Clinical aid decision-making method and device based on medical information input

Abstract

The invention provides a clinical auxiliary decision-making method and device based on medical information entry, and belongs to the technical field of auxiliary decision-making, and the method comprises the steps: constructing a multi-source medical database, obtaining the medical appeal of a target patient, and locking the medical scene of the target patient; setting a complete period for the target patient according to the medical scene and in combination with the multi-source medical database, and sequentially inputting service information of each medical management link of the target patient in the complete period to form complete-period management information; standardizing the provided predictive information and the feedback predictive information under the same medical management link to obtain an initial decision for the medical appeal; according to the method, the features of doctors for solving the medical demands of the target patient are obtained, the initial decision is optimized by combining the historical health features of the target patient to obtain and output the auxiliary decision, the defects of a traditional medical mode can be effectively overcome, and the medical service quality and efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of decision support, and in particular to a clinical decision support method and device based on medical information entry. Background Art

[0002] Under the traditional healthcare model, patient medical information is fragmented, lacking effective integration and coherent management across all healthcare processes. Data interoperability between different medical institutions makes it difficult for doctors to fully understand patient information, impacting diagnostic accuracy and treatment planning.

[0003] While tools like electronic medical records have emerged with the development of healthcare informatization, they are often limited to a single institution or process, making it impossible to manage the entire healthcare cycle. Furthermore, decisions have traditionally been based on individual physician experience and limited medical records, lacking comprehensive, forward-looking patient analysis, leading to inefficient decision-making.

[0004] Therefore, the present invention proposes a clinical decision-making support method and device based on medical information entry. Summary of the Invention

[0005] The present invention provides a clinical decision-making support method and device based on medical information input, which is used to build a multi-source medical database and realize the integration of full-cycle medical management information. It can conduct standardized analysis of patients' medical demands and optimize decision-making based on multiple factors. It can effectively make up for the shortcomings of traditional medical models and improve the quality and efficiency of medical services.

[0006] The present invention provides a clinical decision-making assistance method based on medical information input, comprising:

[0007] Step 1: Build a multi-source medical database, obtain the medical needs of target patients, and identify the medical scenarios of the target patients;

[0008] Step 2: According to the medical scenario and in combination with the multi-source medical database, a full cycle is set for the target patient, and the service information of each medical management link of the target patient in the full cycle is sequentially entered to form full cycle management information, wherein the service information in each medical management link includes the predictive information provided by the doctor to the patient and the predictive feedback information of the patient to himself;

[0009] Step 3: Standardize the provided predictive information and the feedback predictive information in the same medical management link to obtain an initial decision for the medical demand;

[0010] Step 4: Obtain the characteristics of the doctor who can solve the medical needs of the target patient, and combine the historical health characteristics of the target patient to optimize the initial decision to obtain an auxiliary decision and output it, wherein the auxiliary decision is related to the treatment medication.

[0011] Preferably, a multi-source medical database is constructed, including:

[0012] Capturing existing historical patient medical information from different medical platforms and determining information attributes of the medical information, wherein the information attributes include: a complete attribute and an incomplete attribute, and treating the medical information with the complete attribute as first information and treating the medical information with the incomplete attribute as second information;

[0013] Performing a first cluster analysis on all first information to obtain a plurality of first clusters, and simultaneously performing a second cluster analysis on all second information, eliminating cluster results in which the amount of second information present in the second cluster results is less than a specified amount, and retaining the remaining results to obtain a plurality of second clusters;

[0014] Matching and analyzing the cluster information of each second cluster with the cluster information of the first cluster, filtering the cluster information of the second clusters with a matching degree greater than a preset degree and the clustering results of the second clusters according to the matching degree, and adding them to the clustering results of the first cluster to obtain a first new result;

[0015] Performing probability distribution and information reconstruction on each first new result to obtain new information;

[0016] A multi-source medical database is constructed based on all the first information and the new information, wherein the multi-source medical database includes: a plurality of reference information, and each reference information includes a treatment purpose, a treatment cycle consistent with the treatment purpose, treatment drugs for each treatment stage under the treatment cycle, treatment uses, and treatment effects.

[0017] Preferably, information screening is performed on the clustering results of the second cluster according to the matching degree, including:

[0018] Selecting the division distance length consistent with the matching degree from the matching degree-distance comparison table;

[0019] Draw a circle with the division distance as the radius and the point where the corresponding cluster information is located as the origin, and determine the shape relationship between each information point inside the circle except the point corresponding to the cluster information and the two nearest points;

[0020] If it is a straight line relationship, assigning a first mark to the corresponding point according to a first ratio of the length of the corresponding straight line to the divided distance, and in combination with an angular relationship between the straight line relationship and the origin;

[0021]

[0022] Where D1 represents the first marking result; Lz represents the length of the corresponding straight line; Lh represents the division distance; represents the first ratio; θ represents the angle between the two endpoints of the straight line relationship and the origin, and is less than or equal to 180°; a1 represents the first marking threshold;

[0023] If it is a triangular relationship, assign a second label to the corresponding point based on the area ratio of the corresponding triangle to the circle and the distance between the corresponding point and the corresponding information point in the corresponding cluster information;

[0024]

[0025] Where D2 represents the second marking result; SΔ represents the area of ​​the corresponding triangle; π×Lh 2 represents the area of ​​the circle; Ld represents the distance between the corresponding cluster information point and the corresponding information point; a2 represents the second marking threshold;

[0026] According to the first marking result and the second marking result, a first point farthest from the corresponding point of the corresponding cluster information is selected from the valuable marks, and a second point closest to the corresponding point of the corresponding cluster information is selected from the worthless marks;

[0027] If the first point is closer to the cluster information corresponding point than the second point, or the first point and the second point are equally far from the cluster information corresponding point, then the information of the point with the value mark is used as the screening information;

[0028] If the second point is closer to the cluster information corresponding point than the first point, then a first new circle is drawn with the distance between the second point and the cluster information corresponding point as the new radius and the cluster information corresponding point as the origin;

[0029] Divide the first new circle into four quadrants, locate the third point farthest from the origin in the quadrant with the maximum quadrant density, and determine the length to be divided;

[0030]

[0031] Where Lave represents the average value of the length farthest from the origin in the remaining three quadrants; w1 represents the maximum quadrant density; w2 represents the minimum quadrant density; sumw represents the sum of all quadrant densities; σL 2 represents the variance of the distance between each information point and the origin in the first new circle; L3 represents the distance between the third point and the origin; rx represents the length of the new radius; Lr represents the length to be divided;

[0032] A second new circle is drawn with the length to be divided as the radius and the corresponding point of the cluster information as the origin;

[0033] The information of the points with value marks within the second new circle is used as screening information.

