Similar case recommendation method and system

Abstract

The application discloses a similar case recommendation method and system, belongs to the technical field of clinical decision support system, and aims to solve the technical problem that the implementation difficulty of similar case recommendation is extremely great and the feasibility is not high by using a deep learning algorithm and a natural language processing technology.The technical scheme is as follows:the method is specifically as follows: data preprocessing: extracting the symptom and diagnosis information of the case admission record electronic medical record; and processing the case into a standard symptom and standard diagnosis list, and extracting the treatment information of the case for storage in a dictionary to assist in subsequent acquisition of the similarity degree of the case; wherein, the treatment information comprises age and department; obtaining the symptom weight: based on the knowledge graph of the symptom and disease related knowledge, extracting the weight of the symptom in the disease diagnosis; obtaining the similarity degree of the case: according to the symptom and diagnosis list of the target case, and fusing the weight to obtain the similarity degree of other cases and the target case; similar case recommendation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of clinical decision support system (CDSS), in particular to a similar case recommendation method and system. BACKGROUND

[0002] Similar case recommendation is a breakthrough of recommendation system applied in smart medical care, that is, by recommending cases with the same disease and symptoms to a doctor, the doctor's clinical diagnosis and treatment are assisted.

[0003] At present, many similar case recommendation methods are based on deep learning artificial intelligence algorithm, which calculates the cosine similarity and other indicators by understanding the semantics, and uses the natural language processing technology to calculate the text similarity to calculate the case similarity.

[0004] However, it is difficult to use deep learning algorithm and natural language processing technology to realize similar case recommendation, the deep learning model has poor interpretability, and the clinical doctors need to evaluate the similarity of sample cases to obtain training samples, which is prone to errors of different doctors having different evaluation opinions, so the reliability of the training samples is difficult to guarantee; and massive case data support is necessary for the accuracy of the model, and data collection is very difficult for general enterprises and individuals, especially electronic medical records, which are extremely private personal data, are difficult to collect and process; finally, the deep learning model has high requirements for server and other hardware configurations, which further increases the implementation difficulty and cost.

[0005] In summary, the current similar case recommendation has the problem of great difficulty in realizing similar case recommendation by using deep learning algorithm and natural language processing technology, and low feasibility. SUMMARY

[0006] The technical task of the present application is to provide a similar case recommendation method and system to solve the problem of great difficulty in realizing similar case recommendation based on deep learning and natural language processing technology and low feasibility.

[0007] The technical task of the present application is realized in the following way, a similar case recommendation method, the method is as follows:

[0008] Data preprocessing: extracting the symptoms and diagnosis information of the case admission record electronic medical record; and processing the case into a standard symptom and standard diagnosis list, and extracting the case visit information at the time of visiting the doctor and storing it in a dictionary to assist subsequent acquisition of case similarity; wherein the visit information includes age and department;

[0009] Obtaining symptom weight: based on the knowledge graph of the symptoms and disease related knowledge, extracting the weight of the symptoms in the disease diagnosis;

[0010] Obtaining case similarity: according to the symptoms and diagnosis list of the target case, and fusing the weight to obtain the similarity of other cases with the target case;

[0011] Similar case recommendation: according to the calculation results of symptoms, diagnosis and case similarity, the case list is sorted in descending order according to the similarity from large to small, and combined with the age and department of the case, the cases with high similarity, same department and same age (0-6 years old, 6-18 years old, 18-60 years old, more than 60 years old) are recommended.

[0012] As preferred, the data preprocessing further includes normalizing the symptom and diagnosis information of all cases in the case library, specifically as follows:

[0013] Taking the case number as the index, obtaining the symptom list and diagnosis list of all cases in the case library in the cache;

[0014] According to the mapping relationship table of original words and normalized words in the knowledge graph, the normalized word list of case symptoms and diagnoses is obtained by matching the original word name and original word entity type in cache storage; wherein, the original word name includes symptom name and / or diagnosis name; the original word entity type includes symptom and / or diagnosis.

