Clinical aid decision-making method based on artificial intelligence and related products

Abstract

The invention provides a clinical auxiliary decision-making method based on artificial intelligence and a related product, and the method comprises the steps: building a knowledge graph covering medical entities and relationships through learning authoritative medical data and medical records; a doctor inputs patient information, and data standardization and cleaning are carried out; based on the multi-dimensional information of the patient, recommending suspected diagnosis and differential diagnosis; recommending a preset examination scheme according to the suspected diagnosis and differential diagnosis results; when a doctor's advice is issued, the system checks conflicts between inspection and inspection items in real time and provides risk prompts; similar medical records are recommended based on patient clinical information, and diagnosis and treatment reference is provided for doctors. The clinical aid decision-making method based on artificial intelligence and related products have the beneficial effects that the disease list of suspected diagnosis and definite diagnosis is generated by automatically analyzing the illness state of the patient, and examination, inspection and treatment schemes are recommended in a targeted manner, so that the problems of missed diagnosis and misdiagnosis are effectively avoided.

Description

Technical Field

[0001] The present invention belongs to the field of smart medical care, and in particular, relates to an artificial intelligence-based clinical decision-making assistance method and related products. Background Art

[0002] With the continuous development of big data and artificial intelligence technologies, the application of CDSS in the medical field has received widespread attention and has become an important tool for improving medical quality and reducing medical costs;

[0003] Although existing CDSS systems have made some progress in assisting doctors in diagnosis and treatment, they still face many problems in practical applications. For example, existing CDSS systems usually rely on a large amount of structured data and have limited processing capabilities for unstructured data (such as text medical records, doctor's notes, etc.), resulting in limited diagnostic accuracy. Existing systems mostly use fixed rules or shallow models in the reasoning and decision-making process, which makes it difficult to flexibly respond to complex and changing clinical scenarios.

[0004] In addition, existing technologies are also insufficient in terms of personalized treatment plan recommendations and medical record quality control functions, and are unable to provide accurate recommendations based on individual patient differences, resulting in limited effectiveness of the system in improving medical quality and reducing misdiagnosis rates. Summary of the invention

[0005] In view of this, the present invention aims to propose an artificial intelligence-based clinical decision-making support method and related products to at least solve one problem in the background technology.

[0006] To achieve the above object, the technical solution of the present invention is achieved as follows:

[0007] In a first aspect, this solution discloses a clinical decision-making assistance method based on artificial intelligence, comprising:

[0008] By studying authoritative medical materials and medical records, we build a knowledge graph covering medical entities and relationships;

[0009] Doctors enter patient information and perform data standardization and cleaning;

[0010] Recommend suspected diagnosis and differential diagnosis based on multi-dimensional information of patients;

[0011] Recommend a preset examination plan based on the suspected diagnosis and differential diagnosis results;

[0012] When a doctor's order is issued, the system checks conflicts between tests and examination items in real time and provides risk warnings;

[0013] Recommend similar medical records based on the patient's clinical information to provide doctors with diagnosis and treatment references.

[0014] Furthermore, by learning authoritative medical information and medical records, a knowledge graph covering medical entities and relationships is constructed, including:

[0015] By studying authoritative medical textbooks, pharmacopoeias and high-quality hospital medical records, we collect medical data covering diseases, symptoms, signs, tests, drugs, and surgeries, and extract medical entities and relationships from unstructured texts;

[0016] The extracted medical entities and relations are organized into a structured knowledge graph. Graph embedding technology is used in the construction process to represent the nodes and edges in the knowledge graph as vectors, and the representation is learned by optimizing the objective function:

[0017]

[0018] Among them, eh and et are the vector representations of the head entity and the tail entity respectively, Wr is the transformation matrix of the relationship, σ(·) is the sigmoid function, and T is the set of triples in the knowledge graph.

[0019] Furthermore, the doctor enters the patient information and performs data standardization and cleaning, including:

[0020] The doctor inputs the patient's basic and clinical information, parses the input text data, and standardizes the input information using a predefined medical terminology set to ensure consistency of the information;

[0021] The input data is cleaned to remove redundant and inconsistent information, and an interpolation-based algorithm is used to handle missing values ​​to improve data integrity:

[0022]

[0023] Among them, x^i represents the estimation of the missing value, N(i) is the neighbor set of node i, wij is the influence weight of neighbor j on i, and xj is the value of the neighbor node.

