Accurate personalized intelligent aid decision-making method and system for anti-rheumatism medicine

Abstract

The invention relates to an anti-rheumatism medicine precise personalized intelligent aid decision making method and system, and relates to the field of intelligent medical treatment, and the method comprises the steps: obtaining patient diagnosis related data; performing language analysis on the patient diagnosis related data to obtain current keyword information; performing range screening on drugs stored in a preset recommendation database based on the current keyword information in sequence to obtain recommended drugs, wherein the recommendation database stores keyword information matched with the current keyword information, a mapping relationship among the drugs and drug curative effects; searching a corresponding recommended drug curative effect from a preset recommendation database based on the current keyword information and the recommended drug; and forming a recommendation scheme based on the recommended medicine and the curative effect of the recommended medicine, and displaying the recommendation scheme on a preset doctor end. The method has the effects of performing recommendation by automatically forming the recommended medicine combination, not needing a doctor to depend on personal experience for judgment, reducing selection deviation of the treatment scheme and improving the accuracy of medicine recommendation.

Description

Technical Field

[0001] The present application relates to the field of smart healthcare, and in particular to a method and system for precise and personalized intelligent decision-making assistance for anti-rheumatic drugs. Background Art

[0002] When a person feels unwell or suspects they have a disease, the process of seeking diagnosis and treatment at a medical institution is a core part of a medical consultation. This process is even more complex in the diagnosis and treatment of rheumatic diseases (such as rheumatoid arthritis and systemic lupus erythematosus). Patients often undergo multiple rounds of immune marker testing, joint imaging assessments, and medical history review to clarify the disease phenotype (such as seropositive RA or renal involvement of SLE).

[0003] In related technologies, doctors usually follow the process of "inquiring medical history → physical examination → auxiliary diagnosis → plan formulation". Taking rheumatic diseases as an example, doctors need to pay attention to specific indicators such as the duration of morning stiffness, the location of joint swelling, and the titer of autoantibodies (such as anti-CCP antibodies, dsDNA antibodies), and combine them with

[0004] Diagnosis is based on guidelines such as the ACR / EULAR Rheumatoid Arthritis Classification Criteria. Finally, doctors develop treatment plans and prescribe medications based on their personal experience.

[0005] The existing technology has the following problems: the process of prescribing drugs based on personal experience is prone to selection bias due to different abilities and experience levels of each person, which affects the patient's recovery process. There is still room for improvement. Summary of the Invention

[0006] In order to improve the problem that the process of prescribing drugs based on personal experience is prone to selection bias due to different abilities and experience levels of each person, thus affecting the patient's recovery process, this application provides a precise personalized intelligent auxiliary decision-making method and system for anti-rheumatic drugs.

[0007] In the first aspect, the present application provides a precise and personalized intelligent decision-making method for anti-rheumatic drugs, which adopts the following technical solutions:

[0008] A precise and personalized intelligent decision-making method for anti-rheumatic drugs, comprising:

[0009] Obtain patient diagnosis-related data;

[0010] Perform language analysis on patient diagnosis-related data to obtain current keyword information;

[0011] Based on the current keyword information, the medicines stored in the preset recommendation database are range-screened to obtain recommended medicines, wherein the recommendation database stores a mapping relationship between keyword information, medicines, and drug efficacy that matches the current keyword information;

[0012] Based on the current keyword information and recommended drugs, the corresponding recommended drug efficacy is found from the preset recommendation database;

[0013] A recommendation plan is formed based on the recommended drugs and the efficacy of the recommended drugs and displayed on the preset doctor's end.

[0014] By adopting the above technical solution, patient data is received, and then keywords are extracted. The range of therapeutic drugs is narrowed down in sequence according to the order of keywords, and finally a recommended drug combination (drug + efficacy) is formed for recommendation. Doctors do not need to rely on personal experience and judgment, which reduces the selection bias of treatment plans and improves the accuracy of drug recommendations.

[0015] Optionally, a method for updating the recommendation database is also included, the method comprising:

[0016] Obtain the latest scientific research data, authoritative guidelines on recommended medications for related diseases, and the latest trial data, including non-human trial data and virtual human trial data;

[0017] Analyze the latest scientific research data and the latest experimental data to extract the latest keyword information, corresponding new drugs and new drug efficacy;

[0018] Form the latest mapping relationship based on the latest keyword information, new drugs and new drug efficacy;

[0019] When the latest keyword information and new drugs do not match the keyword information or new drugs in the mapping relationship in the recommendation database, the latest mapping relationship is stored in the recommendation database;

[0020] When the latest keyword information matches the keyword information in the mapping relationship in the recommendation database, the latest mapping relationship and the mapping relationship are selected and updated based on the efficacy of the new drug and the efficacy of the drug;

[0021] When a new drug matches a drug in a mapping relationship in a recommendation database, the latest mapping relationship is stored in the recommendation database.

[0022] By adopting the above technical solutions, the content of the database will be updated in real time based on the latest scientific research findings and actual test data, avoiding lags in drug recommendation results and improving the accuracy and timeliness of therapeutic drug recommendations.

[0023] Optionally, the method of forming a recommendation plan and displaying it on a preset doctor's terminal includes:

[0024] Get the adoption probability of the recommended solution;

[0025] When the adoption probability is less than the preset critical appropriate probability, the recommended solution is disassembled to obtain a separate recommended solution;

[0026] Based on the recommended drugs corresponding to the individual recommendation plan, the corresponding drug expert category is searched from the preset doctor database;

[0027] Get recommended expert categories;

[0028] When the recommended expert category and the drug expert category are consistent, a recommendation plan is formed and displayed on the doctor's side;

[0029] No recommendation plan will be formed when the recommended expert category and the drug expert category are inconsistent;

[0030] When the adoption probability is greater than the critical appropriate probability, a recommendation plan is formed and displayed on the doctor's side.

[0031] By adopting the above technical solution, if the probability of adoption of the recommended drug is low, the recommendation will be made based on the doctor's level, ability and professional category, preventing the possibility of the doctor randomly selecting recommended drugs due to insufficient professionalism, thereby improving the safety of drug recommendations.