[0015] As preferred, the symptom weight is obtained as follows:

[0016] Obtain the data related to symptoms and diseases in the knowledge graph, extract the knowledge type (starting entity type-relation-terminating entity type) as all related knowledge of disease-related symptoms-symptoms, and store it in the symptom and disease knowledge table;

[0017] Through the extraction of the related knowledge of symptoms-diseases in the knowledge graph, the number of diseases N i associated with any symptom i and the total number of diseases N in the knowledge base are obtained, so as to calculate the symptom weight.

[0018] More preferably, through the extraction of the related knowledge of symptoms-diseases in the knowledge graph, the number of diseases N i associated with any symptom i and the total number of diseases N in the knowledge base are obtained, so as to calculate the symptom weight as follows:

[0019] Counting the de-duplication of the normalized words of the starting entity, that is, obtaining the total number of diseases N in the knowledge base;

[0020] Extracting the de-duplication of the normalized words of the terminating entity in the symptom and disease knowledge table to obtain the symptom list of the knowledge base;

[0021] According to the symptom list of the knowledge base, the starting entity normalized word set corresponding to each symptom i is extracted in turn, and the number of starting entity normalized words is calculated, that is, the number of diseases associated with each symptom i N i ;

[0022] According to the symptom list of the knowledge base and the number of diseases associated with each symptom i N i , a symptom-weight dictionary is obtained, and the formula is as follows:

[0023]

[0024] Where, W i is the symptom coefficient of the case similarity degree; N-N i represents the number of diseases irrelevant to the symptom; and N represents the total number of diseases in the knowledge base.

[0025] More preferably, the case similarity degree is obtained as follows:

[0026] The symptom and diagnosis lists of the target case are matched with the symptom and diagnosis lists of other cases in the case library, respectively, to obtain the intersection set of the symptoms and diagnoses of the target case and other cases;

[0027] The symptom similarity degree is obtained according to the intersection set of the symptoms of the target case and other cases;

[0028] The diagnosis similarity degree is obtained according to the intersection set of the diagnoses of the target case and other cases;

[0029] The symptom similarity degree and the diagnosis similarity degree are subjected to weighted operation to obtain the case similarity degree;

[0030] The basic information of the case is combined, and the table form with the case number, age, department, symptom similarity degree, diagnosis similarity degree and case similarity degree as the table header is stored in the cache.

[0031] More preferably, the symptom similarity degree is as follows:

[0032] Let the symptom list of the compared case be [i1, i2, i3…i n ]; the symptom intersection set of the compared case and the target case be [i1, i2, i3…i m ]; and the symptom weight be W i , then the calculation formula of the corresponding case and the target case is as follows:

[0033]

[0034] Where, S represents the symptom similarity degree; W i1 +W i2 +W i3 +…+W imThe symptom weight accumulation representing the symptom intersection set obtains the intersection score; n represents the number of symptom list elements of the compared case;

[0035] The diagnosis similarity is specifically as follows:

[0036] Suppose the diagnosis list of the compared case is [z1, z2, z3…z k ]; the diagnosis intersection set of the compared case and the target case is [z1, z2, z3…z j ]; then the symptom similarity degree calculation formula of the corresponding case and the target case is as follows:

[0037]

[0038] Wherein, S' represents the symptom similarity degree; j represents the number of the diagnosis intersection set; k represents the number of the diagnosis list elements of the compared case;

[0039] The case similarity degree is obtained by averaging the symptom similarity degree S and the diagnosis similarity degree S', and the formula is as follows:

[0040]

[0041] Wherein, F represents the case similarity degree.

