An intelligent family doctor recommendation method, system and storage medium

Abstract

This invention belongs to the field of medical big data system technology, specifically relating to an intelligent family doctor recommendation method, system, and storage medium. This invention uses a dynamic comprehensive score to represent the matching degree between doctors and patients. The specific indicators included in the dynamic comprehensive score are capability matching degree, spatiotemporal matching degree, and economic matching degree. This invention can dynamically assess the matching degree between patients and doctors, fully utilize existing medical big data resources, make objective and accurate judgments, thereby assisting patients in efficiently finding the most suitable family doctor, and has excellent application prospects.

Description

Technical Field

[0001] This invention belongs to the field of medical big data system technology, specifically relating to an intelligent family doctor recommendation method, system, and storage medium. Background Technology

[0002] In the traditional medical service model, patients need to go through a series of steps: registration, waiting, consultation, examination, waiting for results, payment, treatment, medication pickup, discharge, and follow-up consultation. The time spent on non-medical procedures far exceeds the time spent on treatment. With the development of medical and internet technologies, introducing data-driven and information-based platform systems into the healthcare industry can help improve medical technology, share digital medical resources and information, and thus solve the aforementioned problems.

[0003] The family doctor service system enables tiered medical care and plays a significant role in reducing medical costs, optimizing the use of health resources, and improving the overall health of the population. A family doctor is a team composed of general practitioners, team secretaries, specialists, and public health physicians, who typically provide services to their contracted patients in a team format.

[0004] Family doctors provide comprehensive, continuous, effective, timely, and personalized healthcare services to their clients. They possess comprehensive and systematic knowledge of prevention, health maintenance, medical treatment, and rehabilitation, and have strong verbal communication, interpersonal, and coordination skills. They can provide timely and effective health services throughout the entire life cycle. Family doctors offer personalized prevention, health maintenance, treatment, rehabilitation, and health education services and guidance, enabling patients to address their daily health problems and healthcare needs, and receive home-based treatment and rehabilitation care without leaving their homes.

[0005] Current technologies lack a comprehensive family doctor recommendation and management system that fully utilizes medical big data. Specifically, this manifests in several ways: the role of artificial intelligence is not considered; the dynamic correlation between family doctor big data and overall medical big data is poor, hindering the provision of family doctor services; and existing family doctor systems still offer standardized services without matching optimal medical resources to each individual patient, potentially leading to situations where patients cannot be matched with a suitable family doctor. Therefore, developing a new intelligent family doctor recommendation system to achieve personalized family doctor matching for individual patients is a pressing issue that needs to be addressed in this field. Summary of the Invention

[0006] To address the problems of existing technologies, this invention provides an intelligent family doctor recommendation method, system, and storage medium.

[0007] An intelligent family doctor recommendation method, wherein the method is based on a dynamic comprehensive score representing the matching degree between doctors and patients, and the calculation formula of the dynamic comprehensive score is as follows:

[0008]

[0009] Among them, the As weight, For specific indicators, Specifically selected from the following three items: Capability Matching Degree Spatiotemporal matching degree Economic matching degree .

[0010] Preferably, the formula for calculating the capability matching degree is:

[0011]

[0012] in, Rate doctors' history To achieve order saturation, The degree of disease association is calculated based on a medical knowledge graph;

[0013] The The calculation method is as follows:

[0014]

[0015] in, This indicates a set of chronic diseases suffered by the patient. Indicates belonging to A specific disease in the context; This indicates the doctor's area of ​​expertise. Indicates belonging to A specific disease in the context; To associate diseases in medical knowledge graphs Related diagnostic and treatment pathways This indicates starting from the current disease node in the medical knowledge graph. The longest path length to the associated node.

[0016] Preferably, the specific disease type is represented based on ICD-10 coding.

[0017] Preferably, the formula for calculating the spatiotemporal matching degree is:

[0018]

[0019] in, The patient's place of residence (latitude and longitude) The latitude and longitude of the doctor's residence; The Haversine distance is calculated based on the latitude and longitude of the doctor's and patient's residences, in km. For distance attenuation factor, Expected response time; The average response time is expressed in hours (h). This represents the maximum response time allowed by the system for doctors.