[0032] Optionally, the method for updating the recommendation database further includes:

[0033] Obtain the user's IP address and patient category when the recommended solution is adopted;

[0034] Reversely search for the corresponding mapping relationship based on the adopted recommendation solution, and define the mapping relationship as the mapping relationship to be updated;

[0035] Receive feedback information from the user's IP input;

[0036] Disassemble feedback information to get actual drug efficacy;

[0037] Based on the patient category and the mapping relationship to be updated, the corresponding reference degree is found from the preset reference database;

[0038] The mapping relationship to be updated is updated based on the actual drug efficacy, user category, and reference degree to obtain an updated mapping relationship, and is stored in the recommendation database as a mapping relationship.

[0039] By adopting the above technical solution, after being recommended and adopted, the missing human experimental data is enriched by recording the patient's daily life, thereby improving the completeness and timeliness of the latest experimental data.

[0040] Optionally, the method of sequentially screening the drugs stored in the preset recommendation database based on the current keyword information to obtain the recommended drugs includes:

[0041] Analyze the current keyword information to obtain the current symptom words and the past symptom words;

[0042] Screening drugs based on the current symptom word to obtain drugs that can treat the current symptom word, and defining the drugs as targeted drugs;

[0043] Screening targeted drugs based on previous disease words to obtain targeted drugs that are beneficial to the previous disease words, and defining the targeted drugs as beneficial targeted drugs;

[0044] When the number of beneficial targeted drugs required by the current symptom word is equal to 1, the beneficial targeted drug is used as the recommended drug;

[0045] When the number of beneficial targeted drugs required by the current symptom word is greater than 1, the corresponding interaction is searched from a preset interaction database based on each beneficial targeted drug corresponding to each current symptom word;

[0046] Based on the interaction, each beneficial targeted drug corresponding to each current symptom word is self-combined to obtain an interfering drug group, a non-interfering drug group and a synergistic drug group;

[0047] When a synergistic drug group exists, the synergistic drug group will be used as the recommended drug;

[0048] When the synergistic drug group does not exist, the non-interfering drug group will be recommended as the drug;

[0049] When only the interfering drug group exists, a preset alarm signal is output.

[0050] By adopting the above technical solution, the method of narrowing the scope is as follows: first, a range of drugs that can treat the disease is formed, and then drugs with side effects are excluded (past medical history, description of symptoms, and drugs already taken). Then, drugs with beneficial effects are given priority recommendation (synergistic drugs already taken, drugs that have good effects on other past medical histories, drugs that treat multiple diseases at the same time, and drugs that can synergize with other drugs for treating the current disease), thereby improving the intelligence and rationality of drug recommendations.

[0051] Optional approaches to recommending a non-interfering drug group include:

[0052] Obtain patient information and purchased medication information;

[0053] Analyze the purchased drug information to obtain the purchased effective drugs and the purchased quantities;

[0054] Determine the efficacy of purchased drugs and the efficacy of non-interfering drugs based on drugs in the purchased effective drugs and non-interfering drug groups;

[0055] Searching for corresponding substitution levels from a preset substitution database based on the purchased effective drugs, the non-interfering drug group, the efficacy of the purchased drugs, and the efficacy of the non-interfering drugs;

[0056] When the degree of substitution is a preset allowable degree of substitution, the non-interfering drug group is defined as a target drug group, and the replaced drug in the non-interfering drug group is defined as a replacement drug;

[0057] Determine required quantities based on replacement medication and patient information;

[0058] When the required quantity is less than or equal to the purchased quantity, the replacement drug is deleted from the target drug group to obtain the first target drug group, which is used as the recommended drug;

[0059] When the demand quantity is greater than the purchased quantity, the difference quantity is obtained based on the demand quantity and the purchased quantity;

[0060] updating the target drug group based on the difference quantity to obtain a second target drug group as the recommended drug;

[0061] When the replacement level reaches the preset negative replacement level, the non-interfering drug group will be used as the recommended drug.

[0062] By adopting the above technical solution, the scope is narrowed down and the drugs that the user has already purchased and is expected to retain are matched, which reduces the amount of drugs purchased and improves the humanization of drug recommendations.

[0063] Optionally, a method of deleting the replacement drug from the target drug group to obtain a first target drug group and using the first target drug group as the recommended drug includes:

[0064] When the degree of substitution is weaker than the preset equal substitution degree, the substituted drug is not deleted from the target drug group, and the target drug group is still used as the recommended drug;

[0065] When the degree of substitution is stronger than the degree of equal substitution, the substitution drug is deleted from the target drug group to obtain the first target drug group, which is used as the recommended drug.

[0066] By adopting the above technical solution, if the purchased drug is not as effective or has fewer side effects as the subsequent drug, the subsequent drug will still be recommended, thereby improving the excellence of drug recommendation and scenario analysis capabilities.

[0067] Optionally, the method of recommending the non-interfering drug group as the drug when the replacement degree is a preset negative replacement degree includes:

[0068] Based on the purchased effective drugs and non-interfering drug groups, the corresponding purchased interactions are found from the interaction database;

[0069] Classifying the non-interfering drug group based on the purchased interactions to obtain a purchased interfering drug group, a purchased non-interfering drug group, and a purchased synergistic drug group;

[0070] When a purchased synergistic drug group exists, the purchased synergistic drug group will be used as the recommended drug;

[0071] When a purchased synergistic drug group does not exist, the purchased non-interfering drug group will be used as the recommended drug;

[0072] Determine the recommended interfering drug for the purchased interfering drug group based on the purchased interactions when neither the purchased non-interfering drug group nor the purchased synergistic drug group exists;

[0073] A non-recommendation signal is formed based on recommended interfering drugs, purchased effective drugs, and purchased interactions;

[0074] The purchased interfering drug group is outputted simultaneously as a recommended drug and a non-recommended signal.

[0075] By adopting the above technical solution, if the purchased drugs and the currently recommended drugs interact with each other, recommendations will be made based on the interaction. For example, interference effects will not be recommended, while synergistic effects will be recommended as a priority, thereby improving the rationality and intelligence of drug recommendations.