[0024] Furthermore, the recommendation of suspected diagnosis and differential diagnosis based on the multi-dimensional information of the patient includes:

[0025] The most likely suspected diagnosis is recommended based on the multi-dimensional information input by the patient. When calculating the probability of each disease, the weighted contribution of each dimension is taken into account:

[0026]

[0027] Among them, P(Di|X) represents the posterior probability of disease Di, X represents the clinical information of the patient, αj is the weight of dimension j, fj(X) is the characteristic function of dimension j, and Z is the normalization factor;

[0028] After the initial diagnosis, the most likely differential diagnosis is recommended by analyzing the similarity of the current diagnosis with other possible diseases and using a distance metric to rank the diseases:

[0029]

[0030] Among them, d(Di,Dj) represents the distance between diseases Di and Dj, xik and xjk are the values ​​of diseases Di and Dj on feature k, respectively, and σk is the standard deviation of feature k.

[0031] Furthermore, the preset examination plan is recommended based on the suspected diagnosis and differential diagnosis results, including:

[0032] Recommendations for examinations and tests: Based on the preliminary diagnosis results, combined with the patient's basic information and current medical history, we recommend the examinations and tests required for diagnosis or differentiation, and balance the accuracy and cost of diagnosis based on a multi-objective optimization algorithm:

[0033]

[0034] Where y is the recommended inspection and testing combination, Cj(yj) is the cost function of inspecting item j, Rj(yj) is the risk function of inspecting item j, and λj is the parameter for weighing cost and risk;

[0035] Evaluation form and drug recommendation: Based on the diagnosis results and patient information, we recommend evaluation forms and drugs. In drug recommendations, we combine the patient's characteristics and historical medication data to recommend the most suitable drugs:

[0036]

[0037] Among them, r^ui represents the predicted score of user u for drug i, μ is the global bias, bu and bi are the biases of the user and drug respectively, qi and pu are the feature vectors of the drug and user;

[0038] Recommendation of surgical method: Based on the diagnosis and the patient's personalized information, the indications and contraindications of surgery are considered and the surgical plan is optimized through a multi-level decision tree model:

[0039]

[0040] In it, T*(X) represents the best surgical decision tree, T is the set of all possible decision trees, δ(T(Xi),yi) is the prediction accuracy of decision tree T for input Xi and true label yi, and wi is the weight of the sample.

[0041] Furthermore, when the doctor's order is issued, the system checks the conflicts between the tests and the examination items in real time and provides risk warnings, including:

[0042] When doctors issue medical orders, they can check conflicts between tests and examination items in real time. By establishing a conflict matrix, they can determine whether new items conflict with existing items based on the patient's characteristics and provide risk warnings:

[0043]

[0044] Among them, Cij represents the conflict score between items i and j, βk is the weight of feature k, and gk(i,j) is the contribution function of feature k to the conflict between items i and j.

[0045] Furthermore, the recommendation of similar medical records based on the patient's clinical information to provide a reference for diagnosis and treatment for doctors includes:

[0046] Based on the patient's clinical information, through the analysis of symptoms, diagnosis, and treatment in multiple dimensions, a similarity measurement algorithm is used to recommend medical records similar to the current patient:

[0047]

[0048] Among them, S(Xp,Xq) represents the similarity between patient p and patient q, γk is the weight of feature k, and sk(xpk,xqk) is the similarity function of feature k.

[0049] In a second aspect, the present solution discloses an electronic device, comprising a processor and a memory that is communicatively connected to the processor and is used to store executable instructions of the processor, wherein the processor is used to execute the artificial intelligence-based clinical decision-making support method described in the first aspect.

[0050] In the third aspect, the present solution discloses a server, comprising at least one processor, and a memory communicatively connected to the processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the processor so that the at least one processor executes the artificial intelligence-based clinical decision-making support method as described in the first aspect.

[0051] In a fourth aspect, the present solution discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the clinical decision-making support method based on artificial intelligence as described in the first aspect.