[0034] Preferably, performing probability distribution and information reconstruction on each first new result to obtain new information includes:

[0035] Retrieving a field blank table from a type-table comparison library according to the treatment type of the cluster information in the first new result, wherein the field blank table includes a plurality of basic placement cells;

[0036] Perform field conversion on each basic expression in each piece of information in the first new result and place it in the comparison basic placement grid, and construct a field difference vector for each column of information in the placement table;

[0037] According to the field placement probability of each column of information, the field difference vector is processed to obtain a new field;

[0038] A broad definition of each new field is obtained based on its unity with the corresponding cluster information, and new information is obtained by freely splicing and combining them.

[0039] Preferably, setting a full cycle for the target patient according to the medical scenario and in combination with a multi-source medical database includes:

[0040] Analyzing the medical scenario to obtain a number of scenario parameters, and comparing and analyzing each scenario parameter with each piece of information in the multi-source medical data to obtain a comparison vector for each piece of information;

[0041] Selecting the best vector and the second best vector from all comparison vectors to obtain first reference information and second reference information;

[0042] A full cycle is set for information fusion of the first reference information and the second reference information.

[0043] Preferably, obtaining an initial decision regarding the medical request includes:

[0044] performing a correlation analysis on the corresponding provided predictive information and the feedback predictive information under the same medical management link according to the interaction relationship between the doctor and the patient, thereby obtaining a first distribution of the first comparison information in the provided predictive information and a second distribution of the second comparison information in the feedback predictive information;

[0045] Obtaining a patient compliance coefficient to a doctor based on the degree of consistency between the first distribution and the second distribution, and determining a relative independence coefficient based on the first independent information in the provided predictive information and the second independent information in the feedback predictive information;

[0046] According to the one-to-one correspondence of compliance coefficient and relative independence coefficient, the service information for the same patient in the same medical management link is weighted and ranked, and the service information with the highest weight is retained;

[0047] Determine the reference information of the entire cycle, split the service information and reference information according to the medical management link, and fuse the information of the same medical management link. Standardize the corresponding fused information according to the management indicators under each medical management link to obtain an initial decision.

[0048] Preferably, optimizing the initial decision to obtain an auxiliary decision includes:

[0049] Perform feature analysis on doctor characteristics and historical health characteristics based on the management indicators under each medical management link to obtain the array to be corrected for each management indicator;

[0050] The initial decision is optimized according to the array to be corrected to obtain an auxiliary decision.

[0051] The present invention provides a clinical decision support device based on medical information input, comprising:

[0052] The scenario analysis module is used to build a multi-source medical database, obtain the medical demands of target patients, and identify the medical scenarios of the target patients;

[0053] a cycle construction module, configured to set a full cycle for the target patient according to the medical scenario and in combination with a multi-source medical database, and sequentially input the service information of each medical management link of the target patient in the full cycle to form full cycle management information, wherein the service information in each medical management link includes the predictive information provided by the doctor to the patient and the predictive information of the patient's feedback to himself;

[0054] An initial decision acquisition module is used to standardize the provided predictive information and the feedback predictive information under the same medical management link to obtain an initial decision for the medical demand;

[0055] The decision optimization module is used to obtain the characteristics of the doctor who can solve the medical needs of the target patient, and combine the historical health characteristics of the target patient to optimize the initial decision to obtain an auxiliary decision and output it, wherein the auxiliary decision is related to the treatment medication.

[0056] Compared with the prior art, the present invention has the following advantages:

[0057] Building a multi-source medical database and integrating full-cycle medical management information, conducting standardized analysis of patients' medical demands and optimizing decisions based on multiple factors can effectively make up for the shortcomings of the traditional medical model and improve the quality and efficiency of medical services.

[0058] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0059] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0061] Figure 1 This is a flow chart of a clinical decision-making support method based on medical information input in an embodiment of the present invention;

[0062] Figure 2 This is a structural diagram of a clinical decision support device based on medical information entry in an embodiment of the present invention. DETAILED DESCRIPTION

[0063] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0064] The present invention provides a clinical decision-making assistance method based on medical information input, such as Figure 1 Shown, including:

[0065] Step 1: Build a multi-source medical database, obtain the medical needs of target patients, and identify the medical scenarios of the target patients;

[0066] Step 2: According to the medical scenario and in combination with the multi-source medical database, a full cycle is set for the target patient, and the service information of each medical management link of the target patient in the full cycle is sequentially entered to form full cycle management information, wherein the service information in each medical management link includes the predictive information provided by the doctor to the patient and the predictive feedback information of the patient to himself;

[0067] Step 3: Standardize the provided predictive information and the feedback predictive information in the same medical management link to obtain an initial decision for the medical demand;

[0068] Step 4: Obtain the characteristics of the doctor who can solve the medical needs of the target patient, and combine the historical health characteristics of the target patient to optimize the initial decision to obtain an auxiliary decision and output it, wherein the auxiliary decision is related to the treatment medication.

[0069] In this embodiment, the multi-source medical database refers to obtaining medical information of different patients from different medical platforms (preliminary information), where different medical platforms include but are not limited to: using web crawler technology to collect historical patient medical information from different medical platforms (such as hospital official websites, online consultation platforms, medical data sharing platforms, etc.), specifically used to record the treatment objectives of different patients, treatment cycles consistent with the treatment objectives, treatment medications for each treatment stage under the treatment cycle, treatment uses and treatment effects and other actual practical results, providing comprehensive and reliable basic medical data.

[0070] In this example, a medical complaint refers to a patient's description of symptoms via the hospital's self-service registration system, or by communicating with a triage nurse about their discomfort. For example, if a patient reports a cough or fever, the triage nurse will record this complaint and enter it into the system. The medical scenario is determined based on the complaint. For example, if a complaint includes coughing and fever, the medical scenario is determined to be a respiratory disease scenario. It should be noted that medical scenarios are related to the outpatient categories set by the hospital, and can also include cardiovascular disease scenarios, dermatology scenarios, or orthopedics scenarios, etc.

[0071] In this embodiment, the full cycle is implemented based on the database and medical scenarios, and is aimed at the full cycle of medication. For example, the full cycle of medication for patients with hypertension includes: the primary medication stage and the stable medication stage.

[0072] Treatment objectives: Control blood pressure levels, reduce the risk of complications such as cardiovascular and cerebrovascular diseases caused by high blood pressure, and maintain blood pressure within the normal range for a long time (generally systolic blood pressure <140 mmHg and diastolic blood pressure <90 mmHg).