[0076] Optionally, the method of outputting the purchased interfering drug group as a recommended drug and a non-recommended signal simultaneously includes:

[0077] Based on the recommended interfering drugs and the purchased effective drugs, a corresponding elimination drug is searched from a preset elimination database, and the elimination drug itself has no drug efficacy;

[0078] Add the eliminated drugs to the purchased non-interfering drug group to form the expected drug group, and analyze whether it is a non-interfering drug group or a synergistic drug group;

[0079] When the expected drug group is a non-interfering drug group or a synergistic drug group, the expected drug group is output as a recommended drug;

[0080] When the drug group is not expected to be a non-interfering drug group or a synergistic drug group, the purchased non-interfering drug group is outputted as a recommended drug and a non-recommended signal at the same time.

[0081] By adopting the above technical solution, by adding an additional drug, the side effects caused by the interaction between the purchased drugs and the currently recommended drugs are suppressed without causing side effects, thereby improving the flexibility of drug recommendations.

[0082] In a second aspect, the present application provides a drug recommendation system, which adopts the following technical solutions:

[0083] A drug recommendation system, comprising:

[0084] The acquisition module is used to obtain patient diagnosis-related data, the latest scientific research data, the latest test data, adoption probability, recommended expert category, user IP address, patient category, patient information, and purchased drug information;

[0085] A memory for storing a program for a control method of any of the above-mentioned methods for precise personalized intelligent decision-making assistance for anti-rheumatic drugs;

[0086] The program in the processor memory can be loaded and executed by the processor to realize the control method of any of the above-mentioned precise personalized intelligent auxiliary decision-making methods for anti-rheumatic drugs.

[0087] By adopting the above technical solution, patient data is received, and then keywords are extracted. The range of therapeutic drugs is narrowed down in sequence according to the order of keywords, and finally a recommended drug combination (drug + efficacy) is formed for recommendation. Doctors do not need to rely on personal experience and judgment, which reduces the selection bias of treatment plans and improves the accuracy of drug recommendations.

[0088] In summary, this application includes at least the following beneficial technical effects:

[0089] 1. By automatically generating recommended drug combinations, doctors no longer need to rely on personal experience and judgment, reducing treatment selection bias and improving the accuracy of drug recommendations;

[0090] 2. Real-time updates based on the latest scientific research findings and actual trial data avoid lags in drug recommendation results and improve the accuracy and timeliness of treatment drug recommendations;

[0091] 3. After narrowing the scope, the drugs that the user has already purchased and is expected to retain are matched, reducing the amount of drugs purchased and improving the humanization of drug recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] Figure 1 This is a flow chart of a precise personalized intelligent decision-making method for anti-rheumatic drugs in an embodiment of the present application.

[0093] Figure 2 This is a flowchart of a method for updating a recommendation database in an embodiment of the present application.

[0094] Figure 3 This is a flowchart of a method for forming a recommendation plan and displaying it on a preset doctor's terminal in an embodiment of the present application.

[0095] Figure 4It is a flowchart of a further method for updating the recommendation database in an embodiment of the present application.

[0096] Figure 5 This is a flowchart of a method for sequentially screening the drugs stored in a preset recommendation database based on current keyword information to obtain recommended drugs in an embodiment of the present application.

[0097] Figure 6 This is a flow chart of a method for recommending a non-interfering drug group as a drug in an embodiment of the present application.

[0098] Figure 7 This is a flowchart of a method for deleting replacement drugs from a target drug group to obtain a first target drug group and using it as a recommended drug in an embodiment of the present application.

[0099] Figure 8 This is a flowchart of a method for recommending a non-interfering drug group as a drug when the replacement degree is a preset negative replacement degree in an embodiment of the present application.

[0100] Figure 9 This is a flowchart of a method for simultaneously outputting a purchased interfering drug group as a recommended drug and a non-recommendation signal in an embodiment of the present application.

[0101] Figure 10 This is a system module diagram of a precise personalized intelligent decision-making method for anti-rheumatic drugs in an embodiment of the present application. DETAILED DESCRIPTION

[0102] In order to make the purpose, technical solutions and advantages of this application more clear, the following Figures 1-10 It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application.

[0103] The embodiments of the present application disclose a precise and personalized intelligent auxiliary decision-making method for anti-rheumatic drugs.

[0104] Reference Figure 1 , a precise personalized intelligent auxiliary decision-making method for anti-rheumatic drugs includes:

[0105] Step 100: Obtain patient diagnosis-related data.

[0106] Patient diagnosis-related data refers to the data obtained after the doctor diagnoses the patient, including data obtained by the doctor through questioning, observation, instrument auscultation, physical examination and measurement of the patient, such as colds and fevers. It should be noted that it also includes a lot of data related to the disease or the patient's body, such as allergy history, etc. It is obtained by the doctor entering it on the corresponding doctor's end. The input method here can be a doctor filling in the blanks, or selecting the corresponding data in the corresponding area, or directly entering the description of the disease. Here, patient diagnosis-related data include rheumatism-related laboratory test results (such as RF, anti-CCP antibodies, ANA, dsDNA antibodies), joint imaging features (such as synovial thickening, bone erosion), disease activity score (DAS28, SLEDAI) and comorbidity information (such as interstitial lung disease, Sjögren's syndrome).

[0107] Step 101: Perform language analysis on patient diagnosis-related data to obtain current keyword information.

[0108] The current keyword information is information that can express symptoms and words that affect drug recommendations. The current keyword information includes rheumatic disease phenotype keywords (such as 'active rheumatoid arthritis', 'SLE renal damage', 'HLA-B*27 positive') and pathological characteristic keywords (such as 'Th17 cell polarization', 'RANKL high expression').

[0109] Step 102: sequentially screening the drugs stored in the preset recommendation database based on the current keyword information to obtain recommended drugs.

[0110] Recommended drugs are drugs that can treat corresponding diseases and are recommended based on the current keyword information. Here, the drug can be one or multiple drugs can be grouped together.