[0042] A similar case recommendation system, the system includes a data layer and a data processing layer; the data layer includes a case library management module, the data processing layer includes a preprocessing module, a symptom weight acquisition module, a case similarity acquisition module and a similar case recommendation module;

[0043] Wherein, the case library management module is used to maintain the basic information of the case, the structured large text of the electronic medical record and the symptom and diagnosis information;

[0044] The preprocessing module is used to normalize the symptom and diagnosis information, and according to the original word and normalized word mapping table of the knowledge graph, the original word to normalized word mapping of the symptom and diagnosis information of all cases including the target case in the case library is carried out, the standard symptom and standard diagnosis information of the case are obtained and stored in the form of list; wherein, the normalization processing is to obtain the normalized word corresponding to the symptom and diagnosis according to the original word and normalized word mapping relationship table of the knowledge graph;

[0045] The symptom weight acquisition module is used to acquire the symptom weight based on the knowledge type of the disease-related symptom-symptom in the knowledge graph, and form a symptom-weight dictionary;

[0046] The case similarity acquisition module is used to calculate the symptom similarity, diagnosis similarity, and case similarity of all other cases based on the symptom and diagnosis list of all cases in the input case database and the symptom weight dictionary, using the symptoms and diagnosis of the target case as the comparison standard, and store them in dictionary form;

[0047] The similar case recommendation module is used to sort cases in reverse order based on symptoms, diagnosis and the calculation results of case similarity, and to obtain a list of cases with the highest to lowest similarity. It also combines the age and department visit information of the cases to give priority to recommending cases with high similarity, from the same department and the same age group (0-6 years, 6-18 years, 18-60 years, 60 years and above).

[0048] As a preferred embodiment, the working process of the symptom weight acquisition module is as follows:

[0049] (1) Obtain data related to symptoms and diseases in the knowledge graph, and extract all related knowledge types (starting entity type-relationship-ending entity type) as disease-related symptoms-symptoms;

[0050] (2) Deduplicate the starting entities of all related knowledge of disease-related symptoms-symptoms and calculate the total number of diseases N in the knowledge base.

[0051] (3) Extract the terminator of all relevant knowledge of disease-related symptoms-symptoms by deduplication to obtain the symptom list of the knowledge base.

[0052] (4) Based on the symptom list in the knowledge base, extract the set of starting entities corresponding to each symptom i by deduplication, and obtain the number of diseases N associated with each symptom i by calculating the number of starting entities corresponding to each term. i ;

[0053] (5) For the symptom list in the knowledge base, sequentially calculate the number of diseases N associated with each acquired symptom. i The symptom-weight dictionary is obtained by combining the total number of diseases N with the symptom-weight dictionary, as shown in the following formula:

[0054]

[0055] Among them, W i Symptom coefficients representing the degree of case similarity; NN i This represents the number of diseases unrelated to this symptom; N represents the total number of diseases in the knowledge base.

[0056] An electronic device includes: a memory and at least one processor;

[0057] The memory contains computer programs;

[0058] The at least one processor executes the computer program stored in the memory, so that the at least one processor executes the similar case recommendation method as described above.

[0059] A computer readable storage medium, the computer readable storage medium stores a computer program, the computer program can be executed by a processor to implement the similar case recommendation method as described above.

[0060] The similar case recommendation method and system of the present application have the following advantages:

[0061] (I) The present application normalizes the symptoms and diagnosis information of the case, calculates the case similarity by fusing the symptom weight, so that the accuracy of similar case recommendation is strong and easy to implement;

[0062] (II) The present application overcomes the difficulty of implementing similar case recommendation based on deep learning and natural language processing technology, compares the symptoms and diagnosis information of the case to make similar degree reasoning operation, and introduces the concept of symptom weight through the related knowledge of symptoms and diseases in the medical knowledge graph, and fuses into the calculation of case similarity, aiming to improve the proportion of rare symptoms in similarity reasoning, while being easy to implement, improve the accuracy of case similarity calculation;

[0063] (III) The present application can recommend similar case information based on the symptoms and diagnosis information extracted from the electronic medical record of the target case, and fuse the weight of the symptoms in disease diagnosis, assist clinical diagnosis and clinical decision-making;

[0064] (IV) Considering that the similarity of the case in the symptoms is not only related to the number of matched symptoms, but also related to the rare degree of the matched symptoms, therefore, the present application adds the concept of symptom weight to improve the accuracy of calculating the similarity of the symptoms. BRIEF DESCRIPTION OF DRAWINGS

[0065] The present application will be further described below in conjunction with the accompanying drawings.

