily behavior feature set corresponding to each trial drug patient, and perform a logical analysis on the clinical adverse drug reaction feature set corresponding to each trial drug patient and the historical medical record feature set and daily behavior feature set corresponding to the trial drug patient, to determine whether the clinical adverse drug reaction feature set corresponding to the trial drug patient and the corresponding historical medical record feature set and daily behavior feature set conform to a logical relationship. If the corresponding clinical adverse drug reaction feature set of the trial drug patient and the corresponding historical medical record feature set and daily behavior feature set have a logical relationship, then the associated information of the trial drug patient is judged to be passed, and there is no suspicious information in the associated information. The associated information of the trial drug patient who has passed the test is added to the verification database. If there is no logical relationship between the clinical adverse drug reaction feature set of the trial patient and the corresponding historical medical record feature set and daily behavior feature set, it is determined that the associated information of the trial patient has failed and there is suspicious information in the associated information; The manual verification module submits the various items of relevant information corresponding to the trial drug patients whose relevant information in the logic verification module is suspicious to the management personnel for verification. The management personnel adds the various items of relevant information corresponding to the trial drug patients that have passed the verification to the verification database, and does not process the various items of relevant information corresponding to the trial drug patients that have not passed the verification; The method for the logic verification module to obtain the clinical adverse drug reaction feature set, historical medical record feature set and daily behavior feature set corresponding to each drug trial patient includes the following steps: S1.
1. Obtain the corresponding clinical adverse drug reaction feature set for each trial patient, wherein the clinical adverse drug reaction features include the time, type and adverse manifestation of the adverse reaction. When the trial drug patient has no adverse reaction, the adverse reaction feature set corresponding to the trial drug patient is empty; S1.
2. Obtain a set of historical medical record features corresponding to each drug trial patient, wherein the historical medical record features include the type of historical medical record and various contraindications corresponding to each historical medical record. When there is no historical medical record for the trial drug patient, the historical medical record feature set corresponding to the trial drug patient is empty; S1.
3. Obtain a set of daily behavior characteristics corresponding to each drug trial patient, wherein the daily behavior characteristics include the corresponding food types in the daily diet of the drug trial patient; The method for the logic verification module to perform logic analysis on the clinical adverse drug reaction feature set corresponding to each trial drug patient and the historical medical record feature set and daily behavior feature set corresponding to the trial drug patient comprises the following steps: S2.
1. Obtain the clinical adverse drug reaction feature set, historical medical record feature set and daily behavior feature set corresponding to each trial drug patient to form a logical verification data set corresponding to the trial drug patient. The logical verification set corresponding to the nth trial drug patient is recorded as An, where An={Bn, Cn, Dn}. Among them, Bn represents the clinical adverse drug reaction feature set corresponding to the nth trial patient. Cn represents the historical medical record feature set corresponding to the nth drug trial patient, Dn represents the daily behavior feature set corresponding to the nth drug trial patient; S2.2, obtain the drug ingredients in the drug category to be accessed, and compare the obtained results with each contraindicated drug corresponding to each historical medical record in Cn corresponding to the nth drug trial patient, and match the comparison results with Bn to obtain the first logical coefficient Q1n corresponding to the nth drug trial patient; S2.3, obtain the food types in the daily diet of the drug trial patient in Dn corresponding to the nth drug trial patient, determine whether the obtained food types can produce adverse reactions with the drug components in the drug types to be visited, and count the food types in Dn that can produce adverse reactions with the drug components in the drug types to be visited, and the adverse reactions produced are the same as the adverse manifestations in Bn, compare with the database, obtain the adverse reaction coefficient corresponding to each food type, accumulate and sum the adverse reaction coefficients corresponding to the statistical food types, and then obtain the second logical coefficient Q2n; S2.4, obtain the logical value Qn of the nth trial patient, , wherein r1 represents the first weighted value, and r2 represents the second weighted value; S2.5, compare Qn with the first preset value, When Qn is greater than or equal to the first preset value, it is determined that the associated information of the trial patient passes. When Qn is less than the first preset value, it is determined that the associated information of the trial patient has failed; The method for obtaining r2 comprises the following steps: S2.4.
1. Obtain the food types in the daily diet of the trial patient in Dn corresponding to the nth trial patient in S2.3; S2.4.
2. Obtain the food types in Dn that can produce adverse reactions with the drug ingredients in the drug type to be accessed, and the adverse reactions produced are different from the adverse manifestations in Bn, and enter the corresponding adverse reaction coefficients of the obtained food types in the database into a blank set one by one, and obtain the set , where m represents the set The number of elements in the set The value of the adverse reaction coefficient corresponding to the yth element in is recorded as ; S2.4.3, obtain the second weighted value r2, when ≠0, the , when = 0, the ; The method of comparing the obtained results in S2.2 with each contraindicated drug corresponding to each historical medical record in Cn corresponding to the nth trial drug patient comprises the following steps: S2.2.
