A traditional chinese medicine reproductive endocrine disease diagnosis auxiliary system based on clinical knowledge graph

Abstract

The application belongs to the technical field of disease auxiliary diagnosis, and particularly relates to a traditional Chinese medicine reproductive endocrine disease diagnosis auxiliary system based on a clinical knowledge graph. When a reproductive knowledge graph is constructed by integrating the medical records of reproductive disease patients, the collection of patient's pathogenic environment information is increased, and correlation analysis is carried out on the pathogenic environment to identify the risk pathogenic environment. Then, a search platform based on the reproductive knowledge graph is created. Thus, by inputting the patient's pathogenic symptoms and matching the reproductive knowledge graph, it is judged whether the matching is successful. When the matching is successful, the preferred treatment path matched with the patient's symptoms is output. When the matching fails, the patient's pathogenic environment information is input and matched with the reproductive knowledge graph, and then the potential reproductive disease of the patient is output. Thus, the ability of the reproductive knowledge graph in early warning and preventive intervention is improved, and the application value of the auxiliary diagnosis using the reproductive knowledge graph is greatly improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of disease auxiliary diagnosis, and specifically relates to a traditional Chinese medicine reproductive endocrine disease diagnosis auxiliary system based on a clinical knowledge graph. BACKGROUND

[0002] With the improvement of public health awareness, women pay more attention to their own health, especially the fast-paced lifestyle, high-pressure environment, environmental pollution and changes in dietary structure in modern society may adversely affect women's health, leading to an increasing trend in the incidence of certain reproductive endocrine diseases. In the diagnosis and treatment of reproductive endocrine diseases, traditional Chinese medicine has gradually become a popular choice with its unique herbal therapy and other non-invasive treatment methods.

[0003] In the current medical environment, in order to improve the accuracy and efficiency of traditional Chinese medicine diagnosis and treatment, especially in the case of insufficient experience of traditional Chinese medicine experts, the industry generally adopts the way of building a clinical knowledge graph to assist in traditional Chinese medicine diagnosis. Traditional Chinese medicine diagnosis mainly relies on the experience of doctors and the comprehensive analysis of the four diagnostic information (inspection, auscultation and olfaction, interrogation, and pulse-taking) of patients, and less uses modern medical examination methods. By establishing a knowledge graph containing information such as disease characteristics and treatment plans, a systematic reference tool can be provided for clinicians to help them better understand the disease and choose appropriate treatment methods, and possibly shorten the treatment period.

[0004] Most systems currently use clinical knowledge graphs for reproductive disease diagnosis mainly focus on the patient's symptoms and treatment methods, often ignoring the patient's disease environment (such as dietary structure, sleep quality, exercise habits, living conditions, etc.). In fact, the occurrence and development of reproductive endocrine disorders are closely related to these life and environmental factors. This one-sided focus makes the existing knowledge graph unable to fully capture the risk factors affecting the occurrence of the disease, limiting its ability in early warning and preventive intervention. This makes doctors only treat the symptoms after the patient has clear symptoms, and it is difficult to predict potential health risks in advance. SUMMARY

[0005] Therefore, the present application aims to provide a traditional Chinese medicine reproductive endocrine disease diagnosis auxiliary system based on a clinical knowledge graph, which collects and summarizes the patient's disease environment information when building the gynecological knowledge graph and identifies the risk disease environment, thereby deepening the use of the gynecological knowledge graph for disease auxiliary diagnosis, effectively solving the problems raised in the background art.

[0006] The objective of this invention can be achieved through the following technical solution: A diagnostic auxiliary system for TCM reproductive endocrine diseases based on clinical knowledge graph, comprising the following modules: a reproductive knowledge graph construction module, used to integrate patient consultation information from different hospitals to form a reproductive department consultation dataset, and construct a reproductive department knowledge graph from it.

[0007] The disease data description and statistics module is used to extract the symptoms, onset environment, and treatment pathways of various reproductive diseases from the constructed reproductive medicine knowledge graph.

[0008] The disease environment association analysis module is used to perform disease environment association analysis based on the disease environment of various reproductive diseases in order to identify risky disease environments.

