Internet Intelligent Medical Consultation System
Through the Internet intelligent medical consultation system, the department affiliation analysis module and guided problem screening are used to solve the problems of low diagnosis and treatment efficiency and inaccurate department matching in the traditional hospital consultation model, and improve the accuracy of department recommendations and patient experience.
Patent Information
- Application Number
- CN202411408155.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-10
- Publication Date
- 2024-12-20
- Estimated Expiration
- 2044-10-10
AI Technical Summary
In the traditional hospital visit model, the inefficient diagnosis and treatment efficiency and inaccurate department matching lead to poor patient visit experience.
An Internet intelligent medical consultation system was designed to obtain patient complaints and medical records through the data collection module. The department affiliate analysis module made initial membership judgment based on the similarity between the patient complaints and the medical records, and screened and updated the membership by guiding the problem until Recommend the department to be selected.
It improves the accuracy of department recommendations, improves the patient's medical experience, and distinguishes the affiliation of different departments by screening and updating problems, reduces the matching of multiple departments, and enhances the robustness of recommendations.
Smart Images
Figure CN118919050B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of medical services, and in particular to an Internet intelligent medical consultation system. Background Art
[0002] With the increasing demand for medical care and the aging population, problems such as uneven matching of medical resources and low efficiency of diagnosis and treatment have become increasingly prominent. The traditional hospital treatment model has problems such as long waiting time between clinics and irregular medical guide services, which bring great pressure on patients. Therefore, it is necessary to improve hospital treatment efficiency and medical quality through intelligent means.
[0003] The medical consultation system automatically analyzes the symptom descriptions entered by patients through natural language processing, machine learning and other technologies, provides preliminary diagnostic suggestions, and intelligently matches the appropriate department, thereby optimizing the patient's medical experience. Patients describe or upload their own chief complaints, that is, the patient's condition, in the medical consultation system. The system compares the patient's chief complaint with the medical records in the hospital database, and recommends the outpatient department to the patient whose medical records are more similar to the patient's chief complaint, so as to improve the patient's medical efficiency.
[0004] When analyzing the similarity between the patient's chief complaint and medical record, since the patient's chief complaint may contain ambiguous information and the characteristic manifestations of the disease may have certain similarities, the similarity between the medical records of different departments and the patient's chief complaint may be high, which may reduce the accuracy of the recommendation for the patient's treatment department and affect the patient's treatment experience. Summary of the invention
[0005] In order to solve the technical problem that the number of departments with a high degree of matching with patients is large due to the diversity of diseases and unclear description of patients' main complaints, resulting in a low accuracy rate of recommending patients' treatment departments, the purpose of the present invention is to provide an Internet intelligent medical consultation system, and the technical solutions adopted are as follows:
[0006] The present invention proposes an Internet intelligent medical consultation system, the system comprising:
[0007] A data collection module is used to obtain the patient's chief complaint, several medical records and guiding questions; the medical records have corresponding departments;
[0008] The department affiliation analysis module is used to determine the initial affiliation of the patient to each department based on the similarity between the medical records of the same department and the patient's chief complaint; the initial guidance set is formed by all the guidance questions; when the difference between the patient's affiliation to different departments does not meet the selection condition, the preferred guidance questions are selected from the guidance set according to the similarity between the guidance questions in the guidance set and the patient's chief complaint and the particularity of each guidance question, and then the guidance set is updated; the patient's reply text to the preferred guidance question is obtained, and the patient's affiliation to each department is updated according to the similarity between the reply text and the medical records of the same department, until the updated patient's affiliation to each department meets the selection condition;
[0009] The treatment department recommendation module is used to recommend the patient's treatment department based on the patient's affiliation to each department when the selection conditions are met.
[0010] Furthermore, the determining of the patient's initial affiliation to each department based on the similarity between the medical records of the same department and the patient's chief complaint includes:
[0011] Obtain textual similarity metrics between the patient’s chief complaint and each medical record;
[0012] The maximum value of the text similarity index between all medical records of each department and the patient's chief complaint is used as the patient's initial membership to each department.
[0013] Furthermore, the obtaining of the text similarity index between the patient's chief complaint and each medical record includes:
[0014] All medical records and patients' chief complaints are recorded as analysis texts, and the analysis texts are segmented and converted into word vectors. All word vectors of the analysis texts constitute a word vector set.