[0066] ATTACHED Figure 1 The flowchart of the similar case recommendation method. DETAILED DESCRIPTION

[0067] The similar case recommendation method and system of the present application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0068] Embodiment 1:

[0069] This embodiment provides a similar case recommendation method, which is specifically as follows:

[0070] S1, normalize the symptom and diagnosis information of all cases in the case library, and obtain the standard symptom list and standard diagnosis list of each case respectively;

[0071] S2, obtain the data related to symptoms and diseases in the knowledge graph, and calculate the symptom weight;

[0072] S3, according to the matching of the standard symptom list of each case and the standard symptom list of the target case, fuse the symptom weight, and calculate the symptom similarity degree; according to the matching of the diagnosis set of each case and the diagnosis list of the target case, calculate the diagnosis similarity degree; take the average of the symptom similarity degree and the diagnosis similarity degree, and calculate the similarity degree of each case and the target case;

[0073] S4, reorder the top 5 cases obtained by the similarity degree, combine the age and department information of the case, and rank the cases in the same department and age range as the target case in the top 5 cases in advance, and obtain the final similar case recommendation result.

[0074] In this embodiment, the normalization of the symptoms and diagnosis information of all cases in the case library in step S1 is as follows:

[0075] S101, take the case number as the index, and obtain the symptom list and diagnosis list of all cases in the case library in the cache;

[0076] S102, according to the mapping relationship table of the original word and the normalized word in the knowledge graph, the normalized word list of the case symptoms and diagnoses is obtained by matching the original word name (symptom name / diagnosis name) and the original word entity type (symptom / diagnosis) in the cache storage; The structure of the original word and the normalized word mapping table of the knowledge graph is as follows:

[0077] Serial number Field name Field meaning 1 original_word Original word 2 entity_type_id Entity type id 3 normalize_term_id Normalized term id 4 normalize_term_cn Normalized Chinese name 5 status Knowledge status 6 update_time Update time

[0078] The symptom weight obtained in step S2 of this embodiment is as follows:

[0079] S201, obtain the data related to symptoms and diseases in the knowledge graph, extract the knowledge type (starting entity type-relation-terminating entity type) as all related knowledge of disease-related symptoms-symptoms, and store it in the symptom and disease knowledge table; as shown in the following table:

[0080]

[0081]

[0082] S202, through the extraction of the related knowledge of symptoms-diseases in the knowledge graph, obtain the number N of diseases associated with any symptom i i and the total number of diseases in the knowledge base N, so as to calculate the symptom weight.

[0083] In step S202 of the embodiment, the number N of diseases associated with any symptom i is obtained by extracting the relevant knowledge of symptoms-diseases in the knowledge graph i and the total number N of diseases in the knowledge base, so as to calculate the symptom weight as follows:

[0084] S20201, the starting entity of the normalized word is counted, that is, the total number N of diseases in the knowledge base is obtained;

[0085] S20202, the ending entity of the normalized word in the symptom-disease knowledge table is extracted, and the symptom list of the knowledge base is obtained;

[0086] S20203, according to the symptom list of the knowledge base, the starting entity normalized word set corresponding to each symptom i is extracted in turn, and the number of starting entity normalized words is calculated, that is, the number N of diseases associated with each symptom i is obtained i ;

[0087] S20204, according to the symptom list of the knowledge base and the number N of diseases associated with each symptom i i , the symptom-weight dictionary is obtained, and the formula is as follows:

[0088]

[0089] Where, W i is the symptom coefficient of the case similarity; N-N i represents the number of diseases irrelevant to the symptom; N represents the total number of diseases in the knowledge base.