[0073] Primary medication adjustment stage: Use nifedipine sustained-release tablets, starting with a low dose based on the patient's blood pressure, such as 10 mg, 1-2 times a day (therapeutic medication);

[0074] Nifedipine sustained-release tablets lower blood pressure (therapeutic use) by blocking calcium ion influx, relaxing vascular smooth muscle, and reducing peripheral vascular resistance;

[0075] After 2 weeks of medication, blood pressure gradually decreased (therapeutic effect);

[0076] Stable medication phase: If blood pressure is well controlled, the dose of nifedipine extended-release tablets may be maintained; if blood pressure fluctuations occur, irbesartan tablets, 75 mg, once daily (therapeutic medication) may be added;

[0077] Irbesartan tablets act on the renin-angiotensin-aldosterone system, inhibiting the binding of angiotensin II to receptors, thus lowering blood pressure. They can be combined with nifedipine extended-release tablets to enhance the antihypertensive effect (for therapeutic use).

[0078] After 2 weeks of combined medication, the blood pressure of most patients can be stably controlled within the target range (systolic blood pressure <140 mmHg and diastolic blood pressure <90 mmHg) (therapeutic effect).

[0079] In this embodiment, providing predictive information refers to the doctor in the corresponding medical scenario informing the patient of the possible development of the disease and treatment in different medical management links based on professional knowledge and experience, such as informing the patient of the functional recovery after reduction of a fracture. Feedback predictive information refers to the patient's feedback on the expected development of health status and treatment expectations in different medical management links, such as the patient's hope to be able to walk within 10 days.

[0080] In this embodiment, a standard information template is established to convert the information provided by doctors and feedback from patients into a unified format. For example, the description of pain is uniformly quantified using the "Visual Analog Scale (VAS)", and the patient's descriptions such as "a little pain" and "very painful" are converted into specific scores. Natural language processing (NLP) technology is used to classify text information, extract key information, and convert it into analyzable data. The information is analyzed using preset rules and algorithms to generate initial decisions. For example, based on the severity of symptoms and body temperature of cold patients, advice on whether medication and rest are needed is given.

[0081] In this embodiment, the doctor characteristics refer to the doctor's professional field, the number of surgeries for diseases in the professional field, the success rate of surgeries, etc.

[0082] In this embodiment, historical health characteristics refer to extracting past medical history, allergy history, family medical history, etc. from the patient's electronic medical record. For example, the patient has a family history of coronary heart disease and a history of hypertension.

[0083] In this embodiment, optimized decision refers to the adjustment of the initial decision based on the doctor's characteristics and health characteristics. For example, for the treatment decision of a patient with coronary heart disease, an experienced doctor who is good at treating coronary heart disease adjusts the dosage and type of medication based on the patient's history of hypertension.

[0084] The beneficial effects of the above technical solution are: building a multi-source medical database and realizing the integration of full-cycle medical management information, conducting standardized analysis of patients' medical demands and combining multiple factors to optimize decision-making, which can effectively make up for the shortcomings of the traditional medical model and improve the quality and efficiency of medical services.

[0085] The present invention provides a clinical decision-making support method based on medical information input, which constructs a multi-source medical database, including:

[0086] Capturing existing historical patient medical information from different medical platforms and determining information attributes of the medical information, wherein the information attributes include: a complete attribute and an incomplete attribute, and treating the medical information with the complete attribute as first information and treating the medical information with the incomplete attribute as second information;

[0087] Performing a first cluster analysis on all first information to obtain a plurality of first clusters, and simultaneously performing a second cluster analysis on all second information, eliminating cluster results in which the amount of second information present in the second cluster results is less than a specified amount, and retaining the remaining results to obtain a plurality of second clusters;

[0088] Matching and analyzing the cluster information of each second cluster with the cluster information of the first cluster, filtering the cluster information of the second clusters with a matching degree greater than a preset degree and the clustering results of the second clusters according to the matching degree, and adding them to the clustering results of the first cluster to obtain a first new result;

[0089] Perform probability distribution and information reconstruction on all first new results to obtain new information;

[0090] A multi-source medical database is constructed based on all the first information and the new information, wherein the multi-source medical database includes: a plurality of reference information, and each reference information includes a treatment purpose, a treatment cycle consistent with the treatment purpose, treatment drugs for each treatment stage under the treatment cycle, treatment uses, and treatment effects.

[0091] In this embodiment, probability distribution refers to the distribution of relevant information of all first new results based on the same indicator, information reconstruction refers to information combination, and the acquisition of new information is to amplify the original information, that is, based on the probability distribution of information such as symptoms, diagnosis and treatment plans in the first new results of patients with a certain type of disease, more comprehensive and reasonable disease diagnosis and treatment path information is reconstructed.

[0092] In this embodiment, complete attributes refer to the specific content that exists under each indicator corresponding to the treatment purpose, the treatment cycle consistent with the treatment purpose, the treatment medications for each treatment stage under the treatment cycle, the treatment purpose, and the treatment effect. The attributes of the corresponding medical information are regarded as complete attributes. If there is no specific content under a certain indicator, the attributes of the corresponding medical information are regarded as incomplete attributes.

[0093] In this embodiment, for the first information, a commonly used clustering algorithm such as the K-Means clustering algorithm is used. First, determine the appropriate number of clusters K (which can be determined by methods such as the elbow rule), calculate the distance between each first information data point (such as the Euclidean distance), and divide the data points into K clusters. For the second information, a clustering algorithm is also used (such as the DBSCAN density clustering algorithm, which is suitable for processing clustering of incomplete data). After clustering, the number of second information in each clustering result is counted, and a specified number threshold is set (such as 5 based on the data scale and experience), and the clustering results with a number less than the threshold are eliminated. A number of first clusters and a number of second clusters after screening are obtained. For example, the medical records of patients with diabetes in the first information are clustered into a first cluster; some records describing the symptoms of diabetes in the second information form a second cluster after cluster screening.

[0094] In this embodiment, the specified number is 10.

[0095] In this embodiment, the similarity between the cluster information of each second cluster and the cluster information of the first cluster is calculated based on the similarity of the information features. For the cluster information of the second cluster (such as the central feature vector of the cluster) and the cluster information of the first cluster, the similarity (such as cosine similarity) between the two is calculated. A preset degree threshold (such as 0.6) is set. When the matching degree is greater than the threshold, the information in the second cluster is filtered according to the matching degree, and the information with a high matching degree is retained. The filtered information is merged into the corresponding first cluster, and then part of the information is filtered from the corresponding second cluster result and added to the cluster result of the first cluster, which is regarded as the first new result.

[0096] In this embodiment, all first information and new information are organized and stored according to a database storage structure (e.g., a relational database table structure, including patient information tables, diagnosis tables, treatment tables, etc.). A database management system (e.g., MySQL, Oracle, etc.) is used for data entry, management, and maintenance, and indexing is established to optimize data query and access performance. This creates a multi-source medical database containing a wealth of medical information, encompassing both complete historical patient medical information and new information generated through processing.