[0015] Select any word vector from the word vector set of the patient's chief complaint and record it as the target word vector, obtain the vector similarity index between the target word vector and each word vector in the word vector set of each medical record, traverse the word vectors in the word vector set of each medical record in a preset order, and stop traversing when the vector similarity index is greater than a preset similarity threshold, and use the vector similarity index when traversal stops as the local similarity index between the target word vector and each medical record;
[0016] If the vector similarity indexes of the target word vector and all the word vectors in the word vector set of each medical record are less than or equal to the preset similarity threshold, the maximum value of the vector similarity indexes of the target word vector and all the word vectors in the word vector set of each medical record is taken as the local similarity index of the target word vector and each medical record;
[0017] The average of the local similarity indexes of all word vectors in the word vector set of the patient's chief complaint and each medical record is used as the text similarity index between the patient's chief complaint and each medical record.
[0018] Furthermore, the candidate conditions include:
[0019] The absolute value of the difference between the two largest values of the patient's membership of all departments is normalized to obtain a clear indicator of department division;
[0020] The candidate condition is: the department division clarity index is greater than a preset clarity threshold, or the number of updates of the patient's affiliation to each department is equal to a preset number threshold.
[0021] Furthermore, selecting the preferred guiding question from the guiding set according to the similarity between the guiding questions in the guiding set and the patient's main complaint and the particularity of each guiding question includes:
[0022] Selecting a guiding question from the guiding set as the analysis question, selecting a word vector from the word vector set of the analysis question as the guiding word vector, and obtaining a local special index of the guiding word vector according to the similarity between the guiding word vector and the word vectors in the word vector set of the remaining guiding questions in the guiding set except the analysis question;
[0023] The maximum value of the vector similarity indexes of the guide word vector and all the word vectors in the word vector set of the patient's chief complaint is recorded as the problem chief complaint similarity index of the guide word vector;
[0024] According to the problem complaint similarity index and the local special index, a local judgment index of the guide word vector is obtained; the cumulative sum of the local judgment indexes of all word vectors in the word vector set of the analysis problem is used as a screening judgment index for the analysis problem;
[0025] The guiding question corresponding to the maximum value of the screening judgment index of all the guiding questions in the guiding set is taken as the preferred guiding question.
[0026] Furthermore, the obtaining of the local special index of the guide word vector includes:
[0027] A control guide set of guide word vectors is formed by all word vectors in the word vector set of the remaining guide questions in the guide set except the analysis question;
[0028] The maximum value of the vector similarity indexes between the guide word vector and all the word vectors in the control guide set is used as the first similarity index of the guide word vector;
[0029] Obtaining the mean of the vector similarity indexes of the guide word vector and all the word vectors in the control guide set, and recording it as the second similarity index of the guide word vector;
[0030] According to the first similarity index and the second similarity index, a local special index of the guide word vector is obtained; the first similarity index and the second similarity index are both negatively correlated with the local special index.
[0031] Furthermore, updating the patient's affiliation to each department according to the similarity between the reply text and the medical records of the same department includes:
[0032] Obtaining the text similarity index between the response text of the preferred guiding question and each medical record;
[0033] The maximum value of the text similarity index between all the medical records of each department and the answer texts of the preferred guiding questions is used as the patient's updated membership to each department.
[0034] Further, the updating of the guide set includes:
[0035] The guidance set after deleting the preferred guidance question is used as the updated guidance set.
[0036] Furthermore, the vector similarity index is the cosine similarity between two word vectors.
[0037] Furthermore, the preset clarity threshold is 0.75.
[0038] The present invention has the following beneficial effects:
[0039] In an embodiment of the present invention, the degree of match between the patient's chief complaint and the department, i.e., the patient's affiliation to the department, is preliminarily determined by the similarity between the medical records of the same department and the patient's chief complaint; since different diseases may have similar clinical manifestations, the similarity between the medical record information of different departments may be high, so that the patient's affiliation to different departments is relatively close. In order to distinguish the patient's affiliation to different departments, the preferred guiding questions with a high degree of relevance to the patient's chief complaint are screened from the guiding questions. The similarity between the guiding question and the patient's chief complaint measures the match between the guiding question and the patient's disease, and is analyzed in combination with the particularity of the guiding question. The patient's affiliation to different departments is distinguished by fewer and more accurate questions, thereby obtaining the preferred guiding question. The preferred guiding question is asked to the patient through the medical consultation system, and the patient's reply text to the preferred guiding question is obtained. The similarity between the reply text and the medical record of each department can deeply analyze the patient's disease manifestation. Using this similarity to update the patient's affiliation to each department can effectively improve the accuracy of the recommendation of the patient's medical department and enhance the patient's medical experience. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0041] Figure 1 A system structure diagram of an Internet intelligent medical consultation system provided by an embodiment of the present invention;
[0042] Figure 2 A system structure diagram of a method for updating a patient's degree of affiliation to a department provided by an embodiment of the present invention;
[0043] Figure 3 A schematic diagram of a computer device of an Internet intelligent medical consultation device provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0044] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation method, structure, features and effects of an Internet intelligent medical consultation system proposed by the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0045] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
[0046] The specific scheme of an Internet intelligent medical consultation system provided by the present invention is described in detail below with reference to the accompanying drawings.