[0111] The recommendation database stores mappings between keywords, drugs, and drug efficacy that match the current keyword information. This is explained in the subsequent steps and will not be repeated here. When the system receives the current keyword information, it searches the preset database for the corresponding recommended drug and outputs it.

[0112] Step 103: Based on the current keyword information and the recommended drugs, the corresponding recommended drug efficacy is searched from a preset recommendation database.

[0113] The recommended drug efficacy is the recommended drug's effectiveness against the condition corresponding to the current keyword information. For example, if the current keyword information is diarrhea and the recommended drug is montmorillonite powder, the recommended effect will definitely be to stop diarrhea. The database stores a mapping relationship between the current keyword information, recommended drugs, and recommended drug efficacy. This is recorded by personnel in this field based on actual clinical responses and experience. When the system receives the corresponding current keyword information and recommended drug, it automatically searches the database for the corresponding recommended drug efficacy and outputs it.

[0114] Step 104: A recommendation plan is formed based on the recommended drugs and the efficacy of the recommended drugs and displayed on a preset doctor's terminal.

[0115] The recommended plan is an integrated plan of recommended drugs and their efficacy.

[0116] It should be noted that different recommended drugs will have different therapeutic effects on the diseases corresponding to the current keyword information, so the recommended therapeutic effects will also be output, rather than blindly launching only recommended drugs and then relying on doctors to make estimates based on their own experience.

[0117] Reference Figure 2 , further comprising a method for updating a recommendation database, the method comprising:

[0118] Step 200: Obtain the latest scientific research data and the latest experimental data.

[0119] The latest scientific research data is data published in research literature or after the development of new drugs. The latest experimental data is data obtained from the testing of new drugs, or it can be new data obtained from the retesting of existing drugs. The latest experimental data includes non-human trial data and virtual human trial data. Non-human trials can be conducted on animals, such as mice. Virtual human trials can simulate human scenarios, then add various parameters to simulate the reactions under various drug trials and quantify the expected benefit-risk ratio.

[0120] The two can be obtained by workers in this field reading and analyzing them according to a fixed frequency and then inputting them into the system.

[0121] In addition, you can also obtain authoritative guidelines for recommended medications for related diseases. As a supplement, foreign authoritative guidelines include the American College of Rheumatology (ACR) guidelines, the European League Against Rheumatism (EULAR) guidelines, the British Society for Rheumatology (BSR) guidelines, and the Asia Pacific League Against Rheumatism (APLAR) guidelines. Domestic authoritative guidelines include the Chinese Rheumatology Guidelines and the Chinese Clinical Guidelines and Expert Consensus.

[0122] Step 201: Analyze the latest scientific research data and the latest experimental data to extract the latest keyword information, corresponding new drugs and new drug efficacy.

[0123] The latest keyword information is the latest information that can express symptoms and words that affect drug recommendations. New drugs are drugs that can treat the symptoms corresponding to the latest keyword information in the latest scientific research data and the latest test data. The efficacy of new drugs is the efficacy of new drugs against the latest keyword information. The extraction method can be consistent with step 101 and is not analyzed here. The latest scientific research data includes updated guidelines in the field of rheumatism (such as ACR / EULAR treatment guidelines) and clinical trial data of biological agents (such as IL-17 inhibitors, anti-CD20 monoclonal antibodies); the new drugs include anti-rheumatic targeted drugs (such as JAK inhibitors, Syk inhibitors).

[0124] Step 202: forming a latest mapping relationship based on the latest keyword information, the new drug and the efficacy of the new drug.

[0125] The latest mapping relationships are the relationships between the latest keyword information, new drugs, and new drug efficacy. These mapping relationships include association data between rheumatic drugs and HLA genotypes and drug-metabolizing enzyme genes (such as CYP2C19), which are used to predict drug efficacy and toxicity risks.

[0126] Step 203: When the latest keyword information and the new drug do not match the keyword information or the new drug in the mapping relationship in the recommendation database, the latest mapping relationship is stored in the recommendation database.

[0127] If the latest keyword information and new drugs do not match the keyword information or new drugs in the mapping relationship in the recommendation database, it means that the latest mapping relationship has no connection with the recommendation database, so they can be directly stored.

[0128] Step 204: When the latest keyword information matches the keyword information in the mapping relationship in the recommendation database, the latest mapping relationship and the mapping relationship are selected and updated based on the efficacy of the new drug and the efficacy of the drug.

[0129] The updating method is to update according to the excellence of the drug's efficacy and the size of its side effects. If the new drug has better efficacy and smaller side effects, the latest mapping relationship will replace the mapping relationship stored in the database. If the new drug's efficacy is not as good as the drug's efficacy and its side effects are greater, then no update will be performed.

[0130] When the latest keyword information matches the keyword information in the mapping relationship in the recommendation database, it means that the two are the same disease, but the drugs are different, so it is necessary to judge which drug is more effective based on the efficacy of the drugs.

[0131] Step 205: When the new drug matches the drug in the mapping relationship in the recommendation database, the latest mapping relationship is stored in the recommendation database.

[0132] When a new drug matches a drug in a mapping relationship in a recommendation database, it means that the same drug can treat different diseases and achieve different therapeutic effects. In this case, it is still stored in the database as a mapping relationship.

[0133] Reference Figure 3 The method for forming a recommendation plan and displaying it on the preset doctor's terminal includes:

[0134] Step 300: Obtain the adoption probability of the recommended solution.

[0135] The adoption probability is the probability that a recommended solution will be adopted after it is released. This probability is statistically derived, meaning that each time the same solution is recommended, its adoption is recorded. Whether a solution is adopted is determined by directly reading whether the doctor clicks the confirmation button. The probability is then calculated by dividing the number of adoptions by the number of recommendations.

[0136] It should be noted that if the number of recommendations is less than or equal to a certain number, for example, 10 times, the adoption probability defaults to zero due to lack of reference value.

[0137] Step 301 : When the adoption probability is less than a predetermined critical suitability probability, the recommended solution is decomposed to obtain a separate recommended solution.