[0009] The treatment effect analysis module is used to analyze the treatment effect based on the treatment path of patients with various reproductive diseases, in order to screen and select the best treatment path.

[0010] The auxiliary diagnostic module is used to create a search platform based on a reproductive knowledge graph. By inputting the patient's symptoms and matching them with the reproductive knowledge graph, the system judges whether the match is successful. When the match is successful, it outputs the optimal treatment path for the patient's symptoms. When the match fails, it inputs the patient's environmental information and matches it with the reproductive knowledge graph, thereby outputting the patient's potential reproductive diseases.

[0011] One improvement to the above technical solution is the construction of a reproductive knowledge graph, as described in the following process:

[0012] The medical records of each patient were retrieved from the reproductive health care dataset, and the symptoms, environment of onset, types of reproductive diseases, and treatment pathways were extracted from them.

[0013] Patients with the same type of reproductive disease were grouped into different patient groups based on their reproductive disease type.

[0014] After comparing the symptoms of each patient in the patient groups of various reproductive diseases and removing duplicates, a set of symptoms for each reproductive disease was formed.

[0015] The disease environment of each patient in the patient group of various reproductive diseases is used to form a disease environment set for each reproductive disease, and the frequency of occurrence of the same disease environment is marked in the set.

[0016] The treatment pathways for each patient in the patient group of various reproductive diseases are combined to form a set of treatment pathways for various reproductive diseases.

[0017] The sets of symptoms, environments, and treatment pathways of various reproductive diseases are considered as knowledge items for each type of reproductive disease.

[0018] For each reproductive disease, a node is created and assigned an identifier and attributes. Edges connecting the nodes are created based on the relationships in the knowledge items. Attribute information is added to the nodes and edges. The nodes and edges created in this way constitute a reproductive knowledge graph.

[0019] One improvement to the above technical solution is that the pathogenesis environment association analysis also includes an assessment of whether there is a pathogenesis environment association. The specific operation is as follows: retrieve the pathogenesis environment set of various reproductive diseases from the reproductive knowledge graph, and extract the frequency of occurrence of different pathogenesis environment information in the marked pathogenesis environment parameters.

[0020] By comparing the frequency of occurrence of different disease environment information in each disease environment parameter, the maximum and minimum occurrence frequencies are extracted. The difference between the maximum and minimum occurrence frequencies is then divided by the maximum occurrence frequency to obtain the degree of repetition difference of each disease environment parameter.

[0021] Compare the repeatability differences of each pathogenesis environment parameter with the set critical difference. If the repeatability differences of all pathogenesis environment parameters are less than the critical difference, it is judged that there is no pathogenesis environment association; otherwise, it is judged that there is a pathogenesis environment association.

[0022] One improvement to the above technical solution is that the process of identifying the risk environment for disease development is as follows:

[0023] When evaluating the existence of a disease-causing environment association, disease-causing environment parameters that repeatedly exhibit differences reaching the critical difference level are recorded as valid disease-causing environment parameters.

[0024] By comparing the frequency of occurrence of different disease-causing environment information in the effective disease-causing environment parameters, the disease-causing environment information with the highest frequency of occurrence is selected as the risk disease-causing environment.

[0025] One improvement to the above technical solution is that the analysis of the treatment effect is performed as follows:

[0026] The treatment pathways for various reproductive diseases are retrieved from the reproductive knowledge graph, and the treatment stages of each treatment pathway are marked. At the same time, the treatment duration and symptom indicators of each treatment stage are extracted.

[0027] The symptom indicators at each treatment stage are compared with those at the previous treatment stage to statistically analyze the degree of symptom improvement at each treatment stage.

[0028] The improvement in symptoms at each treatment stage, combined with the treatment duration, is substituted into the formula. Obtain the treatment efficacy coefficient for each treatment stage. ,in Indicates the degree of symptom improvement during the treatment phase. This indicates the duration of treatment during the treatment phase.

[0029] The treatment effect improvement rate of each treatment stage is calculated by subtracting the treatment effect coefficient of the previous treatment stage from the treatment effect coefficient of the previous treatment stage and then dividing by the treatment effect coefficient of the previous treatment stage.