[0047] Embodiment 1:
[0048] See also Figure 1 , which shows a system block diagram of an Internet intelligent medical consultation system provided by an embodiment of the present invention, the system includes: a data collection module 110, a department affiliation analysis module 120, and a consultation department recommendation module 130.
[0049] The data collection module 110 is used to obtain the patient's chief complaint, several medical records and guiding questions; the medical records have corresponding departments.
[0050] Specifically, patients conduct smart medical consultations through Internet devices. Patients input their main complaints of their own diseases into the medical consultation system, and obtain the main complaints entered by the patients in the medical consultation system and the medical records of different departments in the hospital database. By analyzing the similarity between the patient's main complaints and the medical records of all departments, the department to which the medical record with the greatest similarity belongs is usually recommended to the patient, thereby achieving the recommendation of the patient's treatment department.
[0051] However, different diseases may have similar clinical manifestations, that is, the similarity between different medical records is high, resulting in a high similarity between the patient's chief complaint and multiple medical records. The similarity between the patient's chief complaint and the medical record alone cannot accurately recommend the patient's treatment department. This solution additionally considers guiding questions, that is, problems in the medical record. For example, guiding questions about fever can be symptoms related to fever, such as whether the throat is sore, whether there is a runny nose or nasal congestion; guiding questions about allergies can be symptoms related to allergies, such as whether the skin is itchy, whether breathing is difficult, etc. Guiding questions are set in advance based on the medical records in the hospital database, and the patient's treatment department is further recommended based on the patient's response to the guiding questions.
[0052] The department affiliation analysis module 120 is used to determine the patient's initial affiliation to each department based on the similarity between the medical records of the same department and the patient's chief complaint; an initial guidance set is formed by all guidance questions; when the difference between the patient's affiliations to different departments does not meet the selection condition, based on the similarity between the guidance questions in the guidance set and the patient's chief complaint and the particularity of each guidance question, the preferred guidance questions are selected from the guidance set, and then the guidance set is updated; the patient's reply text to the preferred guidance question is obtained, and the patient's affiliation to each department is updated based on the similarity between the reply text and the medical records of the same department, until the patient's updated affiliation to each department meets the selection condition.
[0053] The similarity between the medical records of the same department and the patient's chief complaint is used to preliminarily judge the degree of match between the patient's chief complaint and the department, that is, the patient's affiliation to the department. Since different diseases may have similar clinical manifestations, the similarity of medical records in different departments may be high, making the patient's affiliation to different departments relatively close. In order to distinguish the patient's affiliation to different departments, the preferred guidance question is obtained. The preferred guidance question is asked to the patient through the medical consultation system to obtain the patient's response to the preferred guidance question; then, based on the similarity between the patient's response to the preferred guidance question and the medical record of each department, the patient's affiliation to the department is updated, and then the patient's treatment department is recommended.
[0054] The patient's chief complaint and medical records are both text data. In order to analyze the similarity between the chief complaint and the medical record, all medical records and the patient's chief complaint are recorded as analysis texts. The analysis texts are segmented by the Jieba word segmentation algorithm, and the word segmentation results of the analysis texts are converted into word vectors using the Word2Vec algorithm. All word vectors of the analysis texts constitute a word vector set. The Jieba word segmentation algorithm and the Word2Vec algorithm are both well-known technologies for those skilled in the art and will not be described in detail here.
[0055] See also Figure 2 , which shows a system structure diagram of a method for updating affiliation of a patient to a department provided by an embodiment of the present invention. The system includes: an affiliation initial analysis unit 121, a guiding question screening unit 122, and an affiliation updating unit 123.
[0056] The initial affiliation analysis unit 121 determines the initial affiliation of the patient to each department according to the similarity between the medical records of the same department and the patient's chief complaint.
[0057] The similarity between the patient's chief complaint and each medical record is measured by the text similarity index. The similarity between the medical records of the same department and the patient's chief complaint is used to analyze the matching degree between the patient's chief complaint and the department, that is, the patient's affiliation to the department.