[0138] A critical suitability probability is a probability greater than this, indicating the probability of being recommended by default. A single recommendation is a recommendation for a single drug. The breakdown method is to directly classify by drug.

[0139] When the adoption probability is less than the critical appropriate probability, it means that it will generally not be adopted at this time. However, there are also cases where the possibility of adoption is abnormal due to the small sample size, so experts are needed to make judgments.

[0140] Step 302: Based on the recommended drugs corresponding to the individual recommendation plan, the corresponding drug expert category is searched from a preset doctor database.

[0141] Drug experts are experts who have extensive knowledge and expertise in recommended drugs. A database stores a mapping between recommended drugs and drug expert categories. These categories are determined by field professionals based on the drug's name or efficacy. When the system receives a recommended drug, it automatically searches the database for the corresponding drug expert category and outputs it.

[0142] Step 303: Obtain recommended expert categories.

[0143] The recommended expert category is the category of the expert whose recommended solution will be displayed. Doctors can enter their own identity information on their own doctor terminal and the system will automatically read and identify it.

[0144] Step 304: When the recommended expert category and the drug expert category are consistent, a recommendation plan is formed and displayed on the doctor's side.

[0145] When the recommended expert category and the drug expert category are consistent, it means that although the probability of adoption is not high, the corresponding doctor is an expert in this field, has strong judgment ability, and is not prone to making mistakes, so a recommendation plan is formed and displayed on the doctor's side.

[0146] Step 305: When the recommendation expert category and the drug expert category are inconsistent, no recommendation plan is formed.

[0147] If the recommended expert category and the drug expert category are inconsistent, it means that the doctor is not an expert in this field and is prone to making wrong judgments and using the recommended plan. However, the recommended plan has not been put into practice and is prone to problems, so it cannot be output as a recommended plan.

[0148] Step 306: When the adoption probability is greater than the critical appropriate probability, a recommendation plan is generated and displayed on the doctor's side.

[0149] When the adoption probability is greater than the critical suitability probability, it means that the adoption probability of the recommended solution is high and the data is reliable.

[0150] Reference Figure 4 , the updating method of the recommendation database further includes:

[0151] Step 400: Obtain the user IP and patient category when the recommendation plan is adopted.

[0152] The user IP address is the IP address of the app connected to the hospital's system. This app is installed on the user's phone and allows them to record their daily feelings and side effects after taking medication. This information is obtained to connect to the corresponding phone and identify the user.

[0153] The patient category is the patient's disease category. The method of obtaining it is to directly read the current keyword information.

[0154] Step 401: reversely search for a corresponding mapping relationship based on the adopted recommendation solution, and define the mapping relationship as a mapping relationship to be updated.

[0155] The search method is to obtain the current keyword information, that is, the symptoms, based on the recommended drugs and the efficacy of the recommended drugs in the recommended plan.

[0156] Step 402: Receive feedback information input by the user IP.

[0157] Feedback information is the information that users input into the corresponding user IP and transmitted. It can include weight, body temperature, or even data obtained after the user goes to the hospital for a physical examination.

[0158] Step 403: Decompose the feedback information to obtain the actual drug efficacy.

[0159] The actual drug efficacy is the drug efficacy obtained from the feedback information. For example, if the condition is hypertension and the feedback information shows low blood pressure, the actual drug efficacy is blood pressure reduction.

[0160] Step 404: Based on the patient category and the mapping relationship to be updated, the corresponding reference degree is searched from the preset reference database.

[0161] The reference degree is the degree to which the data corresponding to the patient category can be referenced for updating the mapping relationship to be updated. Here, since some patient categories correspond to diseases that may only have a partial efficacy of the drug, but not the most important efficacy, the corresponding reference degree is low. The database stores the mapping relationship between patient categories, mapping relationships to be updated, and reference degrees. The reference degree is obtained by staff in this field based on their own experience to evaluate the reference of patient categories and mapping relationships to be updated. When the system receives the corresponding patient category and mapping relationship to be updated, it automatically searches the database for the corresponding reference degree and outputs it.

[0162] Step 405: Based on the actual drug efficacy, user category and reference degree, the to-be-updated mapping relationship is updated to obtain an updated mapping relationship, and the updated mapping relationship is stored in the recommendation database as a mapping relationship.

[0163] The updating method is to provide a reference weight for the mapping relationship corresponding to the actual drug efficacy and the user category according to the percentage of the reference degree, and then put it into the mapping relationship to be updated.

[0164] Reference Figure 5 The method of sequentially screening the drugs stored in the preset recommendation database based on the current keyword information to obtain the recommended drugs includes:

[0165] Step 500: Analyze based on current keyword information to obtain current symptom words and past symptom words.

[0166] Current symptom words are words corresponding to the current symptom. Past symptom words are words corresponding to the symptom the same patient has previously suffered. Examples include "in the past" and "had previously suffered." The analysis method is to read the words related to the symptom and then read the preceding time state words.

[0167] Step 501: Screening drugs based on the current symptom word to obtain a drug that can treat the current symptom word, and defining the drug as a targeted drug.

[0168] The drugs are the drugs that can treat the current symptom word. The screening method is to determine the drugs whose current keyword information is the current symptom word according to the mapping relationship.

[0169] Step 502: Screening targeted drugs based on the previous symptom words to obtain targeted drugs that are beneficial to the previous symptom words, and defining the targeted drugs as beneficial targeted drugs.

[0170] Beneficial targeted drugs are drugs that are beneficial to the previous symptom words among the targeted drugs. If the drug does not exist here, it will be directly output as a targeted drug.

[0171] When the previous disease words include rheumatic-related comorbidities, the beneficial targeted drugs must simultaneously meet the treatment needs of the current rheumatic disease and comorbidities (e.g., for RA combined with interstitial lung disease, DMARDs with lung protective effects should be screened first).

[0172] Step 503: When the number of beneficial targeted drugs required by the current symptom word is equal to 1, the beneficial targeted drug is used as a recommended drug.

[0173] When the number of beneficial targeted drugs required by the current symptom word is equal to 1, there will be no interaction between drugs, so the beneficial targeted drugs can be directly used as recommended drugs.