[0097] The beneficial effects of the above technical solution are: eliminating the second clustering results with too small a number, avoiding the interference of abnormal small clusters on subsequent analysis, improving the reliability and effectiveness of the clustering results, and enriching the content of the first cluster cluster by integrating the second information that was originally incomplete but related to the complete information, making the clustering results more comprehensive, and excavating more potentially valuable information associations, improving the richness and availability of data, and further improving the medical data content based on probability distribution and information reconstruction, providing a better data foundation for subsequent medical decision support and other applications.

[0098] The present invention provides a clinical decision-making support method based on medical information input, which screens the clustering results of the second cluster according to the matching degree, including:

[0099] Selecting the division distance length consistent with the matching degree from the matching degree-distance comparison table;

[0100] Draw a circle with the division distance as the radius and the point where the corresponding cluster information is located as the origin, and determine the shape relationship between each information point inside the circle except the point corresponding to the cluster information and the two nearest points;

[0101] If it is a straight line relationship, assigning a first mark to the corresponding point according to a first ratio of the length of the corresponding straight line to the divided distance, and in combination with an angular relationship between the straight line relationship and the origin;

[0102]

[0103] Where D1 represents the first marking result; Lz represents the length of the corresponding straight line; Lh represents the division distance; represents the first ratio; θ represents the angle between the two endpoints of the straight line relationship and the origin, and is less than or equal to 180°; a1 represents the first marking threshold;

[0104] If it is a triangular relationship, assign a second label to the corresponding point based on the area ratio of the corresponding triangle to the circle and the distance between the corresponding point and the corresponding information point in the corresponding cluster information;

[0105]

[0106] Where D2 represents the second marking result; SΔ represents the area of ​​the corresponding triangle; π×Lh 2 represents the area of ​​the circle; Ld represents the distance between the corresponding cluster information point and the corresponding information point; a2 represents the second marking threshold;

[0107] According to the first marking result and the second marking result, a first point farthest from the corresponding point of the corresponding cluster information is selected from the valuable marks, and a second point closest to the corresponding point of the corresponding cluster information is selected from the worthless marks;

[0108] If the first point is closer to the cluster information corresponding point than the second point, or the first point and the second point are equally far from the cluster information corresponding point, then the information of the point with the value mark is used as the screening information;

[0109] If the second point is closer to the cluster information corresponding point than the first point, then a first new circle is drawn with the distance between the second point and the cluster information corresponding point as the new radius and the cluster information corresponding point as the origin;

[0110] Divide the first new circle into four quadrants, locate the third point farthest from the origin in the quadrant with the maximum quadrant density, and determine the length to be divided;

[0111]

[0112] Where Lave represents the average value of the length farthest from the origin in the remaining three quadrants; w1 represents the maximum quadrant density; w2 represents the minimum quadrant density; sumw represents the sum of all quadrant densities; σL 2 represents the variance of the distance between each information point and the origin in the first new circle; L3 represents the distance between the third point and the origin; rx represents the length of the new radius; Lr represents the length to be divided;

[0113] A second new circle is drawn with the length to be divided as the radius and the corresponding point of the cluster information as the origin;

[0114] The information of the points with value marks within the second new circle is used as screening information.

[0115] In this embodiment, the matching degree-distance comparison table records the lengths of the divided distances corresponding to different matching degrees. For example, assuming the matching degree range is [0.6, 1], it is divided into several intervals, such as [0.6-0.7), [0.7-0.8), [0.8-0.9), and [0.9-1], which correspond to different divided distance lengths. If the current matching degree is 0.85, by querying the matching degree-distance comparison table, the interval [0.8-0.9) in which it is located is found. Assuming that the length of the divided distance corresponding to this interval is 5 (the unit can be set according to the actual situation, such as centimeters, the number of data point intervals, etc.), the selected divided distance length is 5.

[0116] In this embodiment, in a two-dimensional coordinate space (assuming that the data points can be mapped to a two-dimensional plane, if it is high-dimensional data, it can be processed by dimensionality reduction or other methods), it is known that the partition distance length is 5 (step one example data), and the point where the corresponding cluster information is located is the coordinate origin (0,0). For each information point, the distance between it and other surrounding information points is calculated to find the two closest points. According to the relative position relationship of these three points (the information point and its two nearest points), it is determined whether it is a linear relationship or a triangular relationship. For example, if the coordinates of the three points are A(1,1), B(2,2), and C(3,3), respectively, by calculating the slope or other methods, it is found that they are on the same straight line, which can be determined to be a linear relationship; if the coordinates of the three points are D(1,1), E(3,1), and F(2,3), forming a triangle, it is determined to be a triangular relationship.

[0117] In this embodiment, the value of a1 is 0.7, and the value of a2 is 0.6.

[0118] In this embodiment, the ratio of the straight line length to the dividing distance reflects the size of the distribution range of the information point in the straight line direction relative to the demarcated space range. The smaller the ratio, the more concentrated the information points are near the origin, and the more valuable they may be. The angular relationship between the straight line and the origin reflects the directional characteristics of the straight line in space. The angle information is incorporated into the formula through the inverse tangent function, and the impact of the direction on the information value is comprehensively considered. This formula can quantify the value of the information point under the straight line relationship, thereby performing effective labeling and screening. The area ratio of the triangle to the circle reflects the distribution density of the information point in the demarcated circular space. The smaller the ratio, the more concentrated the information point distribution is near the origin, and the higher the possible value. The distance between the corresponding cluster information corresponding point and the corresponding information point reflects the distance relationship between the information point and the core point (cluster information corresponding point). The closer the distance, the higher the possible value. These two factors are combined through this formula to quantify the value of the information point under the triangular relationship.

[0119] In this embodiment, if the second point is closer to the cluster information corresponding point than the first point, it means that the distance from the second point to the origin is smaller than the distance from the first point to the origin.

[0120] In this embodiment, the first new circle is divided into four quadrants according to a plane rectangular coordinate system, namely the first quadrant (x>0, y>0), the second quadrant (x<0, y>0), the third quadrant (x<0, y<0), and the fourth quadrant (x>0, y<0). The number of information points in each quadrant is counted. Assume that there are 10 information points in the first quadrant, 5 information points in the second quadrant, 3 information points in the third quadrant, and 7 information points in the fourth quadrant. The density of each quadrant is calculated separately. Assume that the area of ​​the circle is S=π×rx 2 , first quadrant density Second Quadrant Density Third Quadrant Density Fourth Quadrant Density The maximum quadrant density At this time, the first quadrant is locked as the maximum density quadrant.

[0121] In this embodiment, when there are information points in the remaining three quadrants, if the distances from the origin to the remaining three quadrants are respectively 5, 3, and 6, then Lave = (5+3+6) / 3;

[0122] If there are information points in only two quadrants or one quadrant, then the sum of the farthest lengths in the quadrant corresponding to the information point / the number of quadrants corresponding to the information point is used.

[0123] If there is no information point in the remaining three quadrants, then Lave is 0 and σL 2 is 0.