[0058] Preferably, in some possible implementations of the present embodiment, the method for obtaining the text similarity index includes: selecting any word vector from the word vector set of the patient's chief complaint as the target word vector, obtaining the vector similarity index between the target word vector and each word vector in the word vector set of each medical record, traversing the word vectors in the word vector set of each medical record in a preset order, and stopping the traversal until the vector similarity index is greater than a preset similarity threshold, and using the vector similarity index when the traversal is stopped as the local similarity index between the target word vector and each medical record; if the vector similarity index between the target word vector and all the word vectors in the word vector set of each medical record is less than or equal to the preset similarity threshold, then using the maximum value of the vector similarity index between the target word vector and all the word vectors in the word vector set of each medical record as the local similarity index between the target word vector and each medical record; using the average of the local similarity indexes between all the word vectors in the word vector set of the patient's chief complaint and each medical record as the text similarity index between the patient's chief complaint and each medical record.
[0059] Because the patient's chief complaint and medical record both correspond to multiple word vectors, when calculating the text similarity index between the patient's chief complaint and the medical record, it is necessary to combine the word vectors of the chief complaint and the medical record in pairs, but multiple combinations are not conducive to analyzing the overall similarity between the chief complaint and the medical record. Therefore, this embodiment is based on the word vector of the patient's chief complaint, and performs similarity analysis on it with the word vectors in the word vector set of the medical record, obtains the local similarity index of each word vector of the patient's chief complaint and the medical record, and comprehensively analyzes the overall level of the local similarity index of all word vectors of the patient's chief complaint and the medical record, and obtains the text similarity index between the patient's chief complaint and the medical record. When the text similarity index is larger, the similarity between the patient's chief complaint and the medical record, that is, the greater the degree of matching, the greater the possibility of recommending the patient to visit the department to which the medical record belongs.
[0060] In this embodiment, the cosine similarity between two word vectors is used as the vector similarity index of the two word vectors. In other embodiments, the result of normalizing the Euclidean distance between the two word vectors can also be used as the vector similarity index.
[0061] It should be noted that the word vector set of medical records is composed of word vectors converted from word vectors obtained after word segmentation processing of the medical records, that is, each word segment of the medical record corresponds to a word vector; in this embodiment, the preset order is the order in which the word vectors in the word vector set of medical records correspond to the words in the medical records; the preset similarity threshold is an empirical value of 0.8, and the implementer can set it according to the specific situation.
[0062] For each medical record, traversing the word vectors in the word vector set of the medical record in a preset order is equivalent to obtaining the vector similarity index of the target word vector and the word vectors in the word vector set of the medical record in a preset order. If the vector similarity index is greater than the preset similarity threshold, it means that the similarity between the word vector of the medical record and the target word vector is large enough to represent the overall similarity between the target word vector and the medical record. The traversal is stopped without subsequent analysis, which can effectively reduce the amount of calculation. If the similarity between the target word vector and the word vector in the word vector of the medical record is less than or equal to the preset similarity threshold, it means that the degree of match between the medical record and the target word vector is low. Then the largest vector similarity index is selected to represent the overall level of similarity between the target word vector and the medical record, and the local similarity index between the target word vector and the medical record is obtained.
[0063] This embodiment measures the degree of matching between each department and the patient's chief complaint by the maximum value of the text similarity index between all medical records of each department and the patient's chief complaint; the maximum value of the text similarity index between all medical records of each department and the patient's chief complaint is used as the patient's initial affiliation to each department.
[0064] Guidance question screening unit 122: An initial guidance set is formed by all guidance questions; when the differences in the patient's affiliation to different departments do not satisfy the selection conditions, the preferred guidance questions are selected from the guidance set based on the similarities between the guidance questions in the guidance set and the patient's chief complaint and the particularity of each guidance question, and then the guidance set is updated.
[0065] Since the patient's chief complaint may contain vague information and the characteristic manifestations of the disease have certain similarities, the similarity between the medical records of multiple departments and the patient's chief complaint is relatively high, that is, there are multiple departments with a high degree of match with the patient's chief complaint, which affects the recommendation of the patient's treatment department; however, if the department with the highest degree of match is directly selected for recommendation, it is easy to reduce the robustness of the recommendation and affect the patient's treatment experience.
[0066] In cases where there are multiple departments with a high degree of match to the patient's chief complaint, this program updates the patient's affiliation with the chief complaint and department by proactively asking the patient to respond to guiding questions and adding the patient's own detailed symptom information.
[0067] The difference between the patient's affiliation to different departments reflects the clarity of the department division, and the department division clarity index is obtained. In this embodiment, the absolute value of the difference between the two largest values of the patient's affiliation to all departments is normalized to obtain the department division clarity index; when the department division clarity index is larger, it means that the department can be clearly divided under the current affiliation, and the possibility of recommending the patient to a treatment department based on the current affiliation is greater.