[0090] In step S3 of the embodiment, the case similarity is obtained as follows:

[0091] S301, the symptoms and diagnosis list of the target case are matched with the symptoms and diagnosis list of other cases in the case base, respectively, to obtain the intersection set of the symptoms and diagnosis of the target case and other cases;

[0092] S302, the symptom similarity is obtained according to the intersection set of the symptoms of the target case and other cases;

[0093] S303, the diagnosis similarity is obtained according to the intersection set of the diagnosis of the target case and other cases;

[0094] S304, the symptom similarity and the diagnosis similarity are weighted and operated to obtain the case similarity;

[0095] S305, the basic information of the case is combined, and the table form with the table header of case number, age, department, symptom similarity, diagnosis similarity and case similarity is stored in the cache.

[0096] The symptom similarity in step S302 of the embodiment is specifically as follows:

[0097] Suppose the symptom list of the compared case is [i1, i2, i3…i n ]; the intersection set of the symptoms of the compared case and the target case is [i1, i2, i3…i m ]; and the symptom weight is W i , then the calculation formula of the corresponding case and the target case is as follows:

[0098]

[0099] Wherein, S represents the symptom similarity degree; W i1 +W i2 +W i3 +…+W im represents the intersection score obtained by accumulating the symptom weight of the intersection set; and n represents the number of elements of the symptom list of the compared case.

[0100] The diagnosis similarity in step S303 of the embodiment is specifically as follows:

[0101] Suppose the diagnosis list of the compared case is [z1, z2, z3…z k ]; the intersection set of the diagnoses of the compared case and the target case is [z1, z2, z3…z j ]; then the calculation formula of the symptom similarity degree of the corresponding case and the target case is as follows:

[0102]

[0103] Wherein, S’ represents the symptom similarity degree; j represents the number of the intersection set of the diagnoses; and k represents the number of elements of the diagnosis list of the compared case.

[0104] The case similarity degree in step S304 of the embodiment is obtained by averaging the symptom similarity degree S and the diagnosis similarity degree S’, and the formula is as follows:

[0105]

[0106] Wherein, F represents the case similarity degree.

[0107] According to the front-end requirements, the generated table can be sorted according to the symptom similarity degree, the diagnosis similarity degree, and the case similarity degree, respectively, the top 5 records under different sorting conditions are obtained, and each of the top 5 records is reordered according to the age and the department, the cases in the top 5 records which are in the same department and the same age range as the target case are ranked in advance, and the final similar case recommendation result can be obtained.

[0108] Embodiment 2:

[0109] The embodiment provides a similar case recommendation system, which comprises a data layer and a data processing layer; the data layer comprises a case library management module, and the data processing layer comprises a preprocessing module, a symptom weight acquisition module, a case similarity acquisition module and a similar case recommendation module;

[0110] The case library management module is used for maintaining basic information of cases, structured large text of electronic medical records and symptom and diagnosis information;

[0111] The preprocessing module is used for performing normalization processing on the symptom and diagnosis information, and performing mapping of original words to normalized words on the symptom and diagnosis information of all cases in the case library including the target case according to an original word and normalized word mapping table of a knowledge graph, to obtain standard symptom and standard diagnosis information of the case and store the information in a list form; wherein the normalization processing is performed according to the original word and normalized word mapping relationship table of the knowledge graph to obtain normalized words corresponding to the symptom and the diagnosis;

[0112] The symptom weight acquisition module is used for acquiring symptom weights based on knowledge of the knowledge graph with a knowledge type of disease-related symptom-symptom, and forming a symptom-weight dictionary;

[0113] The case similarity acquisition module is used for calculating the symptom similarity degree, the diagnosis similarity degree and the case similarity degree of all the cases other than the target case based on the symptom and diagnosis lists of all the cases in the case library and the symptom weight dictionary, taking the symptom and diagnosis of the target case as a comparison standard, and storing the degrees in a dictionary form;

[0114] The similar case recommendation module is used for obtaining a case list in descending order of similarity degree from large to small according to the calculation results of the symptom, the diagnosis and the case similarity degree, and preferentially recommending cases with high similarity degree, same department and same age range (0-6 years old, 6-18 years old, 18-60 years old and above 60 years old) in combination with the age and department visiting information of the cases.