[0030] The treatment effect improvement rate of each treatment stage along the treatment path is summed to obtain the treatment effect characterization value of the treatment path.

[0031] One improvement to the above technical solution is that the screening and selection of the preferred treatment path is performed as follows:

[0032] Patients with the same treatment path among the treatment paths for various reproductive diseases are categorized to obtain a number of patients corresponding to each treatment path, and the number of patients with each treatment path is counted.

[0033] The standard deviation of the treatment effect characterization value of each patient in the same treatment pathway is calculated and compared with the set limit standard deviation. If the standard deviation of the treatment effect characterization value of a patient in a certain treatment pathway is less than or equal to the limit standard deviation, the average value of the treatment effect characterization value of each patient in that treatment pathway is taken as the typical treatment effect characterization value of that treatment pathway. Otherwise, the minimum value of the treatment effect characterization value of each patient in that treatment pathway is taken as the typical treatment effect characterization value of that treatment pathway.

[0034] Substitute the typical treatment effect characteristics of each treatment pathway and the number of patients into the evaluation formula. The selection value of each treatment pathway was obtained. In the formula Indicates the treatment pathway number. , Indicates the first Typical therapeutic effect characterization values ​​for each treatment pathway Indicates the first The number of patients for whom a treatment pathway exists.

[0035] The selection value of each treatment path is compared, and the treatment path with the highest selection value is selected as the preferred treatment path.

[0036] Compared to existing technologies, the beneficial effects of this invention are as follows: In constructing a reproductive knowledge graph, this invention systematically integrates patient medical information and incorporates environmental factor data related to the onset of the disease. This allows for in-depth correlation analysis of this information to identify risk environments associated with disease development. Based on this enhanced knowledge graph, the search platform provides possible symptom suggestions and optimal treatment pathway recommendations by matching patient symptoms or onset environment information. This not only improves early warning capabilities but also enhances the effectiveness of preventative interventions, thereby significantly improving the application value and accuracy of knowledge graph-assisted diagnosis. Attached Figure Description

[0037] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a schematic diagram showing the connections of the various modules in the system of the present invention.

[0039] Figure 2 This is a schematic diagram illustrating the construction of the reproductive knowledge graph in this invention. Detailed Implementation

[0040] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] See Figure 1 As shown, this invention proposes a diagnostic auxiliary system for TCM reproductive endocrine diseases based on clinical knowledge graphs, including a reproductive knowledge graph construction module, a disease data description and statistics module, a pathogenesis environment association analysis module, a treatment effect analysis module, and an auxiliary diagnostic module. The reproductive knowledge graph construction module is connected to the disease data description and statistics module, the disease data description and statistics module is connected to the pathogenesis environment association analysis module and the treatment effect analysis module, and the pathogenesis environment association analysis module and the treatment effect analysis module are connected to the auxiliary diagnostic module.

[0042] The reproductive knowledge graph construction module is used to integrate patient visit information from different hospitals to form a reproductive visit dataset, and to construct a reproductive knowledge graph from it.

[0043] To apply the above scheme, the process of constructing a reproductive knowledge graph is as follows: retrieve the medical information of each patient from the reproductive medical treatment dataset, and extract the symptoms, the environment in which the disease occurred, the types of reproductive diseases, and the treatment path.

[0044] It's important to understand that when constructing a reproductive knowledge graph, patient information is primarily retrieved from patient medical records. Symptoms refer to the discomfort or abnormalities described by the patient during their visit, such as irregular bleeding, lower abdominal pain, or abnormal vaginal discharge. The environment in which the disease develops typically refers to external conditions or background that may influence its occurrence. In the reproductive field, this can include lifestyle habits (such as dietary habits and exercise frequency) and living environment (such as hygiene conditions in the place of residence). The environment can be obtained by inquiring about the patient's daily habits. Types of reproductive diseases refer to specific disease names or diagnoses, such as uterine fibroids, ovarian cysts, and cervicitis. Treatment pathways refer to the phased treatment measures taken for a specific disease, which may include drug therapy and physical therapy.

[0045] Patients with the same type of reproductive disease were grouped into different patient groups based on their reproductive disease type.