[0068] It should be noted that, since it is only necessary to recommend the patient to visit the department corresponding to the maximum membership, this embodiment can measure whether the department can be clearly divided by the difference between the two largest values of the patient's membership of all departments. This embodiment uses the Norm function for normalization, and other normalization methods can also be selected, such as function conversion, maximum and minimum normalization, etc., which are not limited here.
[0069] The selection conditions are: the department division clarity index is greater than the preset clarity threshold, or the number of updates of the patient's affiliation to each department is equal to the preset number threshold.
[0070] If the department division clarity index is greater than the preset clarity threshold, it means that the difference between the patient's chief complaint and the degree of matching between different departments is large, and the patient's affiliation to different departments can be clearly divided, and the patient's department can be recommended based on the affiliation in the current situation. The number of updates of the patient's affiliation to the department is equivalent to the number of preferred guidance questions. The purpose of setting the preset number threshold is to prevent patients from not having more descriptive information about their own diseases, that is, symptoms, and to prevent the consultation system from constantly asking new guidance questions that affect the patient's mentality.
[0071] It should be noted that in this embodiment, the preset clarity threshold is an empirical value of 0.75, and the preset quantity threshold is an empirical value of 10, which can be set by the implementer according to the specific situation. The number of updates of the initial membership of the patient to each department is zero.
[0072] When the difference between the patient's affiliation to different departments does not meet the selection condition, that is, the department division clarity index is less than or equal to the preset clarity threshold, it means that the patient's matching degree to multiple departments is relatively close in the current situation, and the recommendation error of the patient's treatment department based on the current affiliation is large. Therefore, this solution needs to select guiding questions related to the patient's disease to obtain additional information about the patient's symptoms.
[0073] The similarity between the guiding questions and the patient's chief complaint in the guiding set measures the match between the guiding questions and the patient's disease. The greater the similarity between the guiding questions and the patient's chief complaint and the stronger the specificity of the guiding questions, the easier it is for the guiding questions to distinguish the degree of match between the patient and different departments, thereby screening the preferred guiding questions, and distinguishing the patient's affiliation to different departments through fewer and more accurate questions.
[0074] Preferably, in some possible implementations of the present embodiment, the method for obtaining the preferred guiding question includes: randomly selecting a guiding question from the guiding set and recording it as the analysis question, randomly selecting a word vector from the word vector set of the analysis question and recording it as the guiding word vector, and obtaining the local special index of the guiding word vector based on the similarity between the guiding word vector and the word vectors in the word vector set of the remaining guiding questions in the guiding set except the analysis question; recording the maximum value of the vector similarity index of the guiding word vector and all the word vectors in the word vector set of the patient's chief complaint as the problem complaint similarity index of the guiding word vector; obtaining the local judgment index of the guiding word vector based on the problem complaint similarity index and the local special index; taking the cumulative sum of the local judgment indicators of all the word vectors in the word vector set of the analysis question as the screening judgment indicator of the analysis question; and taking the guiding question corresponding to the maximum value of the screening judgment index of all the guiding questions in the guiding set as the preferred guiding question.
[0075] The initial guidance set is composed of all guidance questions. The specificity of the guidance word vector is measured by the similarity between the guidance word vector and the word vectors in the word vector set of the remaining guidance questions except the analysis question, and the local specific index is obtained.
[0076] Preferably, in some possible implementations of the present embodiment, the method for obtaining the local special index includes: forming a control guide set of guide word vectors from all word vectors in the word vector set of the remaining guide questions except the analysis question in the guide set; taking the maximum value of the vector similarity indexes of the guide word vector and all the word vectors in the control guide set as the first similarity index of the guide word vector; obtaining the mean value of the vector similarity indexes of the guide word vector and all the word vectors in the control guide set, recorded as the second similarity index of the guide word vector; obtaining the local special index of the guide word vector based on the first similarity index and the second similarity index; the first similarity index and the second similarity index are both negatively correlated with the local special index.
[0077] This embodiment analyzes the particularity of the guide word vector from two aspects. The first similarity index represents the upper limit of the range of the vector similarity index between the guide word vector and the word vector in the control guide set, and the second similarity index represents the overall level of the vector similarity index between the guide word vector and the word vector in the control guide set. If the first similarity index and the second similarity index are larger, it means that the similarity between the guide word vector and the remaining guide questions in the guide set except the analysis question is greater, and the more common the segmentation corresponding to the guide word vector is, the smaller the local special index is. Therefore, the first similarity index and the second similarity index are both negatively correlated with the local special index. In this embodiment, the product of the first similarity index of the guide word vector and the second similarity index is negatively correlated to obtain the local special index of the guide word vector.