[0124] In this embodiment, Taking into account the new radius and the distance from the farthest point to the origin in the maximum quadrant density, it reflects the approximate range scale of the core area, σL 2 This reflects the degree of dispersion of information points within the circle. A greater degree of dispersion indicates a greater adjustment to the partition length. A large variance indicates a dispersed distribution of information points, necessitating a larger or smaller partition length to adequately capture valuable information. A small variance indicates a relatively concentrated distribution of information points, requiring a smaller adjustment. The quadrant density-related factors w1, w2, and sumw reflect the density of information distribution and are relevant to the determination of partition length. High-density areas may require finer partitioning.

[0125] The beneficial effects of the above technical solution are: for information points with linear relationships, it is possible to accurately judge whether they are valuable based on their position and direction characteristics in space, providing a clear basis for subsequent retention or elimination of information, improving the refinement of information screening, simplifying the complex spatial relationship between information points into two typical relationships: straight lines and triangles, facilitating the subsequent use of corresponding mathematical formulas to make value judgments on information points, improving the scientificity and systematicness of information screening, and determining a new screening range based on the distance relationship between the point and the corresponding point of cluster information. The range determined by the point closer to the corresponding point of cluster information is more likely to contain information closely related to the cluster, providing a suitable spatial range definition for subsequent screening. By determining the length to be divided, the information within the circle can be more reasonably divided and processed based on the distribution characteristics of the information points within the circle (distance, discreteness, quadrant density, etc.), providing a quantitative basis for accurate information screening.

[0126] The present invention provides a clinical decision-making support method based on medical information input, which performs probability distribution and information reconstruction on each first new result to obtain new information, including:

[0127] Retrieving a field blank table from a type-table comparison library according to the treatment type of the cluster information in the first new result, wherein the field blank table includes a plurality of basic placement cells;

[0128] Perform field conversion on each basic expression in each piece of information in the first new result and place it in the comparison basic placement grid, and construct a field difference vector for each column of information in the placement table;

[0129] According to the field placement probability of each column of information, the field difference vector is processed to obtain a new field;

[0130] A broad definition of each new field is obtained based on its unity with the corresponding cluster information, and new information is obtained by freely splicing and combining them.

[0131] In this embodiment, it is assumed that the treatment type of the cluster information in the first new result is "diabetes treatment". In the type-table reference library, field blank tables corresponding to different treatment types are pre-stored. By searching for records with the treatment type of "diabetes treatment", the corresponding field blank table is found. The field blank table contains several basic placement cells, such as "symptom description", "drug name", "drug dosage", "blood glucose monitoring frequency" and other basic placement cells. For example, in the first new result, there are two information records about diabetes treatment. Information record 1: "The patient has symptoms of polydipsia and polyphagia, and uses metformin at a daily dose of 1000 mg, and monitors blood glucose 3 times a week." Information record 2: "The patient has symptoms of polyuria, and takes metformin at a daily dose of 1500 mg, and monitors blood glucose 4 times every two weeks."

[0132] Convert each basic expression in each piece of information into a field, for example, "Symptoms of polydipsia and polyphagia" corresponds to the "Symptom Description" field, "Metformin" corresponds to the "Medication Name" field, etc. Place these converted contents in the control basic placement grid to form a placement table, specifically:

[0133] Information Records Symptom description Name of medication Dosage Monitoring frequency Record 1 Drink more and eat more Metformin 1000 mg / day 3 times / week Record 2 Polyuria Metformin 1500 mg / day 4 times / two weeks

[0134] In this embodiment, a field difference vector is calculated for each column of information in the placement table. Taking the "drug dosage" column as an example, the dosages of two records are different, and the difference can be expressed as a vector, such as (1000-1500).

[0135] In this example, statistical analysis is performed to determine the field placement probability for the "Dose" column. For example, in diabetes treatment information, the probability of records with a dose between 1000 and 1500 mg / day is 70%. Combined with the previously calculated "Dose" field difference vector of -500, for example, if the absolute value of the difference vector is within 1000 and the corresponding probability reaches 60%, the average of the two doses is taken as the new field value. Similarly, based on the field placement probability for each column, the field difference vectors for the other columns are processed to obtain new fields.

[0136] If the occurrence probability of "metformin" is 100%, then the field is kept unchanged.

[0137] In this embodiment, for the new field obtained, such as the new field value of "drug dosage" is 1250 mg / day, its consistency with the corresponding cluster information (diabetes treatment cluster information) is judged. Since the dosage value is within the reasonable range of diabetes treatment dosage, it is considered to be consistent. Give this new field a broad definition, such as "recommended dosage for routine treatment of diabetes." Other new fields are processed in the same way, and then these new fields are freely spliced ​​and combined. For example, by splicing and combining with the new field of "symptom description" (assuming it is "typical diabetes symptoms"), the new field of "drug name" (assuming it is "metformin"), etc., new information is obtained: "Under typical diabetes symptoms, the recommended medication in the routine treatment stage is metformin, and the dosage is 1250 mg / day."

[0138] The beneficial effects of the above technical solution are: by retrieving the field blank table from the type-table comparison library, the basic expressions in the medical information are converted into fields and placed in the corresponding grid, which facilitates the construction of field difference vectors and processing to obtain new fields, effectively integrating the related information in different information records under the same treatment type, and processing the field difference vectors according to the field placement probability to obtain new fields. It can explore the potential rules and values ​​in the medical information, and by judging the unity of the new fields and cluster information and giving them a broad definition, new information can be obtained by free splicing and combination, thereby expanding the connotation of the original medical information.

[0139] The present invention provides a clinical decision support method based on medical information entry, which sets a full cycle for the target patient according to the medical scenario and in combination with a multi-source medical database, including:

[0140] Analyzing the medical scenario to obtain a number of scenario parameters, and comparing and analyzing each scenario parameter with each piece of information in the multi-source medical data to obtain a comparison vector for each piece of information;

[0141] Selecting the best vector and the second best vector from all comparison vectors to obtain first reference information and second reference information;

[0142] A full cycle is set for information fusion of the first reference information and the second reference information.

[0143] In this embodiment, it is assumed that the medical scenario is a hypertension scenario, and the scenario parameters involved include: symptom parameters (headache, dizziness, palpitations, etc.), examination item parameters (blood pressure measurement, electrocardiogram, blood biochemistry test, etc.), treatment method parameters (drug therapy, lifestyle intervention, etc., drugs such as calcium channel blockers, angiotensin-converting enzyme inhibitors, etc.), etc.

[0144] Information record A: "The patient has headache symptoms and blood pressure is 160 / 100 mmHg. He is treated with nifedipine extended-release tablets (calcium channel blocker) and a low-salt diet is recommended."

[0145] Information record B: "The patient experienced dizziness, the electrocardiogram showed normal results, and the blood biochemical test showed no abnormalities. Enalapril tablets (angiotensin-converting enzyme inhibitor) were used for treatment, and increased exercise was recommended.