1. List the contraindicated drugs in Cn that have the same drug ingredients as the drug category to be visited, and form a data pair [p1, p2] with the contraindicated drugs and the drug ingredients in the drug category to be visited that are the same as the contraindicated drugs. Among them, p1 represents a contraindicated drug, and p2 represents a drug ingredient in the drug category to be accessed that is the same as the contraindicated drug; S2.2.
2. Obtain the historical medical record type corresponding to p1 in each data pair, compare it with the database, and match the final contraindication value corresponding to p1. When the historical medical record type corresponding to p1 is one, compare the database to match the taboo value corresponding to p1 in the historical medical record. This taboo value is the final taboo value corresponding to p1. When there are multiple types of historical medical records corresponding to p1, the database is compared to match the taboo values corresponding to p1 in each type of historical medical record. The maximum value of the obtained taboo values is the final taboo value of p1. The method for obtaining the first logical coefficient Q1n corresponding to the nth drug trial patient in S2.2 comprises the following steps: S2.2-1. Obtain each data pair in S2.2.1; S2.2-2. The adverse performance caused by the drug ingredients in the drug category to be visited in each data pair that are the same as the contraindicated drugs is recorded as e, and e is compared with Bn. When e is included in Bn, the corresponding final taboo value in the data pair corresponding to e is obtained. When e is not included in Bn, the corresponding final taboo value in the data pair corresponding to e is not processed; S2.2-3. Calculate the sum of all final contraindication values obtained in S2.2-2 to obtain the first logical coefficient Q1n corresponding to the nth trial patient. When all final contraindication values obtained in S2.2-2 are empty, the first logical coefficient Q1n corresponding to the nth trial drug patient is equal to 0; The method for obtaining r1 comprises the following steps: S2.4-1. Obtain all data pairs corresponding to the nth trial patient in S2.2.1; S2.4-2, screen the data pairs whose corresponding bad performance e in S2.4-1 is not included in Bn, and enter the final taboo values corresponding to each screened data pair into a blank set one by one, and obtain , where m1 represents the set The number of elements in, n represents the nth trial patient, the set The final taboo value corresponding to the y1th element in is recorded as ; S2.4-3, obtain the first weighted value r1, when ≠0, the , when = 0, the .
2. The artificial intelligence-based pharmacovigilance system access system according to claim 1, characterized in that: The method for the pharmacovigilance report generation module to obtain the pharmacovigilance report corresponding to each drug trial patient comprises the following steps: S3.1, obtain the data pairs corresponding to each e contained in Bn in S2.2-2; S3.2, extract the contraindicated drugs in each data pair in S3.1 and enter them one by one into the suspected drug section of the pharmacovigilance report corresponding to the nth trial patient; S3.
3. Extract the adverse manifestations corresponding to each data pair in S3.1 respectively, and enter them one by one into the adverse manifestation part of the pharmacovigilance report corresponding to the nth trial patient; S3.
4. Obtain the personal information of the nth trial patient and enter it into the personal information section of the pharmacovigilance report corresponding to the nth trial patient. The pharmacovigilance report corresponding to the trial patient includes a suspicious drug section, an adverse reaction section and a personal information section.
3. The artificial intelligence-based pharmacovigilance system access system according to claim 1, characterized in that: When the pharmacovigilance report generation module determines the priority of the pharmacovigilance report based on the content of the pharmacovigilance report, Sort the logical values of the drug trial patients from large to small. The drug trial patients with higher serial numbers have higher priority for the pharmacovigilance reports. When the logical values of the drug trial patients are the same, the number of data pairs corresponding to the drug trial patients is further compared, and the drug trial patients with more corresponding data pairs have a higher priority for the corresponding pharmacovigilance report; If the numbers of data pairs corresponding to the trial patients are the same, the priorities of the drug vigilance reports corresponding to the trial patients with the same numbers of data pairs will be randomly sorted.
4. An artificial intelligence-based pharmacovigilance system access method using an artificial intelligence-based pharmacovigilance system access system according to any one of claims 1 to 3, characterized in that: The method comprises the following steps: S1. Query the type of medicine to be accessed through the medicine query module; S2, obtaining various related information of the trial patients of the drug type to be accessed in the drug query module through the patient related information acquisition module; S3. In the verification module, logic verification is performed on various related information of the trial patients of the drug type to be accessed obtained in the patient related information acquisition module to determine whether there is suspicious information in the related information, and manual verification is performed on the suspicious information; S4. The pharmacovigilance report generation module aggregates the related information of the trial patients of the drug type to be accessed that has passed the verification and check in the verification and check module, obtains the pharmacovigilance report corresponding to each trial patient, and determines the priority of the pharmacovigilance report based on the content of the pharmacovigilance report.