[0078] In the embodiment of the present invention, the correlation between the first similarity index, the second similarity index and the local special index may also be constructed through other basic mathematical operations, which will not be limited or elaborated herein.
[0079] It should be noted that this embodiment uses an indicator function with a natural constant as the base for negative correlation mapping. Other negative correlation mapping methods may also be selected, such as function conversion, taking the inverse, etc., which are not limited here.
[0080] The similarity between the guiding word vector and the word vector in the word vector set of the patient's chief complaint reflects the matching degree between the word segmentation result of the guiding question and the patient's chief complaint; the local feature index shows the particularity of the local word segmentation result of the guiding question; if the matching degree between the guiding question and the patient's chief complaint is higher and the guiding question is more special, it means that the content of the guiding question is easier to distinguish the patient's matching situation with the department, that is, the degree of affiliation, and the screening judgment index is larger. Therefore, the problem complaint similarity index and the local judgment index are both positively correlated with the screening judgment index.
[0081] In a specific implementation of the embodiment of the present invention, the screening judgment index is expressed by the formula:
[0082]
[0083] In the formula, Screening and judging indicators for analyzing problems; The total number of word vectors in the word vector set for the analysis problem; The problem complaint similarity index of the a-th word vector in the word vector set for analyzing the problem; The local special index of the ath word vector in the word vector set of the analysis problem; It is the local judgment indicator of the ath word vector in the word vector set of the analysis problem; Norm is the normalization function.
[0084] According to the method for obtaining the screening and judgment indicators of the analysis questions, the screening and judgment indicators of each guiding question are obtained.
[0085] The guiding problem corresponding to the maximum value of the screening judgment index of all guiding problems in the guiding set is taken as the preferred guiding problem. The guiding set after deleting the preferred guiding problem is taken as the updated guiding set.
[0086] Affiliation updating unit 123: obtains the patient's reply text to the preferred guidance question, and updates the patient's affiliation to each department according to the similarity between the reply text and the medical records of the same department, until the patient's updated affiliation to each department meets the selection condition.
[0087] The preferred guiding questions are asked to the patients through the medical inquiry system, and the patients' reply texts to the preferred guiding questions are obtained. The patient's affiliation to each department is updated according to the similarity between the reply texts and the medical records of each department.
[0088] Preferably, in some possible implementations of the present embodiment, the method for updating the patient's affiliation to a department includes: obtaining a text similarity index between the reply text of the preferred guiding question and each medical record; and taking the maximum value of the text similarity indexes between all medical records of each department and the reply text of the preferred guiding question as the patient's updated affiliation to each department.
[0089] It should be noted that the method for obtaining the text similarity index between the preferred guiding question and the medical record is the same as the method for obtaining the text similarity index between the patient's chief complaint and the medical record; it is equivalent to replacing the patient's chief complaint with the preferred guiding question in the process of calculating the text similarity index between the patient's chief complaint and the medical record.
[0090] As an example, suppose there are three departments in the hospital, K1, K2, and K3. Each department has multiple medical records, and there are three guiding questions, Q1, Q2, and Q3. Based on the similarity between the medical records of the same department and the patient's chief complaint, the initial membership of the patient to each department is obtained, which are , , ;in, is the patient's initial affiliation to department K1, is the patient's initial affiliation to department K2, is the patient's initial membership to department K3. The initial guidance set The updating process of the patient's affiliation to the department is as follows:
[0091] The first update process is: according to , , Get the initial department division clarity index W0. When the difference between the initial affiliation of patients to different departments does not meet the selection condition, that is, the department division clarity index Preset clear threshold Or the number of updates of the patient's departmental affiliation is less than the preset threshold, indicating that the similarity between the patient's main complaint and the medical records of different departments is large, and the departments cannot be clearly divided. It is necessary to select the optimal guidance question from the guidance set YJ0. If the guidance question Q1 is the optimal guidance question, Q1 is deleted from the guidance set YJ0 to obtain the updated guidance set . Obtain the patient's reply text to the preferred guiding question Q1, then obtain the text similarity index between the reply text to the guiding question Q1 and all the medical records of each department, and take the maximum value of the text similarity index between the reply text to the guiding question Q1 and the medical records of each department as the patient's updated membership to each department, which are respectively , , .in, The patient's affiliation with department K1 after the first update. The patient's affiliation with department K2 after the first update. It is the patient's affiliation to department K3 after the first update.