[0046] After comparing the symptoms of each patient in various reproductive disease groups and removing duplicates, a set of symptoms for each reproductive disease was formed. This ensured that the key symptoms of each reproductive disease were accurately identified and summarized.

[0047] The disease-causing environments of each patient in various reproductive disease groups were grouped into disease-causing environment sets for each disease, and the frequency of occurrence of the same disease-causing environment was marked in the sets. By marking the frequency, the prevalence of each environmental factor can be intuitively understood, providing a basis for further analysis.

[0048] The treatment pathways for each patient in the patient group of various reproductive diseases are combined to form a set of treatment pathways for various reproductive diseases.

[0049] The sets of symptoms, environments, and treatment pathways of various reproductive diseases are considered as knowledge items for each type of reproductive disease.

[0050] For each reproductive disease, a node is created and assigned an identifier and attributes (such as reproductive disease type). Edges connecting the nodes are then created based on the relationships in the knowledge items. Attribute information is added to both the nodes and edges. The resulting nodes and edges constitute a reproductive knowledge graph. (See [link to relevant documentation]). Figure 2 As shown.

[0051] It should be added that the constructed reproductive knowledge graph can be visually displayed using visualization tools.

[0052] The disease data description and statistics module is used to extract the symptoms, onset environment, and treatment pathways of various reproductive diseases from the constructed reproductive knowledge graph.

[0053] The disease environment association analysis module is used to perform disease environment association analysis based on the disease environment of various reproductive diseases in order to identify risky disease environments.

[0054] In the above-mentioned approach, before conducting the disease environment association analysis, it is necessary to determine whether there is a disease environment association. The specific operation is as follows: retrieve the disease environment set of various reproductive diseases from the reproductive knowledge graph, and extract the frequency of occurrence of different disease environment information in the marked disease environment parameters.

[0055] By comparing the frequency of occurrence of different disease-related environmental information among various disease-related environmental parameters, the maximum and minimum frequencies of occurrence are extracted. The difference between the maximum and minimum frequencies is then divided by the maximum frequency to obtain the degree of repetition difference for each disease-related environmental parameter. Calculating the degree of repetition can quantify the distributional differences between different disease-related environmental information, helping to identify which environmental factors show significant differences in the patient population.

[0056] The repetition rate of each disease-causing environment parameter is compared with the set critical rate of difference. For example, the critical rate of difference is 0.6. If the repetition rate of all disease-causing environment parameters is less than the critical rate of difference, it is judged that there is no disease-causing environment association. Otherwise, it is judged that there is a disease-causing environment association.

[0057] It's important to understand that the occurrence of reproductive diseases is often related to the patient's living environment. This is because various lifestyle and environmental factors can directly or indirectly affect women's reproductive health. Specifically: Dietary habits: Unhealthy diets (such as high-fat, high-sugar, low-fiber diets) can lead to obesity, increasing the risk of polycystic ovary syndrome (PCOS), endometriosis, and other diseases. Exercise habits: Lack of exercise can lead to weight gain and metabolic disorders, increasing the likelihood of endocrine-related diseases. Conversely, excessive exercise can cause menstrual irregularities and ovulation problems, affecting fertility. Regular sleep patterns: Long-term sleep deprivation and staying up late can disrupt the body's biological clock, leading to hormonal imbalances, affecting the menstrual cycle and ovarian function, and increasing the risk of reproductive diseases.

[0058] The recurrence variability of the aforementioned environmental parameters can be understood as a measure of the frequency or pattern of disease occurrence under a certain environmental condition, and a threshold (i.e., critical variability) is set to determine whether the environmental condition may be associated with the disease.

[0059] When the repeatability of all environmental factors affecting disease incidence is less than the critical variability, it means that the incidence of disease under that environmental condition does not change significantly across different observations or experiments, or is relatively stable without a clear increasing trend. In this case, researchers may conclude that there is no direct causal relationship between this specific environmental condition and the disease under study, or that its influence is very weak. Conversely, if the repeatability of certain environmental parameters affects disease incidence exceeds the critical variability, it indicates that the incidence of disease under that environmental condition fluctuates significantly across different time points or different samples, potentially suggesting that this environmental condition has some influence on the occurrence of the disease.