[0052] Compared with the prior art, the artificial intelligence-based clinical decision-making assistance method and related products described in the present invention have the following beneficial effects:

[0053] (1) The artificial intelligence-based clinical decision-making support method and related products described in the present invention automatically analyze the patient's condition, generate a list of suspected and confirmed diseases, and recommend examinations, tests and treatment plans in a targeted manner, effectively avoiding the occurrence of missed diagnosis and misdiagnosis;

[0054] (2) The artificial intelligence-based clinical decision-making assistance method and related products described in the present invention can greatly improve the integrity and accuracy of medical records through intelligent medical record management and quality control functions, thereby reducing medical disputes caused by medical record problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0056] Figure 1 Schematic diagram of the artificial intelligence-based clinical decision-making assistance method described in an embodiment of the present invention. DETAILED DESCRIPTION

[0057] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0058] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0059] CDSS system based on AI deep learning is a clinical decision support system that follows evidence-based medicine and is independently developed by learning from authoritative textbooks, pharmacopoeias and high-quality hospital medical records, based on multiple AI technologies such as medical knowledge graphs, natural language processing, cognitive computing, and integrating probabilistic graphical reasoning, rule reasoning and multi-model decision systems based on deep learning. It is used to improve diagnostic accuracy and reduce misdiagnosis and missed diagnosis.

[0060] The system provides the following functions:

[0061] 1. Recommended diagnosis: Doctors provide real-time suspected diagnosis and differential diagnosis suggestions by entering clinical characteristics such as symptoms and signs.

[0062] The recommended diagnosis includes two aspects: recommending suspected diagnosis and recommending differential diagnosis. The reasons for the recommended diagnosis and a detailed introduction to the disease can also be given.

[0063] (1) Recommendation of suspected diagnosis

[0064] It supports recommending suspected diagnoses based on 12 dimensions: causes, symptoms, signs, tests, examinations, medical history, allergy history, medication history, surgical history, family history, marital history, and personal history.

[0065] The CDSS system recommends the most relevant diseases (including probabilities), related symptoms and signs based on the chief complaint, symptoms mentioned in the current medical history (including the causes, duration, location, nature, degree, aggravating and alleviating factors of the symptoms), diseases, and physical examination and test results. The recommended diagnosis is based on 12 dimensions including causes, symptoms, signs, tests, examinations, medical history, allergy history, medication history, surgical history, family history, marital history, and personal history.

[0066] (2) Recommended differential diagnosis

[0067] Supports recommendation of differential diagnosis based on current diagnosis. Once the initial diagnosis is confirmed, the system will recommend differential diagnosis information based on the chief complaint and relevant information mentioned in the current medical history, and guide further differential analysis of the initial diagnosis to exclude other possibilities.

[0068] 2. Auxiliary treatment: Recommend examinations, tests, surgeries, scales, and drugs based on the initial diagnosis of the disease. Through configurable personalized backfill treatment plans, improve clinical writing efficiency and accuracy, which is convenient and fast.

[0069] (1) Recommended examination

[0070] Recommend examination items for the purpose of diagnosis or differentiation based on the initial diagnosis. Once the initial diagnosis is confirmed, the system will recommend further examination items based on the patient's basic information, chief complaint, current medical history and other medical records, as well as the preliminary diagnosis and completed examination items, and give corresponding reasons to assist doctors in making better decisions.

[0071] (2) Recommended inspection

[0072] Recommend examination items for the purpose of diagnosis or differentiation based on the initial diagnosis. Once the initial diagnosis is confirmed, the system will recommend further examination items based on the patient's basic information, chief complaint, current medical history and other medical records, as well as the preliminary diagnosis and completed examination items, and give corresponding reasons to assist doctors in making better decisions.

[0073] (3) Recommended evaluation form

[0074] Supports recommendation of required assessment forms based on symptoms, diagnosis and other conditions. After entering the medical history / initial diagnosis, the system will recommend the assessment form required by the doctor based on the patient's basic information, chief complaint, current medical history, past medical history, allergy history and other medical record information, examination and test results, diagnosis results and other content, so that the doctor can calculate the information that needs to be quantified more accurately and efficiently.

[0075] (4) Recommended drugs

[0076] It supports recommending appropriate drugs based on age, gender, symptoms, diagnosis results, etc. After the initial diagnosis is confirmed, the system will recommend reasonable medications based on the patient's basic information, chief complaint, current medical history, past medical history, allergy history, examination and test results, diagnosis results, etc., and give corresponding reasons to assist doctors in making better decisions.