[0020] Preferably, the formula for calculating the economic matching degree is:

[0021]

[0022] in, Budgeting for patients, Doctor's house call fees Assess patient compliance.

[0023] Preferably, the weights are dynamically adjusted, and the dynamic adjustment method is as follows:

[0024] like The importance of the specific indicators, ranked from highest to lowest, is as follows: capability matching degree, spatiotemporal matching degree, and economic matching degree.

[0025] like =3, then the importance ranking of the specific indicators is that the ability matching degree is higher than the economic matching degree and the time-space matching degree;

[0026] like The importance of the specific indicators, ranked from highest to lowest, is as follows: economic matching degree, capability matching degree, and spatiotemporal matching degree.

[0027] in, In terms of urgency.

[0028] Preferably, the dynamic adjustment of the weights is performed as follows:

[0029] like The weights of economic matching, capability matching, and spatiotemporal matching are as follows: ;

[0030] like =3, and the weights of economic matching degree, capability matching degree and spatiotemporal matching degree are W=[0.3,0.4,0.3] respectively;

[0031] like The weights of economic matching, capability matching, and spatiotemporal matching are as follows: .

[0032] Preferred, according to Sort the list of doctors in descending order and recommend the top-ranked doctor to the patient; if there are multiple doctors... If all of them rank first, then all of them are recommended for the patient to choose from.

[0033] This invention also provides an intelligent family doctor recommendation method system, which integrates modules for implementing the above-mentioned intelligent family doctor recommendation method, including:

[0034] The data acquisition module is configured to collect data from patients and doctors and integrate it with medical knowledge graph data.

[0035] The dynamic comprehensive score calculation module is configured to calculate the dynamic comprehensive score.

[0036] The recommendation module is configured to recommend doctors to patients based on the dynamic comprehensive score.

[0037] The present invention also provides a computer-readable storage medium having stored thereon a computer program for implementing the above-described intelligent family doctor recommendation method.

[0038] This invention addresses the issue of family doctor recommendations by constructing a system for assessing the matching degree between patients and doctors. This system includes ability matching, spatiotemporal matching, and economic matching, and dynamically adjusts the weights of key indicators under different circumstances. The assessment and recommendation system of this invention can fully utilize existing medical big data resources to objectively and accurately judge the degree of doctor-patient matching, thereby helping patients efficiently find the most suitable family doctor. The method and system of this invention can be integrated into various medical-related management systems, application software, and mobile apps, showing great application potential.

[0039] Obviously, based on the above description of the present invention, and according to common technical knowledge and conventional methods in the field, various other modifications, substitutions or alterations can be made without departing from the basic technical concept of the present invention.

[0040] The following detailed embodiments further illustrate the above-described content of the present invention. However, this should not be construed as limiting the scope of the present invention to the following examples. All technologies implemented based on the above-described content of the present invention fall within the scope of the present invention. Detailed Implementation

[0041] It should be noted that the algorithms for data acquisition, transmission, storage and processing steps not specifically described in the embodiments, as well as the hardware structures and circuit connections not specifically described, can all be implemented using content already disclosed in the prior art.

[0042] Example 1: Intelligent Family Doctor Recommendation System and Method

[0043] The system and method of this embodiment are used to recommend the most suitable family doctor to a patient. The system components include:

[0044] The data acquisition module is configured to collect data from patients and doctors and integrate it with medical knowledge graph data.

[0045] The dynamic comprehensive score calculation module is configured to calculate the dynamic comprehensive score.

[0046] The recommendation module is configured to recommend doctors to patients based on the dynamic comprehensive score.

[0047] The specific steps for using this system to make family doctor recommendations include the following:

[0048] Step 1: Data Input and Preprocessing

[0049] 1. Input of patient quantitative indicators: Input the patient's basic information, including budget. (Unit: Yuan) Collection of Chronic Diseases (A set of diseases based on ICD-10 codes, such as E11.9 = type 2 diabetes), urgency level (Graded into 1-5 levels, 1 = routine consultation, 5 = critical illness, automatically determined based on symptom keywords), compliance score Latitude and longitude of the patient's residence .