[0060] However, it is important to note that the above explanation applies only to scientific research and data analysis. In actual clinical practice, doctors do not rely solely on such statistical data to make diagnostic or treatment decisions.

[0061] In a further feasible way of the above scheme, the identification of risky disease environments is carried out as follows: when judging the existence of disease environment association, disease environment parameters that repeatedly appear with a difference degree reaching the critical difference degree are recorded as valid disease environment parameters.

[0062] By comparing the frequency of occurrence of different environmental factors among the effective environmental parameters for disease incidence, the environmental factor with the highest frequency of occurrence is selected as the risk environmental factor. Selecting the environmental factor with the highest frequency of occurrence as the risk factor conforms to the "majority principle," meaning that the most common environmental factor is more likely to be the true risk factor.

[0063] The treatment effect analysis module is used to analyze the treatment effect based on the treatment path of patients with various reproductive diseases, so as to screen the optimal treatment path.

[0064] Preferably, the treatment effect analysis follows this process: Retrieve a set of treatment pathways for various reproductive diseases from a reproductive knowledge graph, mark the treatment stages of each pathway, and extract the treatment duration and symptom indicators for each stage. By marking the treatment stages, the unstructured treatment process can be transformed into structured data, facilitating subsequent quantitative analysis.

[0065] The symptom indicators at each treatment stage were compared with those at the previous treatment stage to statistically analyze the degree of symptom improvement at each treatment stage. .

[0066] The improvement in symptoms at each treatment stage, combined with the treatment duration, is substituted into the formula. Obtain the treatment efficacy coefficient for each treatment stage. ,in Indicates the degree of symptom improvement during the treatment phase. This indicates the duration of each treatment phase. Introducing treatment duration allows for a comprehensive assessment of the efficiency of each treatment phase, considering not only the degree of improvement in the condition but also the time cost.

[0067] The treatment effect improvement rate of each treatment stage is calculated by subtracting the treatment effect coefficient of the previous treatment stage from the treatment effect coefficient of the previous treatment stage and then dividing by the treatment effect coefficient of the previous treatment stage.

[0068] The treatment effect improvement rate of each treatment stage along the treatment path is summed to obtain the treatment effect characterization value of the treatment path.

[0069] It is important to know that by calculating the treatment effect improvement rate, we can capture the improvement of each treatment stage relative to the previous node, reflect the dynamic changes in the treatment process, and use the treatment effect coefficient of the previous node as the denominator to make the improvement rate value between 0 and 1, which is convenient for subsequent cumulative calculation. The cumulative method can reflect the cumulative improvement effect of each node in the treatment process and help identify the most effective treatment path.

[0070] Further preferably, the selection of preferred treatment pathways involves the following steps: Patients sharing the same treatment pathway from the treatment pathway sets for various reproductive diseases are categorized into several patients corresponding to each treatment pathway, and the number of patients associated with each treatment pathway is counted. This allows us to understand the frequency and prevalence of the treatment pathway, providing a basis for subsequent evaluation.

[0071] The standard deviation of the treatment effect characterization value of each patient in the same treatment pathway is calculated and compared with the set limit standard deviation. If the standard deviation of the treatment effect characterization value of a patient in a certain treatment pathway is less than or equal to the limit standard deviation, the average value of the treatment effect characterization value of each patient in that treatment pathway is taken as the typical treatment effect characterization value of that treatment pathway. Otherwise, the minimum value of the treatment effect characterization value of each patient in that treatment pathway is taken as the typical treatment effect characterization value of that treatment pathway.

[0072] The above-mentioned calculation of standard deviation can assess the consistency of the efficacy of the same treatment pathway across different patients. A smaller standard deviation indicates that the treatment pathway is more consistent in efficacy for most patients, suggesting higher reliability. When the standard deviation is large, selecting the minimum value as the typical efficacy representative value is a conservative approach, ensuring that the treatment pathway still has some efficacy even under the most unfavorable circumstances. This approach considers both the consistency of the treatment pathway and the minimum efficacy under extreme conditions, offering good flexibility and robustness.