[0174] Step 504: When the number of beneficial targeted drugs required by the current symptom word is greater than 1, the corresponding interactions are searched from a preset interaction database based on each beneficial targeted drug corresponding to each current symptom word.

[0175] An interaction is an effect, beyond the therapeutic effects of the drugs themselves, that occurs when two drugs are used together. This effect can be beneficial or detrimental. A database stores a mapping between two drugs and their interactions. This is categorized by medical professionals based on common medical knowledge and extensive experimental data. When the system receives information about two beneficial targeted drugs, it automatically searches the database for the corresponding interaction and outputs it.

[0176] The interaction database stores data on specific interactions between rheumatology drugs (such as gastrointestinal toxicity of MTX combined with nonsteroidal anti-inflammatory drugs, and contraindications of biological agents and live vaccines).

[0177] Step 505: Based on the interaction, each beneficial drug corresponding to each current symptom word is self-combined to obtain an interfering drug group, a non-interfering drug group and a synergistic drug group.

[0178] The interfering drug group consisted of at least two beneficial targeted drugs whose therapeutic effects interfered with each other, with the remaining drug combinations being either mutually neutral or synergistic. The non-interfering drug group consisted of drug combinations in which none of the drugs in the group interacted with each other. The synergistic drug group consisted of at least two beneficial targeted drugs whose therapeutic effects synergized with each other, with the remaining drug combinations being mutually neutral. The analysis method was to determine drug interactions within the drug group.

[0179] Step 506: If a synergistic drug group exists, use the synergistic drug group as a recommended drug.

[0180] When a synergistic drug group exists, it indicates that it can promote synergistic effects, and it can be output as a recommended drug.

[0181] Step 507: When the synergistic drug group does not exist, the non-interfering drug group is used as the recommended drug.

[0182] When the synergistic drug group does not exist, it means that there are only the interfering drug group and the non-interfering drug group. In this case, only the non-interfering drug group can be output as the recommended drug.

[0183] Step 508: Output a preset alarm signal when only the interfering drug group exists.

[0184] The alarm signal indicates that the recommendation cannot be made and requires human intervention. If there is only one interfering drug group, then once used, they will interfere with each other and cannot be recommended, so the alarm signal is output.

[0185] Between step 505 and step 506, given that rheumatic diseases have complex autoimmune mechanisms, large individual phenotypic differences, and require long-term dynamic management, combined with the multi-dimensional pathological characteristics and treatment needs of rheumatic diseases, the following steps may also be performed:

[0186] Step 5051: Use the quantum-inspired graph attention network (Q-GAT) to analyze the patient's rheumatic disease phenotype, realize multimodal data parallel processing through the quantum superposition principle and quantum gate operation, and generate a quantum state phenotype map containing characteristics such as inflammation type and bone damage degree.

[0187] Quantum-inspired Graph Attention Network (Q-GAT): An algorithm that integrates quantum computing principles with graph neural networks, processes multi-node features in parallel through quantum superposition states, and uses the attention mechanism to dynamically assign pathological feature weights.

[0188] Quantum state phenotype map: The disease phenotype is characterized by the probability distribution of quantum bits (qubits), including the complex expression of multidimensional features such as inflammation type (such as Th1 / Th17 polarization) and bone destruction degree (RANKL expression).

[0189] The processing process is summarized as follows: 1. Quantum feature encoding: Multimodal data is mapped into quantum state vectors, and feature parallelization is achieved through quantum amplitude encoding (the square of the amplitude corresponds to the feature probability). 2. Quantum gate operation: Hadamard gates are used to generate superposition states, and CNOT gates are used to achieve feature entanglement. Disease-specific quantum circuits are designed, and feature weights are optimized using a variational quantum algorithm (VQE). 3. Measurement and decoding: Quantum measurements are performed on the entangled state, and the feature probability distribution is obtained through wave function collapse. These are converted to classical probability values ​​through maximum likelihood estimation, generating a phenotypic map containing 128-dimensional pathological features.

[0190] In step 5052, a drug-phenotype association scoring system is constructed using topological data analysis (TDA). Medical literature and real-world medication data are mapped into geometric structures in a topological space. By calculating the persistence graph distance between drugs and phenotypic characteristics in the topological space, a drug-phenotype matching scoring matrix with topological invariance is generated.

[0191] Among them, Topological Data Analysis (TDA) is a mathematical method that analyzes high-dimensional data through geometric features (such as holes and connectivity) to capture the inherent structural stability of the data. The Persistence Diagram is a core tool of TDA, representing the "birth" and "death" of data features using points on a two-dimensional plane, quantifying the significance and stability of structural features. The Drug-Phenotype Matching Score Matrix is ​​based on topological invariants and quantifies the degree of compatibility between drugs and specific disease phenotypes. Matrix elements represent the potential of drugs to intervene in specific pathological features.

[0192] The processing process is summarized as follows:

[0193] 1. Data topology mapping: Map drug mechanisms of action (e.g., TNF-α inhibition) and phenotypic characteristics (e.g., Th17 cell ratio) into a high-dimensional point cloud. Construct a Vietoris-Rips complex and generate nested geometric structures using different scale parameters ε.

[0194] 2. Persistence Homology Computation: Compute homology groups of different dimensions (0-dimensional connected components, 1-dimensional holes, and 2-dimensional voids) to generate persistence graphs. Extract key topological features (such as hole lifetime) as invariants of drug-phenotype associations.

[0195] 3. Distance Metrics and Rating Matrix Generation: Calculate the Wasserstein distance between drugs and phenotypes in the persistence graph space. A smaller distance indicates a higher degree of match. Construct an m×n scoring matrix, where m represents the drug type and n represents the phenotypic characteristic. Element values ​​are normalized to [0, 1] using a sigmoid function.

[0196] Step 5053, using Logistic Chaotic Mapping to generate random sequences, chaotic search for beneficial targeted drugs, according to the drug-phenotype matching score matrix and the complementarity of drug action mechanism, in the chaotic x n+1 =rx n (1-x n ) state to dynamically adjust the screening threshold and generate a preliminary drug combination candidate set containing potential advantageous combinations.