[0050] The value of Ep can be assigned with reference to the "Emergency Triage Standards" (2018), with a value of 1-5. The specific standards are as follows:

[0051] 5: Keyword matching: "chest pain", "difficulty breathing", "loss of consciousness", "massive bleeding", "severe burns", "convulsions", "stroke symptoms", "cardiac arrest";

[0052] 4: Keyword matching: "high fever", "severe abdominal pain", "persistent vomiting", "severe headache", "fracture", "difficulty breathing but not severe", "allergic reaction";

[0053] 3: Keyword matching: "fever", "cough", "diarrhea", "mild pain", "rash", "dizziness", "nausea";

[0054] 2: Keyword matching: "chronic disease follow-up visit", "routine examination", "prescription", "health consultation", "immunization";

[0055] 1: Keyword matching: "interpretation of physical examination report", "health management consultation", "lifestyle guidance", "psychological support".

[0056] Patient adherence is calculated as a weighted average of medication adherence, examination completion rate, and follow-up completion rate. Medication adherence is scored by weighting the examination completion rate (number of examinations completed on time after the initial visit / total number of visits) and the follow-up completion rate (number of follow-up visits attended on time after the initial visit / total number of visits). The weights used in calculating patient adherence can be adaptively set based on the matching effect between the calculation results and the system.

[0057] 2. Quantitative Indicator Input for Doctors: Retrieves doctor information from the doctor database, including consultation fees. (Unit: Yuan / time), Specialized Fields (ICD-10 mapping of disease specialties, such as I10 = hypertension), physician history score (Patient rating average, 0-5 points), order saturation (Current workload percentage, 0%-100%, 100% = full load), average response time (Unit: hours), latitude and longitude of the doctor's residence .

[0058] The method for calculating order saturation is as follows:

[0059] Order saturation = Actual number of patients received / Theoretical maximum number of patients received;

[0060] Actual number of patients treated = odd number The average consultation time per case (e.g., 30 minutes) and the theoretical maximum number of cases are based on the doctor's working hours (e.g., 8 hours per day).

[0061] Step 2: Deep matching based on medical knowledge graph

[0062] 1. Calculate the disease association degree

[0063] Collect patients' chronic diseases With medical specialty fields Mapping to a medical knowledge graph (which uses existing medical knowledge graphs such as Clinical Knowledge Graph, Hetionet, and Bio2RDF, and can be built based on ICD-10 and clinical pathway libraries, containing relationships between diseases and treatment pathways), calculate the disease association degree:

[0064]

[0065] in, To link diseases in a medical knowledge graph Related treatment pathways typically include common complications, related diseases requiring co-treatment, relevant examinations, medications, and treatment plans (i.e., related to...). A set of directly related diseases). Indicates from the current disease node The longest path length to the associated node (such as complications or treatment methods). The greater the depth, the weaker the association.

[0066] For example, when a diabetic patient (E11.9) is matched with an endocrinologist, the numerator is the number of intersections between the patient's disease and the doctor's specialty path (e.g., diabetes + complications); the denominator is the total number of the patient's diseases × path depth (the deeper the depth, the weaker the association, because a deeper depth indicates that the doctor may have less direct association with the disease and more indirect association).

[0067] For example:

[0068] Patient disease collection: = {E11.9 (Type 2 Diabetes), I10 (Primary Hypertension)};

[0069] Collection of doctors' specialties: = {E11.9 (Type 2 Diabetes), I25.1 (Coronary Artery Disease)};

[0070] The path to obtaining doctors' expertise from the constructed medical knowledge graph is as follows:

[0071] Path(E11.9) = {

[0072] "Core Treatments": ["Glucose Monitoring", "Insulin Therapy", "Oral Hypoglycemic Agents"], #Depth=1.0

[0073] "Direct Complications": ["Diabetic Nephropathy", "Retinopathy", "Neuropathy"], #Depth=1.2

[0074] "Metabolic Association": ["Hypertension Management", "Dyslipidemia Management"], #Depth=1.5