[0073] Substitute the typical treatment effect characteristics of each treatment pathway and the number of patients into the evaluation formula. The selection value of each treatment pathway was obtained. In the formula Indicates the treatment pathway number. , Indicates the first Typical therapeutic effect characterization values ​​for each treatment pathway Indicates the first The number of patients for whom a treatment pathway exists.

[0074] The selection value of each treatment path is compared, and the treatment path with the highest selection value is selected as the preferred treatment path.

[0075] The above calculation method can find a balance between efficacy and accessibility, ensuring that the selected treatment path is not only effective but also applicable to a large number of patients. By comparing the selection value of each treatment path, the treatment path with the best overall effect and the widest applicability can be selected as the preferred option in clinical practice.

[0076] The auxiliary diagnostic module is used to create a search platform based on a reproductive knowledge graph. By inputting the patient's symptoms and matching them with the reproductive knowledge graph, the system judges whether the match is successful. When the match is successful, it outputs the optimal treatment path for the patient's symptoms. When the match fails, it inputs the patient's environmental information and matches it with the reproductive knowledge graph, thereby outputting the patient's potential reproductive diseases.

[0077] Specifically, the process for determining whether a match is successful is as follows: the patient's symptoms are matched with the set of symptoms for various reproductive diseases in the reproductive knowledge graph. If a certain symptom falls within the set of symptoms for a certain reproductive disease, then that reproductive disease is taken as the primary reproductive disease for initial matching.

[0078] The number of initial reproductive diseases resulting from patient symptoms during matching is counted, and the percentage of patients with symptoms matched for each initial reproductive disease is obtained. This percentage is then compared with the set effective percentage. If the percentage of patients with symptoms matched for an initial reproductive disease reaches the effective percentage, the match is considered successful; otherwise, the match is considered unsuccessful.

[0079] More specifically, when a match is successful, the operation of outputting the optimal treatment path for the patient's symptoms is as follows: count the number of successfully matched first-time reproductive diseases. If there is only one successfully matched first-time reproductive disease, extract the optimal treatment path corresponding to the corresponding reproductive disease from the reproductive knowledge graph. If there is more than one successfully matched first-time reproductive disease, extract the optimal treatment path corresponding to each first-time reproductive disease from the reproductive knowledge graph and arrange them in descending order of the proportion of the patient's symptoms corresponding to the first-time reproductive disease.

[0080] Furthermore, when a match fails, the patient's disease environment information is entered and matched with the reproductive knowledge graph. See the following process: Arrange the various reproductive diseases existing in the reproductive knowledge graph in order of priority for first-time couples.

[0081] According to the order of reproductive diseases, the patient's disease-causing environment information is matched with the risk-causing environment of the corresponding reproductive disease in the reproductive knowledge graph. If the patient's disease-causing environment information is consistent with the risk-causing environment of a certain reproductive disease, then that reproductive disease is identified as a predisposing reproductive disease.

[0082] Further, the output of the patient's potential reproductive diseases is implemented as follows: count the number of potential reproductive diseases. If there is only one potential reproductive disease, then that reproductive disease is taken as the patient's potential reproductive disease. If there is more than one potential reproductive disease, then compare the patient's disease environment information with the disease environment set corresponding to each potential reproductive disease in the reproductive knowledge graph, count the number of patient disease environment information that falls within the disease environment set for each potential reproductive disease, and then take the potential reproductive disease corresponding to the largest number of occurrences as the patient's potential reproductive disease.

[0083] This invention, in constructing a reproductive knowledge graph, systematically integrates patient medical information and incorporates environmental factor data related to the onset of the disease. Through in-depth correlation analysis of this information, it identifies risk environments associated with disease development. Based on this enhanced knowledge graph, a search platform provides possible symptom suggestions and optimal treatment pathways by matching patient symptoms or onset environment information. This not only improves early warning capabilities but also enhances the effectiveness of preventative interventions, thereby significantly improving the application value and accuracy of knowledge graph-assisted diagnosis.