[0146] Compare and analyze the scenario parameters with each piece of information. For example, consider the symptom parameter. Record A contains "headache." Compare this to the scenario parameters "headache, dizziness, palpitations." A comparison vector for the symptom is constructed. For example, using binary notation, the presence of the symptom is marked as 1, and its absence as 0, resulting in a vector (1, 0, 0). Record B contains "dizziness," resulting in a vector (0, 1, 0). For examination item parameters, record A contains "blood pressure measurement," resulting in a vector (1, 0, 0), while record B contains "electrocardiogram" and "blood biochemistry test," resulting in a vector (1, 1, 0). For treatment parameters, record A uses "calcium channel blockers" and "low-salt diet intervention," resulting in a vector (1, 0, 1, 0), while record B uses "angiotensin-converting enzyme inhibitors" and "exercise intervention," resulting in a vector (0, 1, 0, 1). Finally, the comparison vector for each piece of information is obtained. Assuming that weighted summation is performed (weights are set according to the importance of each parameter, such as symptom 0.3, examination item 0.3, and treatment method 0.4), the comprehensive comparison vector of information record A is (0.6, 0.3, 0.7), and the comprehensive comparison vector of information record B is (0.5, 0.6, 0.6) (the numerical value here represents the degree of matching, ranging from 0 to 1).

[0147] Suppose there are eight information records, resulting in eight comparison vectors: (0.6, 0.3, 0.7), (0.5, 0.6, 0.6), (0.4, 0.4, 0.5), (0.5, 0.5, 0.5), (0.6, 0.5, 0.6), (0.7, 0.4, 0.7), (0.5, 0.3, 0.5), (0.6, 0.6, 0.6). By summing the values ​​of each dimension of the vector, the scores of each vector are 1.6, 1.7, 1.3, 1.5, 1.7, 1.8, 1.3, and 1.8, respectively. The information corresponding to the vector with the highest score is selected as the optimal vector and the first reference information. (Here, the two information corresponding to the vector with a score of 1.8 are used as the required information.)

[0148] Suppose the first piece of information is "The patient experienced headaches and had a blood pressure of 160 / 100 mmHg. He was treated with nifedipine extended-release tablets, initially at a dose of 10 mg twice daily. After two weeks, his blood pressure dropped to 140 / 90 mmHg. A low-salt diet was recommended." The second piece of information is "The patient's blood pressure was 150 / 95 mmHg, with no obvious symptoms. He was treated with enalapril tablets, initially at a dose of 5 mg once daily. After three weeks, his blood pressure stabilized at 135 / 85 mmHg. He was advised to increase exercise and maintain a regular sleep and rest schedule." These two pieces of information are then merged to extract common and complementary information. Regarding the treatment cycle, considering the onset of action and blood pressure control of both drugs, the initial two-to-three weeks are designated as a dose adjustment period, during which blood pressure is closely monitored and the dose is fine-tuned based on blood pressure readings. The next four-to-eight weeks are designated as a blood pressure stabilization observation period, during which the drug dose is maintained and lifestyle interventions are continued. Finally, depending on the long-term stability of blood pressure, a maintenance period may be entered, during which the drug regimen and lifestyle adjustments are further optimized. For example, if blood pressure remains stable within the normal range (systolic pressure <140 mmHg and diastolic pressure <90 mmHg), consideration may be given to reducing the drug dose or trying single-drug maintenance therapy.

[0149] The beneficial effects of the above technical solution are: by generating a comparison vector, the degree of match between hypertension medical scenario parameters and information in multi-source medical data can be quantified. By determining the first and second reference information, the most valuable reference information can be accurately located within a large amount of hypertension-related medical information. By setting a full cycle for information fusion between the first and second reference information, effective content related to hypertension treatment medication use, blood pressure monitoring, lifestyle intervention, and other aspects can be integrated from different reference information.

[0150] The present invention provides a clinical decision-making support method based on medical information input, which obtains an initial decision for the medical request, including:

[0151] performing a correlation analysis on the corresponding provided predictive information and the feedback predictive information under the same medical management link according to the interaction relationship between the doctor and the patient, thereby obtaining a first distribution of the first comparison information in the provided predictive information and a second distribution of the second comparison information in the feedback predictive information;

[0152] Obtaining a patient compliance coefficient to a doctor based on the degree of consistency between the first distribution and the second distribution, and determining a relative independence coefficient based on the first independent information in the provided predictive information and the second independent information in the feedback predictive information;

[0153] According to the one-to-one correspondence of compliance coefficient and relative independence coefficient, the service information for the same patient in the same medical management link is weighted and ranked, and the service information with the highest weight is retained;

[0154] Determine the reference information of the entire cycle, split the service information and reference information according to the medical management link, and fuse the information of the same medical management link. Standardize the corresponding fused information according to the management indicators under each medical management link to obtain an initial decision.

[0155] The interactive relationship between doctors and patients refers to the fact that the two aspects of information, the treatment process and treatment expectations conveyed by the doctor, and the patient's concern for their own health and feedback expectations, have a matching set relationship in some key information. For example, the first control information: taking medicine on time can effectively control blood pressure and avoid complications; the second control information: will try to take medicine on time, but worry about the side effects of the medicine; at this time, the frequency and distribution of keywords such as "take medicine on time", "control blood pressure" and "complications" in the first control information are counted to obtain the first distribution, and the frequency and distribution of keywords such as "take medicine on time" and "drug side effects" in the second control information are counted to obtain the second distribution.

[0156] The degree of consistency refers to whether there are keywords in the second distribution with similar meanings to those in the first distribution. For example, the individual consistency coefficient of taking medicine on time is 1, the individual consistency coefficient of complications and drug side effects is 1, and the individual consistency coefficient of controlling blood pressure is 0. At this time, the degree of consistency is: (1+1+0) / (1+1+1). It should be noted that the degree of consistency = the total number of keywords with similar meanings involved in the first distribution and the second distribution / the total number of keywords involved in the first distribution.

[0157] At this time, the compliance coefficient=the degree of consistency×ln(2+the number of remaining keywords excluding those with similar meanings in the second distribution / the total number of keywords in the second distribution).

[0158] In this embodiment, independent information refers to information that has no interactive relationship.

[0159] In this embodiment, the relative independence coefficient=1-the correlation coefficient between the independent information and the information in the corresponding predictive information that has an interactive relationship, and the correlation coefficient is calculated using a semantic distance algorithm.

[0160] In this embodiment, the weight is calculated as: compliance coefficient×ln(2+relative independence coefficient of the second independent information / (1+relative independence coefficient of the first independent information)).