[0092] The second update process is: according to , , The initial department division clarity index W0 is updated to obtain the updated department division clarity index W1. When the difference between the updated affiliations of patients to different departments does not meet the selection condition, the department division clarity index W1 is Preset clear threshold Or if the number of updates of the patient's department membership is less than the preset threshold, it is necessary to update the updated guidance set If the guidance problem Q2 is the new preferred guidance problem, the guidance problem Q2 is deleted from the updated guidance set YJ1 to update the updated guidance set YJ1 and obtain the updated guidance set . Obtain the patient's reply text to the preferred guiding question Q2, then obtain the text similarity index between the reply text to the guiding question Q2 and all the medical records of each department, and take the maximum value of the text similarity index between the reply text to the guiding question Q2 and the medical records of each department as the patient's updated membership to each department, which are respectively , , .in, The patient's affiliation with department K1 after the second update. The patient's affiliation with department K2 after the second update. It is the patient's affiliation to department K3 after the second update.
[0093] The third update process is: according to , , The updated department division clarity index W1 is further updated to obtain the updated department division clarity index W2. When the difference between the updated affiliations of the patient to different departments meets the selection condition, the department division clarity index W2 is Preset clear threshold Or the number of updates of the patient's affiliation to the department is equal to the preset threshold, which means that the patient's response to the guiding question Q3 can clearly divide the department or the patient has no other description of his own disease. , , Recommend the patient to a medical department. , , The degree of affiliation of the patient to each department when the selection conditions are met.
[0094] It should be noted that, in the process of updating the patient's affiliation to each department, the updating processes except the first update are the same; in the first update process, the patient's affiliation to the department is obtained through the similarity between the patient's chief complaint and the medical record of each department, and in the updating processes of the remaining update times, the patient's affiliation to the department is obtained through the similarity between the patient's reply text to the preferred guidance questions and the medical record of each department, and the other steps are the same.
[0095] The treatment department recommendation module 130 is used to recommend a treatment department for a patient based on the patient's affiliation with each department when the selection conditions are met.
[0096] In this embodiment, the medical consultation system recommends to the user the department corresponding to the maximum value of the patient's membership degree among all departments when the candidate conditions are met, and the patient goes to the department for medical consultation.
[0097] In other embodiments, all departments are arranged in descending order according to the patient's affiliation to all departments when the selection condition is met, to obtain a department recommendation sequence; the department with a smaller subscript in the department recommendation sequence has a higher degree of recommendation for the patient to visit the department, and the department with a larger subscript in the department recommendation sequence has a lower degree of recommendation for the patient to visit the department.
[0098] So far, the present invention is completed.
[0099] Embodiment 2:
[0100] Figure 3 A computer device schematic diagram of an Internet intelligent medical consultation device provided by an embodiment of the present invention. For example, Figure 3 As shown, the computer device includes: a memory 201, a processor 202, and a computer program 203 stored in the memory 201 and running on the processor 202, wherein when the processor 202 executes the computer program 203, the computer device can execute any one of the Internet intelligent medical consultation systems introduced above.
[0101] In addition, an embodiment of the present application also protects a device, which may include a memory and a processor, wherein the memory stores executable program code, and the processor is used to call and execute the executable program code to execute an Internet intelligent medical consultation system provided by an embodiment of the present application.
[0102] In this embodiment, the functional modules of the device can be divided according to the above method example. For example, each functional module can be corresponded, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware. It should be noted that the division of modules in this embodiment is schematic and is only a logical function division. There may be other division methods in actual implementation.
[0103] It should be understood that the device provided in this embodiment is used to execute the above-mentioned Internet intelligent medical consultation system, and thus can achieve the same effect as the above-mentioned implementation method.
[0104] In the case of an integrated unit, the device may include a processing module and a storage module. When the device is applied to a device, the processing module may be used to control and manage the actions of the device. The storage module may be used to support the device to execute mutual program codes, etc.
[0105] The processing module may be a processor or a controller, which may implement or execute various exemplary logic blocks, modules and circuits disclosed in the present application. The processor may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc. The storage module may be a memory.
[0106] Embodiment 3:
[0107] This embodiment also provides a computer-readable storage medium, in which a computer program code is stored. When the computer program code is executed on a computer, the computer executes the above-mentioned related method steps to implement an Internet intelligent medical consultation system provided by the above-mentioned embodiment.
[0108] Among them, the device and computer-readable storage medium provided in this embodiment are used to execute the corresponding system provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding system provided above, and will not be repeated here.
[0109] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0110] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.
[0111] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.