[0075] "Long-term Management": ["Foot Care", "Nutritional Guidance", "Sports Rehabilitation"]#Depth=1.3}

[0076] Path(I25.1) = {

[0077] "Core Treatment": ["Antibacterial Therapy", "Beta-Blockers", "Revascularization"], # Depth=1.0

[0078] "Risk Factors": ["Hypertension Management", "Diabetes Management", "Lipid Management"], # Depth=1.4

[0079] "Complications": ["Heart Failure", "Arrhythmia", "Prevention of Sudden Cardiac Death"], # Depth=1.6

[0080] "Rehabilitation Management": ["Cardiac Rehabilitation", "Lifestyle Intervention"]# Depth=1.3}

[0081] The depth and weight of each level are quantitatively assessed based on the direct relevance of clinical pathways when constructing the knowledge graph. For example, if Path_Depth_E11.9 = (1.0... 3 + 1.2 3 + 1.5 2 + 1.3 3) / 11 = 1.27, and using the same calculation method, Path_Depth_I25.1 = 1.36, so the doctor's average Path_Depth = 1.315.

[0082] According to the formula

[0083]

[0084] For doctors specializing in E11.9 (diabetes): ∩ Path(E11.9) = {E11.9, I10} ∩{E11.9, diabetic nephropathy, retinopathy, neuropathy, hypertension management, dyslipidemia management, ...}, matching diseases: E11.9 (direct match), I10 (associated through "hypertension management"), with an intersection of 2.

[0085] For doctors specializing in I25.1 (coronary artery disease): ∩ Path(I25.1) = {E11.9, I10} ∩{I25.1, Hypertension Management, Diabetes Management, Lipid Management, Heart Failure, ...}, matching diseases: I10 (associated through "Hypertension Management"), E11.9 (associated through "Diabetes Management"), with an intersection of 2.

[0086] The intersection after merging is |{E11.9, I10}| = 2.

[0087] denominator = 2, Path_Depth = 1.315, = 0.76.

[0088] 2. Dynamic weight adaptive mechanism

[0089] like The importance of the specific indicators, ranked from highest to lowest, is as follows: capability matching degree, spatiotemporal matching degree, and economic matching degree.

[0090] like =3, then the importance ranking of the specific indicators is that the ability matching degree is higher than the economic matching degree and the time-space matching degree;

[0091] like The importance of the specific indicators, ranked from highest to lowest, is as follows: economic matching degree, capability matching degree, and spatiotemporal matching degree.

[0092] For example, in one embodiment, based on the patient's urgency level Automatically adjust matching dimension weights:

[0093] like The weights of economic matching, capability matching, and spatiotemporal matching are as follows: ;

[0094] like =3, and the weights of economic matching degree, capability matching degree and spatiotemporal matching degree are W=[0.3,0.4,0.3] respectively;

[0095] like The weights of economic matching, capability matching, and spatiotemporal matching are as follows: .

[0096] Step 3: Multimodal matching degree calculation

[0097] 1. Ability matching degree

[0098]

[0099] in, Rate the doctor's history, for example, a score of 4.8. (1+4.8)≈1.76, To determine the saturation level for order fulfillment, if the saturation level is 80%, then (1 / 1 + 0.8) ≈ 0.55.

[0100] 2. Spatiotemporal matching degree

[0101]

[0102] in, Calculate the Haversine distance (km) based on the latitude and longitude of the doctor's and patient's addresses. This is the distance attenuation factor (default is 0.02). This represents the expected response time. The average response time is expressed in hours (h). This represents the maximum allowable response time for doctors, i.e., the maximum tolerable time difference. For non-emergency services, maxRT can be set to 2 hours; for emergency services, maxRT can be set to half an hour. For example: Ep=1, maxRT=2; Ep=2, maxRT=1.5; Ep=1, maxRT=1.0; Ep=1, maxRT=0.7; Ep=1, maxRT=0.5.