[0197] Logistic Chaotic Map: A Nonlinear Iterative System (x n+1 =rx n (1-x n )), by adjusting the parameter r (such as r = 3.9 to enter the chaotic state), a non-periodic, ergodic random sequence is generated for the global search of the solution space.

[0198] Chaotic search: Utilizes the randomness and ergodicity of chaotic sequences to dynamically explore potential advantageous solutions in the drug combination space, avoiding the local optimal trap of traditional algorithms.

[0199] The processing process is summarized as follows:

[0200] 1. Chaotic sequence initialization:

[0201] Set the initial value of the chaotic map x0∈(0,1) to generate a chaotic sequence {xi} of length N, which is converted into a dynamic screening threshold sequence {Ti=Tmin+(Tmax-Tmin)xi} (such as the threshold range [4,8] points) through linear transformation.

[0202] 2. Dynamic Screening and Mechanism Complementarity Assessment: For each targeted drug, a comprehensive score is calculated: Matching Score (60% weight) + Mechanism Complementarity Score (40%, e.g., DMARDs and biologics are assigned a score of 3). Drugs with a comprehensive score greater than the current threshold Ti are included in the preliminary candidate set. Combinations with overlapping mechanisms of action (e.g., two JAK inhibitors) or with absolute contraindications are excluded.

[0203] 3. Candidate set generation: Through the ergodic nature of chaotic sequences, we can efficiently search the drug combination space of 1,000 levels and generate a potentially advantageous candidate set containing 10-20 combinations, reducing the computational effort by 95% compared to the exhaustive method.

[0204] In step 5054, the artificial immune system algorithm (AIS) is used to treat drug combinations as antibodies and rheumatic disease characteristics as antigens. Through clonal selection, mutation, and immune memory mechanisms, the optimal drug combination solution set is iteratively generated under the multi-objective optimization conditions of reducing disease activity, reducing adverse reactions, and controlling costs, as well as the special constraints of rheumatic treatment.

[0205] Among them, the Artificial Immune System (AIS) algorithm simulates a computational model of the biological immune system's ability to recognize and eliminate antigens, achieving optimized search through clonal selection, mutation, and immune memory mechanisms. Antibody-Antigen Matching considers drug combinations as "antibodies" and rheumatic disease characteristics as "antigens," quantifying the effect of drug combinations on disease characteristics through affinity functions. Multi-Objective Optimization simultaneously optimizes three objective functions: reducing disease activity (such as the DAS28 score), minimizing adverse reactions (such as the risk of liver damage), and controlling treatment costs.

[0206] The processing process is summarized as follows:

[0207] 1. Antibody initialization: Encode the initial set of drug combination candidates into an antibody population, where each antibody represents a drug combination (e.g., "MTX + adalimumab"). The antibody gene loci contain parameters such as drug type, dosage, and dosing frequency.

[0208] 2. Affinity Calculation: Design a multi-dimensional affinity function: Affinity = w1·Efficacy + w2·Safety + w3·Cost. The weights w1, w2, and w3 are dynamically adjusted by the doctor based on the patient's disease stage (e.g., w1 = 0.6 during the active stage).

[0209] 3. Immune operation iteration:

[0210] Clone selection: replication of high-affinity antibodies (clone number is proportional to affinity).

[0211] Mutation: Introduce parameter perturbations (such as dosage ±10%) to the clones to explore better solutions.

[0212] Immune memory: retain historically optimal antibodies to avoid falling into local optimality.

[0213] 4. Constraint handling: Penalty function mechanism: Antibody affinity is reduced to 0 for violations of medication contraindications (such as the combined use of NSAIDs and anticoagulants). Metabolic pathway constraints: Potential liver injury risk combinations are filtered through the CYP450 enzyme inhibition model.

[0214] Step 5055: Process the patient's disease history data through a temporal convolutional network (TCN) to predict the disease development trend, and select a drug combination solution that matches the predicted disease status from the optimal drug combination solution set.

[0215] Among them, the Temporal Convolutional Network (TCN) is a deep learning model designed specifically for time series data. It captures long-range temporal dependencies through causal and dilated convolutions and is suitable for disease trend prediction. Disease Development Trend: A prediction of future disease states based on historical data, such as the probability distribution of rheumatic activity (DAS28 > 3.2) or remission (DAS28 ≤ 2.6). The Optimal Drug Combination Solution Set: The Pareto-optimal solution generated in step 5054, consisting of 3-5 candidate solutions that balance efficacy, safety, and cost.

[0216] The processing process is summarized as follows:

[0217] 1. Time series preprocessing: Standardize historical data (Z-score) and fill missing values ​​(e.g., using bidirectional LSTM interpolation). Construct input sequence: Use the previous 12 months of data as input and predict the trend for the next 3 months.

[0218] 2. TCN Model Construction: Network Architecture: Three dilated convolutional layers (dilation factors [1, 2, 4]) to capture temporal dependencies of 1-8 months. Loss Function: A combination of mean squared error (MSE) and binary cross entropy to simultaneously predict both the continuous indicator (CRP) and discrete states (active phase / remission phase).

[0219] Dynamic matching strategy:

[0220] Trend classification: Output the probability of future disease status (such as the probability of active stage 65%).

[0221] Regimen screening: During the active phase, prioritize combinations with rapid onset of action (half-life <72 hours) and high anti-inflammatory potency (e.g., TNF-α inhibitors + hormones). During the remission phase, prioritize maintenance regimens with high safety (e.g., MTX + hydroxychloroquine, excluding long-term toxic drugs). Adaptive adjustments: Trigger regimen lock-in or secondary optimization based on predicted confidence (e.g., >80%).

[0222] Reference Figure 6 , methods for recommending non-interfering drug groups include:

[0223] Step 600: Obtain patient information and purchased drug information.

[0224] Patient information refers to the patient's own information, including weight, age, etc. Purchased medication information refers to information about medications that the patient has purchased, including the quantity of medications when they were prepared, the time of purchase, the effective period of the purchased medications, and the dosage used in previous purchases.