Claims
1. An Internet intelligent medical consultation system, It is characterized in that The system includes: A data collection module is used to obtain the patient's chief complaint, several medical records and guiding questions; the medical records have corresponding departments; The department affiliation analysis module is used to determine the initial affiliation of the patient to each department based on the similarity between the medical records of the same department and the patient's chief complaint; the initial guidance set is formed by all the guidance questions; when the difference between the patient's affiliation to different departments does not meet the selection condition, the preferred guidance questions are selected from the guidance set according to the similarity between the guidance questions in the guidance set and the patient's chief complaint and the particularity of each guidance question, and then the guidance set is updated; the patient's reply text to the preferred guidance question is obtained, and the patient's affiliation to each department is updated according to the similarity between the reply text and the medical records of the same department, until the updated patient's affiliation to each department meets the selection condition; The treatment department recommendation module is used to recommend the patient's treatment department based on the patient's affiliation with each department when the selection conditions are met; Determining the patient's initial affiliation with each department based on the similarity between the medical records of the same department and the patient's chief complaint includes: Obtain textual similarity metrics between the patient’s chief complaint and each medical record; The maximum value of the text similarity index between all medical records of each department and the patient's chief complaint is used as the patient's initial membership degree to each department; The text similarity index between the patient's chief complaint and each medical record includes: All medical records and patients' chief complaints are recorded as analysis texts, and the analysis texts are segmented and converted into word vectors. All word vectors of the analysis texts constitute a word vector set. Select any word vector from the word vector set of the patient's chief complaint and record it as the target word vector, obtain the vector similarity index between the target word vector and each word vector in the word vector set of each medical record, traverse the word vectors in the word vector set of each medical record in a preset order, and stop traversing when the vector similarity index is greater than a preset similarity threshold, and use the vector similarity index when traversal stops as the local similarity index between the target word vector and each medical record; If the vector similarity indexes of the target word vector and all the word vectors in the word vector set of each medical record are less than or equal to the preset similarity threshold, the maximum value of the vector similarity indexes of the target word vector and all the word vectors in the word vector set of each medical record is taken as the local similarity index of the target word vector and each medical record; The average of the local similarity indexes of all word vectors in the word vector set of the patient's chief complaint and each medical record is used as the text similarity index between the patient's chief complaint and each medical record; The step of selecting a preferred guiding question from the guiding set according to the similarity between the guiding questions in the guiding set and the patient's main complaint and the particularity of each guiding question comprises: Selecting a guiding question from the guiding set as the analysis question, selecting a word vector from the word vector set of the analysis question as the guiding word vector, and obtaining a local special index of the guiding word vector according to the similarity between the guiding word vector and the word vectors in the word vector set of the remaining guiding questions in the guiding set except the analysis question; The maximum value of the vector similarity indexes of the guide word vector and all the word vectors in the word vector set of the patient's chief complaint is recorded as the problem chief complaint similarity index of the guide word vector; According to the problem complaint similarity index and the local special index, a local judgment index of the guide word vector is obtained; the cumulative sum of the local judgment indexes of all word vectors in the word vector set of the analysis problem is used as a screening judgment index for the analysis problem; The guiding question corresponding to the maximum value of the screening judgment index of all guiding questions in the guiding set is taken as the preferred guiding question; The local special index for obtaining the guide word vector includes: A control guide set of guide word vectors is formed by all word vectors in the word vector set of the remaining guide questions in the guide set except the analysis question; The maximum value of the vector similarity indexes of the guide word vector and all the word vectors in the control guide set is used as the first similarity index of the guide word vector; Obtaining the mean of the vector similarity indexes of the guide word vector and all the word vectors in the control guide set, and recording it as the second similarity index of the guide word vector; According to the first similarity index and the second similarity index, a local special index of the guide word vector is obtained; the first similarity index and the second similarity index are both negatively correlated with the local special index.
2. According to claim 1, an Internet intelligent medical consultation system, It is characterized in that The candidate conditions include: The absolute value of the difference between the two largest values of the patient's membership of all departments is normalized to obtain a clear indicator of department division; The candidate condition is: the department division clarity index is greater than a preset clarity threshold, or the number of updates of the patient's affiliation to each department is equal to a preset number threshold.
3. According to claim 1, an Internet intelligent medical consultation system, It is characterized in that The updating of the patient's affiliation to each department according to the similarity between the reply text and the medical records of the same department includes: Obtaining the text similarity index between the response text of the preferred guiding question and each medical record; The maximum value of the text similarity index between all the medical records of each department and the answer texts of the preferred guiding questions is used as the patient's updated membership to each department.
4. The Internet intelligent medical consultation system according to claim 1, It is characterized in that The updating of the guide set includes: The guidance set after deleting the preferred guidance question is used as the updated guidance set.
5. The Internet intelligent medical consultation system according to claim 1, It is characterized in that The vector similarity index is the cosine similarity between two word vectors.
6. The Internet intelligent medical consultation system according to claim 2, It is characterized in that The preset clarity threshold is 0.75.