[0115] The working process of the symptom weight acquisition module in the embodiment is specifically as follows:

[0116] (1) acquire data related to symptoms and diseases in the knowledge graph, and extract all related knowledge with a knowledge type (starting entity type-relation-ending entity type) of disease-related symptom-symptom;

[0117] (2) remove and count the starting entity normalized words of all related knowledge of disease-related symptom-symptom, to obtain the total number N of diseases in the knowledge base;

[0118] (3) remove and extract the ending entity normalized words of all related knowledge of disease-related symptom-symptom, to obtain a symptom list of the knowledge base;

[0119] (4), sequentially extracting the starting entity normalized word set corresponding to each symptom i from the symptom list of the knowledge base, and obtaining the disease number N of each symptom i associated disease by calculating the number of starting entity normalized words i ;

[0120] (5), for the symptom list of the knowledge base, sequentially obtaining the symptom-weight dictionary according to the disease number N of each symptom associated obtained i and the total disease number N obtained, the formula is as follows:

[0121]

[0122] Where, W i is the symptom coefficient of the case similarity degree; N-N i represents the number of diseases irrelevant to the symptom; N represents the total number of diseases in the knowledge base.

[0123] Embodiment 3:

[0124] The embodiment also provides an electronic device, comprising a memory and a processor;

[0125] The memory stores computer execution instructions;

[0126] The processor executes the computer execution instructions stored in the memory, so that the processor executes the similar case recommendation method in any embodiment of the application.

[0127] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), ready-to-program gate arrays (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The processor can be a microprocessor, or the processor can be any conventional processor.

[0128] The memory can be used to store computer programs and / or modules, and the processor can realize various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function, etc.; the data storage area can store data created according to the use of the terminal, etc. In addition, the memory can also include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a secure digital (SD) card, a flash memory card, at least one magnetic disk storage period, a flash memory device, or other volatile solid-state memory device.

[0129] Embodiment 4:

[0130] The embodiments of the present application also provide a computer readable storage medium, in which a plurality of instructions are stored, and the instructions are loaded by a processor to make the processor execute the similar case recommendation method in any of the embodiments of the present application. Specifically, a system or device equipped with a storage medium can be provided, in which a software program code realizing the functions of any of the above embodiments is stored, and a computer (or CPU or MPU) of the system or device reads and executes the program code stored in the storage medium.

[0131] In this case, the program code read from the storage medium itself realizes the functions of any of the above embodiments, and thus the program code and the storage medium storing the program code constitute a part of the present application.

[0132] The embodiments of the storage medium for storing the program code include a floppy disk, a hard disk, a magneto-optical disk (e.g., CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RYM, DVD-RW, DVD+RW), a magnetic tape, a nonvolatile memory card, and a ROM. Alternatively, the program code can be downloaded from a server computer via a communication network.

[0133] In addition, it should be understood that not only the program code read by the computer, but also the operating system or the like operating on the computer can be caused to perform part or all of the actual operations based on the instructions of the program code, thereby realizing the functions of any of the above embodiments.

[0134] In addition, it should be understood that the program code read from the storage medium can be written into a memory provided in an expansion board inserted into the computer or a memory provided in an expansion unit connected to the computer, and then part or all of the actual operations can be performed based on the instructions of the program code by a CPU or the like mounted on the expansion board or the expansion unit, thereby realizing the functions of any of the above embodiments.