[0161] Determine reference information for the entire hypertension treatment cycle, such as diagnostic criteria and medication principles at different stages. Separate the retained service information and reference information according to the medical management process (such as diagnosis, medication, and follow-up). In the medication treatment process, integrate service information (such as "taking medication on time can effectively control blood pressure and avoid complications") with the reference information for that process (such as the dosage adjustment principles for hypertension medication). Then, standardize the integrated information based on the management indicators of that medical management process (such as blood pressure control targets and medication compliance requirements). For example, convert the blood pressure control target into a specific numerical range (systolic blood pressure <140 mmHg and diastolic blood pressure <90 mmHg), and convert the medication compliance requirement into a standard such as an 80% medication compliance rate. Ultimately, the initial decision for the patient in the medication treatment process is made, such as recommending that the patient take medication on time and strictly follow the current dosage to achieve the blood pressure control target and medication compliance requirements.

[0162] The beneficial effects of the above technical solution are: by obtaining information distribution through correlation analysis, it is possible to clearly understand the presentation and differences in key content between the information provided by doctors and the information fed back by patients, calculate the obedience coefficient and the relative independence coefficient, quantify the degree of patient obedience to the doctor's advice and the degree of difference between the independent information of doctors and patients, obtain initial decisions through information fusion and standardization, and integrate service information and reference information within the medical management link.

[0163] The present invention provides a clinical decision support method based on medical information input, which optimizes the initial decision to obtain an auxiliary decision, including:

[0164] Perform feature analysis on doctor characteristics and historical health characteristics based on the management indicators under each medical management link to obtain the array to be corrected for each management indicator;

[0165] The initial decision is optimized according to the array to be corrected to obtain an auxiliary decision.

[0166] In this embodiment, it is assumed that in the drug adjustment management link of hypertension treatment, the management indicators include blood pressure control targets (systolic blood pressure <140 mmHg and diastolic blood pressure <90 mmHg), drug safety indicators (such as normal range of liver and kidney function indicators), etc. Analysis of doctor characteristics: The doctor's characteristics are assumed to be good at hypertension treatment, with 100 surgical cases (the surgical volume here can be compared to the number of cases dealing with complex hypertension conditions), and a surgical success rate of 90%. According to the management indicators, the doctor's ability and experience in this link are analyzed. For example, for the blood pressure control target, the proportion of doctors who achieved this target in past cases is evaluated, assuming it is 80%. These evaluation results are organized into array form. For the blood pressure control target management indicator, the array to be revised may be [80% (achievement rate), 100 (number of cases), 99% (success rate)].

[0167] Analysis of historical health characteristics: Assume that the patient's historical health characteristics include a family history of coronary heart disease and mild renal insufficiency (serum creatinine level 120 μmol / L, normal range 53-106 μmol / L). Based on the drug safety index (normal range of liver and kidney function indicators), analyze the impact of the patient's historical health characteristics on this index. Organize the relevant data into an array. For the drug safety management index, the array to be revised may be [120 (serum creatinine level) family history of coronary heart disease (qualitative information can be quantified as 1 for presence and 0 for absence)].

[0168] Based on the set of physician characteristics to be revised, the previous compliance rate was 80%, indicating that the physician had some experience in blood pressure control but did not fully meet the target. Considering the patient's family history of coronary artery disease, appropriate adjustments to medication dosages or combination therapy may be considered to reduce the risk of cardiovascular events due to poor blood pressure control. Based on the set of historical health characteristics to be revised, the patient's serum creatinine level was above the normal range. Although nifedipine extended-release tablets have a relatively minor effect on renal function, caution is still warranted. Consideration may be given to appropriately reducing the dose of nifedipine extended-release tablets and combining them with medications that have a minimal effect on renal function and provide cardioprotective effects, such as irbesartan tablets (initial dose 75 mg once daily).

[0169] At this time, the initial decision of "continue with the current dose of 10 mg of nifedipine extended-release tablets twice a day" was optimized to the final auxiliary decision of "adjust the dose of nifedipine extended-release tablets to 5 mg twice a day, combined with 75 mg of irbesartan tablets once a day, and closely monitor blood pressure and renal function indicators."

[0170] The beneficial effect of the above technical solution is: by analyzing the doctor's characteristics and historical health characteristics based on management indicators to obtain a set of data to be revised, the doctor's ability and experience and the patient's own health status related information can be used to optimize the initial decision according to the set of data to be revised to obtain an auxiliary decision, which can fully consider the doctor's professional ability and the individual health differences of the patient, and improve the accuracy of clinical decision-making.

[0171] The present invention provides a clinical decision support device based on medical information input, such as Figure 2 Shown, including:

[0172] The scenario analysis module is used to build a multi-source medical database, obtain the medical demands of target patients, and identify the medical scenarios of the target patients;

[0173] a cycle construction module, configured to set a full cycle for the target patient according to the medical scenario and in combination with a multi-source medical database, and sequentially input the service information of each medical management link of the target patient in the full cycle to form full cycle management information, wherein the service information in each medical management link includes the predictive information provided by the doctor to the patient and the predictive information of the patient's feedback to himself;

[0174] An initial decision acquisition module is used to standardize the provided predictive information and the feedback predictive information under the same medical management link to obtain an initial decision for the medical demand;

[0175] The decision optimization module is used to obtain the characteristics of the doctor who can solve the medical needs of the target patient, and combine the historical health characteristics of the target patient to optimize the initial decision to obtain an auxiliary decision and output it, wherein the auxiliary decision is related to the treatment medication.

[0176] The beneficial effects of the above technical solution are: building a multi-source medical database and realizing the integration of full-cycle medical management information, conducting standardized analysis of patients' medical demands and combining multiple factors to optimize decision-making, which can effectively make up for the shortcomings of the traditional medical model and improve the quality and efficiency of medical services.

[0177] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A clinical decision-making support method based on medical information input, characterized in that: include: Step 1: Build a multi-source medical database, obtain the medical needs of target patients, and identify the medical scenarios of the target patients; Step 2: According to the medical scenario and in combination with the multi-source medical database, a full cycle is set for the target patient, and the service information of each medical management link of the target patient in the full cycle is sequentially entered to form full cycle management information, wherein the service information in each medical management link includes the predictive information provided by the doctor to the patient and the predictive feedback information of the patient to himself; Step 3: Standardize the provided predictive information and the feedback predictive information in the same medical management link to obtain an initial decision for the medical demand; Step 4: Obtain the characteristics of the doctor who can solve the medical needs of the target patient, and combine the historical health characteristics of the target patient to optimize the initial decision to obtain an auxiliary decision and output it, wherein the auxiliary decision is related to the treatment medication.