[0084] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A traditional Chinese medicine reproductive endocrine disease diagnosis auxiliary system based on a clinical knowledge graph, characterized by , comprising the following modules: a reproductive knowledge graph construction module for integrating patient visit information from different hospitals to form a reproductive visit dataset and constructing a reproductive knowledge graph therefrom; a disease data descriptive statistics module for extracting the onset symptoms, onset environment, and treatment path of each reproductive condition from the constructed reproductive knowledge graph; an onset environment correlation analysis module for performing onset environment correlation analysis based on the onset environment of each reproductive condition to identify a risk onset environment; a treatment effect analysis module for performing treatment effect analysis based on the treatment path of each reproductive condition corresponding to a patient to screen an optimal treatment path; an auxiliary diagnosis module for creating a search platform based on the reproductive knowledge graph, whereby the patient's onset symptoms are inputted and matched with the reproductive knowledge graph to determine whether the matching is successful, and when the matching is successful, the optimal treatment path matching the patient's symptoms is outputted, and when the matching fails, the patient's onset environment information is inputted and matched with the reproductive knowledge graph, and then the patient's potential reproductive condition is outputted; the construction of the reproductive knowledge graph refers to the following process: the visit information of each patient is retrieved from the reproductive visit dataset, and the onset symptoms, onset environment, reproductive condition type, and treatment path are extracted therefrom; the reproductive condition types of each patient are classified into the same condition type to form patient groups of each reproductive condition; the onset symptoms of each patient in the patient groups of each reproductive condition are compared and de-duplicated to form the onset symptom set of each reproductive condition; the onset environment of each patient in the patient groups of each reproductive condition is formed into the onset environment set of each reproductive condition, and the occurrence frequency of the same onset environment is marked in the set; the treatment path of each patient in the patient groups of each reproductive condition is formed into the treatment path set of each reproductive condition; the onset symptom set, onset environment set, and treatment path set of each reproductive condition are taken as the knowledge items of each reproductive condition; a node is created for each reproductive condition, and an identifier and attributes are assigned thereto, and edges connecting the nodes are created according to the relationships in the knowledge items, and attribute information is added to the nodes and edges, whereby the nodes and edges constitute the reproductive knowledge graph; the onset environment correlation analysis further includes the judgment of whether there is an onset environment correlation, and the specific operation is as follows: the onset environment set of each reproductive condition is retrieved from the reproductive knowledge graph, and the occurrence frequency of different onset environment information in each onset environment parameter is extracted; the occurrence frequency of different onset environment information in each onset environment parameter is compared, and the maximum occurrence frequency and the minimum occurrence frequency are extracted therefrom, and the difference between the maximum occurrence frequency and the minimum occurrence frequency is divided by the maximum occurrence frequency to obtain the repeated occurrence difference of each onset environment parameter; the repeated occurrence difference of each onset environment parameter is compared with the set critical difference, and if the repeated occurrence difference of all onset environment parameters is less than the critical difference, it is judged that there is no onset environment correlation, otherwise it is judged that there is an onset environment correlation; the identification of the risk onset environment is as follows: when it is judged that there is an onset environment correlation, the onset environment parameter whose repeated occurrence difference reaches the critical difference is recorded as an effective onset environment parameter; Compare the occurrence frequencies of different onset environment information in the effective onset environment parameters, and select the onset environment information with the maximum occurrence frequency as the risk onset environment; The treatment effect analysis is described as follows: Retrieve the treatment path set of various reproductive disorders from the reproductive knowledge graph, and mark the treatment stages of each treatment path. Meanwhile, extract the treatment duration and disorder indicators of each treatment stage; Compare the disorder indicators of each treatment stage with those of the previous treatment stage to calculate the disorder improvement degree of each treatment stage; combining the degree of improvement of the condition of each treatment phase with the length of treatment obtaining a treatment effect coefficient for each treatment phase , wherein denotes the degree of improvement of the condition of the treatment phase, denotes the length of treatment of the treatment phase; Calculate the treatment effect improvement rate of each treatment stage by taking the difference between the treatment effect coefficient of each treatment stage and that of the previous treatment stage, and dividing the treatment effect coefficient of the previous treatment stage; Cumulatively add the treatment effect improvement rate of each treatment stage in the treatment path to obtain the treatment effect representation value of the treatment path; The preferred treatment path is selected as follows: Classify the patients with the same treatment path in the treatment path set of various reproductive disorders to obtain a number of patients corresponding to each treatment path, and count the number of patients existing in each treatment path; Calculate the standard deviation of the treatment effect representation values of the patients in the same treatment path, and compare it with the set limited standard deviation. If the standard deviation of the treatment effect representation values of the patients in a treatment path is less than or equal to the limited standard deviation, take the average value of the treatment effect representation values of the patients in the treatment path as the typical treatment effect representation value of the treatment path, otherwise take the minimum value of the treatment effect representation values of the patients in the treatment path as the typical treatment effect representation value of the treatment path; Substitute the typical treatment effect characteristics of each treatment pathway and the number of patients into the evaluation formula. The selection value of each treatment pathway was obtained. In the formula Indicates the treatment pathway number. , Indicates the first Typical therapeutic effect characterization values ​​for each treatment pathway Indicates the first The number of patients for whom each treatment pathway exists; Compare the selection value degrees of each treatment path, and select the treatment path with the maximum selection value degree as the preferred treatment path.