[0135] Finally, it should be noted that the above embodiments are merely used to illustrate the technical solutions of the present application, but not to limit the present application; although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions recorded in the above embodiments can be modified, or part or all of the technical features can be replaced by equivalents; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A similar case recommendation method characterized by comprising: The method specifically comprises the following steps: Data preprocessing: extracting the symptom and diagnosis information of the case admission record electronic medical record; and processing the case as a unit into a standard symptom and standard diagnosis list, while extracting the case's visit information and storing it in a dictionary to assist in subsequent acquisition of the case similarity; wherein the visit information includes age and department; Obtaining symptom weight: based on the knowledge graph related knowledge of symptoms and diseases, extracting the weight of symptoms in disease diagnosis; Obtaining case similarity: according to the symptom and diagnosis list of the target case, and fusing the weight to obtain the similarity of other cases to the target case; Similar case recommendation: according to the calculation results of symptoms, diagnoses and case similarity, the case list is sorted in descending order from high to low similarity, and combined with the age and department visit information of the case, the case with high similarity, same department and same age is recommended; Wherein, the symptom weight is obtained as follows: Obtain the data related to symptoms and diseases in the knowledge graph, extract all related knowledge with the knowledge type of disease-related symptom-symptom, and store it in the symptom and disease knowledge table; By extracting the related knowledge of symptoms-diseases in the knowledge graph, the number of diseases associated with any symptom i is obtained and the total number of diseases in the knowledge base N, so as to calculate the symptom weight; specifically as follows: Count the de-duplication of the normalized words of the starting entity, that is, obtain the total number of diseases in the knowledge base N; Extract the de-duplicated normalized words of the terminal entity in the symptom and disease knowledge table to obtain the symptom list of the knowledge base; According to the symptom list of the knowledge base, the starting entity normalized word set corresponding to each symptom i is extracted in turn, and the number of starting entity normalized words is calculated, so as to obtain the disease number of the disease associated with each symptom i ; According to the symptom list of the knowledge base and the disease number associated with each symptom i , a symptom-weight dictionary is obtained, as follows: ； wherein, a symptom coefficient as a case similarity degree; represents the number of diseases irrelevant to the symptom; N represents the total number of diseases in the knowledge base; The case similarity is obtained as follows: Match the symptom and diagnosis list of the target case with the symptom and diagnosis list of other cases in the case library to obtain the intersection set of the symptoms and diagnoses of the target case and other cases; Obtain the symptom similarity according to the intersection set of the symptoms of the target case and other cases; Obtain the diagnosis similarity according to the intersection set of the diagnoses of the target case and other cases; Weighted operation of symptom similarity and diagnosis similarity to obtain case similarity; Combine the basic information of the case, store it in the cache in the form of a table with the table headers of case number, age, department, symptom similarity, diagnosis similarity and case similarity; The symptom similarity is as follows: Let the list of symptoms for the compared cases be [ The set of symptom intersections between the compared cases and the target case is [ ]. Symptom weighting is... The formula for calculating the corresponding case and the target case is as follows: ； wherein S represents a symptom similarity degree; a symptom weight of a symptom intersection set is accumulated to obtain an intersection score; represents a number of symptom list elements of a compared case; The diagnosis similarity is as follows: Let the diagnosis list of the compared case be [ ]; the intersection set of the diagnosis of the compared case and the target case be [ ]; then the similarity degree of the symptoms of the corresponding case and the target case is calculated according to the following formula: ； wherein, represents the degree of similarity of symptoms; represents the number of diagnostic intersection sets; represents the number of diagnostic list elements of the compared cases; The case similarity is obtained by averaging the symptom similarity S and the diagnosis similarity S', and the formula is as follows: ； Wherein, F represents the case similarity. 2.The similar case recommendation method of claim 1, wherein, The data preprocessing also includes normalizing the symptom and diagnosis information of all cases in the case library, specifically as follows: Take the case number as the index to obtain the symptom list and diagnosis list of all cases in the case library in the cache; According to the mapping relationship table of the original words and the normalized words in the knowledge graph, the normalized word list of the case symptoms and diagnoses is obtained by matching the original word name and the original word entity type and stored in the cache; wherein the original word name includes the symptom name and / or the diagnosis name; the original word entity type includes the symptom and / or the diagnosis.