[0077] (5) Surgery recommendation

[0078] It supports recommending appropriate surgical methods based on age, gender, symptoms, examination, test results, diagnosis results, etc. After the initial diagnosis is confirmed, the system will recommend surgical treatment plans that doctors can perform based on the patient's basic information, chief complaint, current medical history, past medical history, allergy history, examination test results, diagnosis results, etc., and list the surgical indications and contraindications.

[0079] 3. Intelligent quality control: Equipped with intelligent medical quality control services, the system can verify the doctor's orders and medical records in real time to avoid clinical errors at the first opportunity.

[0080] (1) Inspection item conflict reminder

[0081] It supports conflict judgment of test items based on information such as gender, age, symptoms, signs, and diagnosis, and provides reminders for risk situations. After entering the patient's basic information, chief complaint, current medical history and other medical records, the system will determine the risk factors of the new test items (gender, age, pregnancy status, high-risk symptoms) based on the relevant information. If there is a test risk, the early warning mechanism will be triggered and a reminder of the corresponding level will be given.

[0082] (2) Check project conflict reminder

[0083] It supports conflict judgment of examination items based on information such as gender, age, symptoms, signs, and diagnosis, and provides reminders for risk situations. After entering the patient's basic information, chief complaint, current medical history and other medical records, the system will determine the risk factors of the new examination items (gender, age, pregnancy status, high-risk symptoms) based on the relevant information. If there is an examination risk, the early warning mechanism will be triggered and a reminder of the corresponding level will be given.

[0084] 4. Recommend similar medical records: Query similar medical records based on the same clinical manifestations of the current patient and provide real clinical cases.

[0085] The system supports recommending cases similar to the current description from three dimensions: symptoms, diagnosis, and treatment, providing doctors with evidence for diagnosis and treatment.

[0086] 5. Knowledge base: Provides authoritative medical textbooks and the latest clinical guidelines, including knowledge modules such as diseases, drugs, tests, examinations, and surgeries, and is equipped with core retrieval technology to make medical searches more intelligent and friendly.

[0087] The construction of the knowledge base aims to apply the latest concepts of knowledge management, integrate advanced information, digital and network technologies, and build a comprehensive system for medical and clinical knowledge management, information exchange and medical knowledge sharing. The knowledge base of the clinical decision support system is generally a set of clinical guidance opinions developed by the system to help doctors and patients make appropriate treatments and choices for specific clinical problems and decide on appropriate health care services.

[0088] Those of ordinary skill in the art will appreciate that the units and method steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0089] In the several embodiments provided in the present application, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the division of the units described above is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. The above-mentioned units may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present invention.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

[0091] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A clinical decision-making assistance method based on artificial intelligence, characterized in that: include: Build a knowledge graph covering medical entities and relationships; Doctors enter patient information and perform data standardization and cleaning; Recommend suspected diagnosis and differential diagnosis based on multi-dimensional information of patients; Recommend a preset examination plan based on the suspected diagnosis and differential diagnosis results; When a doctor's order is issued, the system checks conflicts between tests and examination items in real time and provides risk warnings; Recommend similar medical records based on the patient's clinical information to provide doctors with diagnosis and treatment references.

2. The artificial intelligence-based clinical decision-making assistance method according to claim 1, characterized in that: The construction covers a knowledge graph of medical entities and relationships, including: By studying authoritative medical textbooks, pharmacopoeias and high-quality hospital medical records, we collect medical data covering diseases, symptoms, signs, tests, drugs, and surgeries, and extract medical entities and relationships from unstructured texts; The extracted medical entities and relations are organized into a structured knowledge graph. Graph embedding technology is used in the construction process to represent the nodes and edges in the knowledge graph as vectors, and the representation is learned by optimizing the objective function: Among them, eh and et are the vector representations of the head entity and the tail entity respectively, Wr is the transformation matrix of the relationship, σ(·) is the sigmoid function, and T is the set of triples in the knowledge graph.