2. The traditional Chinese medicine reproductive endocrine disease diagnosis auxiliary system based on a clinical knowledge graph according to claim 1, characterized in that: The judgment of whether the matching is successful is as follows: Match the onset symptoms of the patient with the onset symptom set of various reproductive disorders in the reproductive knowledge graph. If an onset symptom falls within the onset symptom set of a reproductive disorder, take the reproductive disorder as the primary matching reproductive disorder; Count the number of primary matching reproductive disorders formed in the matching of the patient's onset symptoms, and obtain the proportion of the patient's onset symptoms matched by each primary matching reproductive disorder, and then compare it with the set effective proportion. If the proportion of the patient's onset symptoms matched by a primary matching reproductive disorder reaches the effective proportion, the matching is judged to be successful, otherwise the matching is judged to be failed.

3. The traditional Chinese medicine reproductive endocrine disease diagnosis auxiliary system based on a clinical knowledge graph according to claim 2, characterized in that: When the matching is successful, the preferred treatment path matched by the patient's symptoms is output as follows: Count the number of primary matching reproductive disorders that are successful in matching. If there is only one primary matching reproductive disorder that is successful in matching, extract the preferred treatment path corresponding to the corresponding reproductive disorder from the reproductive knowledge graph. If there is more than one primary matching reproductive disorder that is successful in matching, extract the preferred treatment path corresponding to each primary matching reproductive disorder from the reproductive knowledge graph, and arrange them in the order of the proportion of the patient's onset symptoms corresponding to the primary matching reproductive disorder from large to small.

4. The traditional Chinese medicine reproductive endocrine disease diagnosis auxiliary system based on a clinical knowledge graph according to claim 3, characterized in that: When the matching fails, the patient's onset environment information is matched with the reproductive knowledge graph as follows: Arrange the various reproductive disorders existing in the reproductive knowledge graph in the order of priority of the primary matching reproductive disorder; According to the arrangement order of the reproductive disorders, the patient's onset environment information is matched with the risk onset environment of the corresponding reproductive disorder in the reproductive knowledge graph. If the patient's onset environment information is consistent with the risk onset environment of a reproductive disorder, the reproductive disorder is regarded as a predisposed reproductive disorder.

5. The traditional Chinese medicine reproductive endocrine disease diagnosis auxiliary system based on a clinical knowledge graph according to claim 4, characterized in that: The output of the patient's potential reproductive disorder is implemented as follows: The number of predisposed reproductive disorders is counted. If there is only one predisposed reproductive disorder, the predisposed reproductive disorder is regarded as the patient's potential reproductive disorder. If there is more than one predisposed reproductive disorder, the patient's onset environment information is compared with the corresponding onset environment set of each predisposed reproductive disorder in the reproductive knowledge graph. The number of patient onset environment information falling within the onset environment set in each predisposed reproductive disorder is counted, and then the predisposed reproductive disorder corresponding to the maximum falling number is taken as the patient's potential reproductive disorder.

Images