3. A similar case recommendation system characterized by, The system includes a data layer and a data processing layer; the data layer includes a case library management module, and the data processing layer includes a preprocessing module, a symptom weight acquisition module, a case similarity acquisition module and a similar case recommendation module; Wherein, the case library management module is used to maintain the basic information of the case, the structured large text of the electronic medical record and the symptom and diagnosis information; The preprocessing module is configured to normalize the symptom and diagnosis information, and map the original words to the normalized words according to the mapping table of the original words and the normalized words in the knowledge graph, to obtain the standard symptoms and standard diagnosis information of the cases and store the information in a list form; The symptom weight acquisition module is configured to acquire the symptom weight based on the knowledge type of disease-related symptom-symptom in the knowledge graph, and form a symptom-weight dictionary; The case similarity acquisition module is configured to calculate the symptom similarity, diagnosis similarity and case similarity of all the cases in the case database based on the symptom and diagnosis lists of all the cases in the case database and the symptom weight dictionary, and take the symptom and diagnosis of the target case as the comparison standard, and store the results in a dictionary form; The similar case recommendation module is configured to sort the cases in descending order according to the calculation results of the symptom, diagnosis and case similarity, and recommend the cases with high similarity, same department and same age group as the priority. The working process of the symptom weight acquisition module is specifically as follows: (1) acquire the data of the symptoms related to the diseases in the knowledge graph, and extract all the related knowledge with the knowledge type of disease-related symptom-symptom; (2) count the number of the diseases in the knowledge base N by removing the duplicate original words of the starting entities of all the related knowledge of disease-related symptom-symptom; (3) extract the symptom list of the knowledge base by removing the duplicate original words of the ending entities of all the related knowledge of disease-related symptom-symptom; (4) Extract the starting entity normalized term set corresponding to each symptom i in turn according to the symptom list of the knowledge base, and obtain the disease number associated with each symptom i by calculating the number of starting entity normalized terms ; (5) For the symptom list of the knowledge base, the number of diseases associated with each acquired symptom is sequentially obtained according to the total number of acquired diseases and the total number of acquired diseases to obtain a symptom-weight dictionary, as follows: ； wherein, a symptom coefficient as a case similarity degree; represents the number of diseases irrelevant to the symptom; N represents the total number of diseases in the knowledge base; The working process of the case similarity acquisition module is specifically as follows: (1) match the symptom and diagnosis lists of the target case with the symptom and diagnosis lists of the other cases in the case database, and obtain the intersection sets of the symptoms and diagnoses of the target case and the other cases, respectively; (2) acquire the symptom similarity according to the intersection set of the symptoms of the target case and the other cases; (3) acquire the diagnosis similarity according to the intersection set of the diagnoses of the target case and the other cases; (4) perform a weighted operation on the symptom similarity and the diagnosis similarity to obtain the case similarity degree; (5) store the basic information of the cases in a table form in the cache, with the case number, age, department, symptom similarity degree, diagnosis similarity degree and case similarity degree as the table headers; The symptom similarity is specifically as follows: Let the list of symptoms for the compared cases be [ The set of symptom intersections between the compared cases and the target case is [ ]. Symptom weighting is... The formula for calculating the corresponding case and the target case is as follows: ； wherein S represents a symptom similarity degree; a symptom weight of a symptom intersection set is accumulated to obtain an intersection score; represents a number of symptom list elements of a compared case; The diagnosis similarity is specifically as follows: Let the diagnosis list of the compared case be [ ]; the intersection set of the diagnosis of the compared case and the target case be [ ]; then the similarity degree of the corresponding case and the target case is calculated according to the following formula: ； wherein, represents the degree of similarity of symptoms; represents the number of diagnostic intersection sets; represents the number of diagnostic list elements of the compared cases; The case similarity degree is obtained by averaging the symptom similarity S and the diagnosis similarity S', and the formula is as follows: ； Wherein, F represents the case similarity degree.

4. An electronic device, comprising: The similar case recommendation method comprises the following steps: a memory and at least one processor; The memory has a computer program stored thereon; The at least one processor executes the computer program stored in the memory, so that the at least one processor executes the similar case recommendation method according to claim 1 or 2.

5. A computer-readable storage medium, characterized in that, The computer readable storage medium has a computer program stored therein, and the computer program can be executed by the processor to implement the similar case recommendation method according to claim 1 or 2.

Images