3. The artificial intelligence-based clinical decision-making assistance method according to claim 1, characterized in that: The doctor enters the patient information and performs data standardization and cleaning, including: The doctor inputs the patient's basic and clinical information, parses the input text data, and standardizes the input information using a predefined medical terminology set to ensure consistency of the information; The input data is cleaned to remove redundant and inconsistent information, and an interpolation-based algorithm is used to handle missing values ​​to improve data integrity: Among them, x^i represents the estimation of the missing value, N(i) is the neighbor set of node i, wij is the influence weight of neighbor j on i, and xj is the value of the neighbor node.

4. The artificial intelligence-based clinical decision-making assistance method according to claim 1, characterized in that: Based on the patient's multi-dimensional information, the suspected diagnosis and differential diagnosis are recommended, including: The most likely suspected diagnosis is recommended based on the multi-dimensional information input by the patient. When calculating the probability of each disease, the weighted contribution of each dimension is taken into account: Among them, P(Di|X) represents the posterior probability of disease Di, X represents the clinical information of the patient, αj is the weight of dimension j, fj(X) is the characteristic function of dimension j, and Z is the normalization factor; After the initial diagnosis, the most likely differential diagnosis is recommended by analyzing the similarity of the current diagnosis with other possible diseases and using a distance metric to rank the diseases: Among them, d(Di,Dj) represents the distance between diseases Di and Dj, xik and xjk are the values ​​of diseases Di and Dj on feature k, respectively, and σk is the standard deviation of feature k.

5. The artificial intelligence-based clinical decision-making assistance method according to claim 1, characterized in that: Based on the suspected diagnosis and differential diagnosis results, a preset examination plan is recommended, including: Recommendations for examinations and tests: Based on the preliminary diagnosis results, combined with the patient's basic information and current medical history, we recommend the examinations and tests required for diagnosis or differentiation, and balance the accuracy and cost of diagnosis based on a multi-objective optimization algorithm: Where y is the recommended inspection and testing combination, Cj(yj) is the cost function of inspecting item j, Rj(yj) is the risk function of inspecting item j, and λj is the parameter for weighing cost and risk; Evaluation form and drug recommendation: Based on the diagnosis results and patient information, we recommend evaluation forms and drugs. In drug recommendations, we combine the patient's characteristics and historical medication data to recommend the most suitable drugs: Among them, r^ui represents the predicted score of user u for drug i, μ is the global bias, bu and bi are the biases of the user and drug respectively, qi and pu are the feature vectors of the drug and user; Recommendation of surgical method: Based on the diagnosis and the patient's personalized information, the indications and contraindications of surgery are considered and the surgical plan is optimized through a multi-level decision tree model: In it, T*(X) represents the best surgical decision tree, T is the set of all possible decision trees, δ(T(Xi),yi) is the prediction accuracy of decision tree T for input Xi and true label yi, and wi is the weight of the sample.

6. The artificial intelligence-based clinical decision-making assistance method according to claim 1, characterized in that: When the doctor's order is issued, the system checks the conflicts between the tests and the examination items in real time and provides risk warnings, including: When doctors issue medical orders, they can check conflicts between tests and examination items in real time. By establishing a conflict matrix, they can determine whether new items conflict with existing items based on the patient's characteristics and provide risk warnings: Among them, Cij represents the conflict score between items i and j, βk is the weight of feature k, and gk(i,j) is the contribution function of feature k to the conflict between items i and j.

7. The artificial intelligence-based clinical decision-making assistance method according to claim 1, characterized in that: The recommendation of similar medical records based on the patient's clinical information to provide doctors with diagnosis and treatment references includes: Based on the patient's clinical information, through the analysis of symptoms, diagnosis, and treatment in multiple dimensions, a similarity measurement algorithm is used to recommend medical records similar to the current patient: Among them, S(Xp,Xq) represents the similarity between patient p and patient q, γk is the weight of feature k, and sk(xpk,xqk) is the similarity function of feature k.

8. An electronic device, comprising a processor and a memory connected to the processor for storing instructions executable by the processor, characterized in that: The processor is used to execute the artificial intelligence-based clinical decision-making assistance method described in any one of claims 1-7.

9. A server, characterized in that: It includes at least one processor and a memory communicatively connected to the processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the processor so that the at least one processor executes the artificial intelligence-based clinical decision-making support method as described in any one of claims 1-7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the artificial intelligence-based clinical decision-making assistance method described in any one of claims 1 to 7 is implemented.

Images