[0103] 3. Economic matching degree

[0104]

[0105] Among them, when the patient's budget Doctor's unit price is greater than or equal to When the budget is insufficient, the match score is 1. When the budget is insufficient, the match score is reduced proportionally and multiplied by a coefficient (0.5 + 0.5 × patient compliance score). This means that patients with high compliance can achieve a high degree of economic matching even if their budget is limited.

[0106] Step 4: Dynamic Comprehensive Scoring and Recommendation

[0107] .

[0108] Step 5: Explainable Recommendation Output

[0109] 1. According to Sort the list of doctors in descending order and recommend the top-ranked doctor to the patient.

[0110] 2. If there are multiple doctors If all doctors rank first, multiple doctors will be recommended for the patient to choose from.

[0111] As can be seen from the above embodiments, the present invention can dynamically assess the matching degree between patients and doctors, make full use of existing medical big data resources, make objective and accurate judgments, thereby helping patients efficiently find the most suitable family doctor, and has a very good application prospect.

Claims

1. An intelligent family doctor recommendation method, characterized in that: The dynamic comprehensive score represents the degree of matching between doctors and patients. The formula for calculating the dynamic comprehensive score is as follows: Among them, the As weight, For specific indicators, Specifically selected from the following three items: Capability Matching Degree Spatiotemporal matching degree Economic matching degree ; The formula for calculating the capability matching degree is: in, Rate doctors' history To achieve order saturation, The degree of disease association is calculated based on a medical knowledge graph; The The calculation method is as follows: in, This indicates a set of chronic diseases suffered by the patient. Indicates belonging to A specific disease in the context; This indicates the doctor's area of ​​expertise. Indicates belonging to A specific disease in the context; To associate diseases in medical knowledge graphs Related diagnostic and treatment pathways This indicates starting from the current disease node in the medical knowledge graph. The longest path length to the associated node; The formula for calculating the spatiotemporal matching degree is: in, The patient's place of residence (latitude and longitude) The latitude and longitude of the doctor's residence; The Haversine distance is calculated based on the latitude and longitude of the doctor's and patient's residences, in km. For distance attenuation factor, Expected response time; The average response time is expressed in hours (h). This represents the maximum response time allowed by the system for doctors. The formula for calculating the economic matching degree is: in, Budgeting for patients, Doctor's house call fees Assess patient compliance scores; according to Sort the list of doctors in descending order and recommend the top-ranked doctor to the patient; if there are multiple doctors... If all of them rank first, then all of them are recommended for the patient to choose from.

2. The intelligent family doctor recommendation method according to claim 1, characterized in that: The specific types of diseases are represented based on ICD-10 codes.

3. The intelligent family doctor recommendation method according to claim 1, characterized in that: The weights are dynamically adjusted, and the dynamic adjustment method is as follows: like The importance of the specific indicators, ranked from highest to lowest, is as follows: capability matching degree, spatiotemporal matching degree, and economic matching degree. like =3, then the importance ranking of the specific indicators is that the ability matching degree is higher than the economic matching degree and the time-space matching degree; like The importance of the specific indicators, ranked from highest to lowest, is as follows: economic matching degree, capability matching degree, and spatiotemporal matching degree. in, In terms of urgency.

4. The intelligent family doctor recommendation method according to claim 3, characterized in that: The method for dynamically adjusting the weights is as follows: like The weights of economic matching, capability matching, and spatiotemporal matching are as follows: ; like =3, and the weights of economic matching degree, capability matching degree and spatiotemporal matching degree are W=[0.3,0.4,0.3] respectively; like The weights of economic matching, capability matching, and spatiotemporal matching are as follows: .

5. An intelligent family doctor recommendation system, characterized in that, The system integrates a module for implementing the intelligent family doctor recommendation method according to any one of claims 1-4, comprising: The data acquisition module is configured to collect data from patients and doctors and integrate it with medical knowledge graph data. The dynamic comprehensive score calculation module is configured to calculate the dynamic comprehensive score. The recommendation module is configured to recommend doctors to patients based on the dynamic comprehensive score.

6. A computer-readable storage medium, characterized in that, It stores a computer program for implementing the intelligent family doctor recommendation method according to any one of claims 1-4.

Images