2. The clinical decision-making support method based on medical information input according to claim 1, characterized in that: Build a multi-source medical database, including: Capturing existing historical patient medical information from different medical platforms and determining information attributes of the medical information, wherein the information attributes include: a complete attribute and an incomplete attribute, and treating the medical information with the complete attribute as first information and treating the medical information with the incomplete attribute as second information; Performing a first cluster analysis on all first information to obtain a plurality of first clusters, and simultaneously performing a second cluster analysis on all second information, eliminating cluster results in which the amount of second information present in the second cluster results is less than a specified amount, and retaining the remaining results to obtain a plurality of second clusters; Matching and analyzing the cluster information of each second cluster with the cluster information of the first cluster, filtering the cluster information of the second clusters with a matching degree greater than a preset degree and the clustering results of the second clusters according to the matching degree, and adding them to the clustering results of the first cluster to obtain a first new result; Performing probability distribution and information reconstruction on each first new result to obtain new information; A multi-source medical database is constructed based on all the first information and the new information, wherein the multi-source medical database includes: a plurality of reference information, and each reference information includes a treatment purpose, a treatment cycle consistent with the treatment purpose, treatment drugs for each treatment stage under the treatment cycle, treatment uses, and treatment effects.

3. The clinical decision-making support method based on medical information input according to claim 2, characterized in that: The clustering results of the second cluster are screened based on the matching degree, including: Selecting the division distance length consistent with the matching degree from the matching degree-distance comparison table; Draw a circle with the division distance as the radius and the point where the corresponding cluster information is located as the origin, and determine the shape relationship between each information point inside the circle except the point corresponding to the cluster information and the two nearest points; If it is a straight line relationship, assigning a first mark to the corresponding point according to a first ratio of the length of the corresponding straight line to the divided distance, and in combination with an angular relationship between the straight line relationship and the origin; Where D1 represents the first marking result; Lz represents the length of the corresponding straight line; Lh represents the division distance; represents the first ratio; θ represents the angle between the two endpoints of the straight line relationship and the origin, and is less than or equal to 180°; a1 represents the first marking threshold; If it is a triangular relationship, assign a second label to the corresponding point based on the area ratio of the corresponding triangle to the circle and the distance between the corresponding point and the corresponding information point in the corresponding cluster information; Where D2 represents the second marking result; SΔ represents the area of ​​the corresponding triangle; π×Lh 2 represents the area of ​​the circle; Ld represents the distance between the corresponding cluster information point and the corresponding information point; a2 represents the second marking threshold; According to the first marking result and the second marking result, a first point farthest from the corresponding point of the corresponding cluster information is selected from the valuable marks, and a second point closest to the corresponding point of the corresponding cluster information is selected from the worthless marks; If the first point is closer to the cluster information corresponding point than the second point, or the first point and the second point are equally far from the cluster information corresponding point, then the information of the point with the value mark is used as the screening information; If the second point is closer to the cluster information corresponding point than the first point, then a first new circle is drawn with the distance between the second point and the cluster information corresponding point as the new radius and the cluster information corresponding point as the origin; Divide the first new circle into four quadrants, locate the third point farthest from the origin in the quadrant with the maximum quadrant density, and determine the length to be divided; Where Lave represents the average value of the length farthest from the origin in the remaining three quadrants; w1 represents the maximum quadrant density; w2 represents the minimum quadrant density; sumw represents the sum of all quadrant densities; σL 2 represents the variance of the distance between each information point and the origin in the first new circle; L3 represents the distance between the third point and the origin; rx represents the length of the new radius; Lr represents the length to be divided; A second new circle is drawn with the length to be divided as the radius and the corresponding point of the cluster information as the origin; The information of the points with value marks within the second new circle is used as screening information.

4. The clinical decision-making support method based on medical information input according to claim 3, characterized in that: Probability distribution and information reconstruction are performed on each first new result to obtain new information, including: Retrieving a field blank table from a type-table comparison library according to the treatment type of the cluster information in the first new result, wherein the field blank table includes a plurality of basic placement cells; Perform field conversion on each basic expression in each piece of information in the first new result and place it in the comparison basic placement grid, and construct a field difference vector for each column of information in the placement table; According to the field placement probability of each column of information, the field difference vector is processed to obtain a new field; A broad definition of each new field is obtained based on its unity with the corresponding cluster information, and new information is obtained by freely splicing and combining them.

5. The clinical decision-making support method based on medical information entry according to claim 1, characterized in that: According to the medical scenario and in combination with the multi-source medical database, a full cycle is set for the target patient, including: Analyzing the medical scenario to obtain a number of scenario parameters, and comparing and analyzing each scenario parameter with each piece of information in the multi-source medical data to obtain a comparison vector for each piece of information; Selecting the best vector and the second best vector from all comparison vectors to obtain first reference information and second reference information; A full cycle is set for information fusion of the first reference information and the second reference information.

6. The clinical decision-making support method based on medical information input according to claim 1, characterized in that: Obtain an initial decision regarding the medical need, including: performing a correlation analysis on the corresponding provided predictive information and the feedback predictive information under the same medical management link according to the interaction relationship between the doctor and the patient, thereby obtaining a first distribution of the first comparison information in the provided predictive information and a second distribution of the second comparison information in the feedback predictive information; Obtaining a patient compliance coefficient to a doctor based on the degree of consistency between the first distribution and the second distribution, and determining a relative independence coefficient based on the first independent information in the provided predictive information and the second independent information in the feedback predictive information; According to the one-to-one correspondence of compliance coefficient and relative independence coefficient, the service information for the same patient in the same medical management link is weighted and ranked, and the service information with the highest weight is retained; Determine the reference information of the entire cycle, split the service information and reference information according to the medical management link, and fuse the information of the same medical management link. Standardize the corresponding fused information according to the management indicators under each medical management link to obtain an initial decision.

7. The clinical decision-making support method based on medical information entry according to claim 1, characterized in that: Optimize the initial decision to obtain auxiliary decision, including: Perform feature analysis on doctor characteristics and historical health characteristics based on the management indicators under each medical management link to obtain the array to be corrected for each management indicator; The initial decision is optimized according to the array to be corrected to obtain an auxiliary decision.

8. A clinical decision support device based on medical information input, characterized in that: include: The scenario analysis module is used to build a multi-source medical database, obtain the medical demands of target patients, and identify the medical scenarios of the target patients; a cycle construction module, configured to set a full cycle for the target patient according to the medical scenario and in combination with a multi-source medical database, and sequentially input the service information of each medical management link of the target patient in the full cycle to form full cycle management information, wherein the service information in each medical management link includes the predictive information provided by the doctor to the patient and the predictive information of the patient's feedback to himself; An initial decision acquisition module is used to standardize the provided predictive information and the feedback predictive information under the same medical management link to obtain an initial decision for the medical demand; The decision optimization module is used to obtain the characteristics of the doctor who can solve the medical needs of the target patient, and combine the historical health characteristics of the target patient to optimize the initial decision to obtain an auxiliary decision and output it, wherein the auxiliary decision is related to the treatment medication.

Images