[0225] Step 601: Analyze the purchased drug information to obtain the purchased effective drugs and the purchased quantity.

[0226] Valid purchased medications are those currently valid medications corresponding to the purchased medication information. The purchased quantity is the number of valid medications purchased. The analysis method here is to determine the current validity period based on the expiration date, and then calculate the amount of medication used based on the user's purchase. For example, if a medication is taken once a day for 10 days and the original purchase quantity is 30, then if the medication is still valid, the output quantity of the valid purchased medication is 20.

[0227] Step 602: Determine the efficacy of the purchased drugs and the efficacy of the non-interfering drugs based on the purchased effective drugs and the drugs in the non-interfering drug group.

[0228] The efficacy of purchased drugs is the efficacy of the purchased drugs. The efficacy of non-interfering drugs is the efficacy of drugs in the non-interfering drug group. This is determined by searching for the corresponding mapping relationship and then obtaining the corresponding efficacy.

[0229] Step 603: Based on the purchased effective drugs, the non-interfering drug group, the purchased drug efficacy and the non-interfering drug efficacy, the corresponding replacement degree is searched from a preset replacement database.

[0230] The degree of substitution is the degree to which a purchased effective drug can replace a drug in the non-interfering drug group. A database stores a mapping between the efficacy of purchased drugs, the efficacy of non-interfering drugs, and the degree of substitution. This is determined by personnel in this field based on actual efficacy and their own knowledge and experience. When the system receives the corresponding purchased effective drug, non-interfering drug group, the efficacy of the purchased drug, and the efficacy of the non-interfering drug, it automatically searches the database for the corresponding degree of substitution and outputs it.

[0231] When the patient information includes characteristics of women of childbearing age, the replacement database needs to exclude teratogenic antirheumatic drugs (such as leflunomide) and give priority to pregnancy category B drugs such as hydroxychloroquine.

[0232] Step 604: When the replacement degree is a preset allowable replacement degree, the non-interfering drug group is defined as a target drug group, and the replaced drugs in the non-interfering drug group are defined as replacement drugs.

[0233] The permitted substitution level indicates the extent to which a purchased, effective medication can be substituted for a medication in the non-interfering medication group to achieve the same therapeutic effect. A substitution level of permitted substitution indicates that the purchased, effective medication can still be used in this treatment.

[0234] Step 605: Determine the required quantity based on the replacement medication and patient information.

[0235] The required quantity is the amount needed to replace the medication. This is generally set according to standard quantities. Some settings may be based on patient information, for example, increasing the dosage by 0.5 ml for every 5 kg of body weight. This is determined based on the medication's own requirements.

[0236] Step 606: When the required quantity is less than or equal to the purchased quantity, the replacement drug is deleted from the target drug group to obtain a first target drug group as the recommended drug.

[0237] When the required quantity is less than or equal to the purchased quantity, it means that the effective drugs purchased at this time can not only replace the replacement drugs, but also meet the quantity requirements. Therefore, the target drug group can completely remove the replacement drugs and recommend them in the form of the first target drug group, reducing the amount of drug recommendations and the waste of repeated drug recommendations.

[0238] Step 607: When the demand quantity is greater than the purchased quantity, a difference quantity is obtained based on the demand quantity and the purchased quantity.

[0239] The difference quantity is the quantity required to reach the required quantity. It is calculated by subtracting the purchased quantity from the required quantity.

[0240] When the demand quantity is greater than the purchased quantity, it means that although the purchased effective drugs can replace the alternative drugs, the quantity is insufficient, so it is still necessary to recommend, but the recommended quantity can be smaller.

[0241] Step 608: Update the target drug group based on the difference quantity to obtain a second target drug group as the recommended drugs.

[0242] The number of replacement drugs in the second target drug group is the difference number.

[0243] Step 609: When the replacement level is a preset negative replacement level, the non-interfering drug group is used as the recommended drug.

[0244] The degree of negation of substitution is defined as the degree to which an already purchased effective drug cannot be substituted for a drug in the non-interfering drug group, or the degree to which the substitution cannot achieve the same drug efficacy. If substitution is not possible, only the non-interfering drug group will be automatically recommended.

[0245] Reference Figure 7 Methods for deleting the replacement drug from the target drug group to obtain a first target drug group and using the first target drug group as a recommended drug include:

[0246] Step 700: When the degree of substitution is weaker than a preset equal substitution degree, the substitution drug is not deleted from the target drug group, and the target drug group is still used as the recommended drug.

[0247] The degree of equal substitution is the degree to which a purchased effective drug can achieve the same therapeutic effect by replacing a drug in the non-interfering drug group. If the degree of substitution is weaker than the equal substitution degree, it indicates that the desired therapeutic effect cannot be achieved at this time, so the target drug group is still recommended.

[0248] Step 701: When the degree of substitution is stronger than the degree of equal substitution, the substitution drug is deleted from the target drug group to obtain a first target drug group as the recommended drug.

[0249] When the degree of substitution is stronger than the degree of equal substitution, it indicates that the effect after substitution is better, so the substitution drug is deleted from the target drug group to obtain the first target drug group, which is used as the recommended drug.

[0250] Here, the difference from steps 600-609 is that if the purchased drugs here are not as effective or have fewer side effects as the subsequent drugs, the subsequent drugs will still be recommended, which improves the excellence and scenario analysis capabilities of drug recommendations. The purpose of the former is to match drugs that the user has already purchased and is expected to retain, reduce the amount of drugs purchased, and improve the humanity of recommended drugs.

[0251] Reference Figure 8 , when the degree of substitution is a preset negative substitution degree, the method of recommending the non-interfering drug group as the drug includes:

[0252] Step 800: Based on the purchased effective drugs and the non-interfering drug group, the corresponding purchased interactions are searched from the interaction database.

[0253] The purchased interactions are interactions between the purchased effective drugs and the non-interfering drug group. The establishment of the database has been introduced in the preceding step 504 and will not be described in detail here.

[0254] Step 801: Classify the non-interfering drug group based on the purchased interactions to obtain a purchased interfering drug group, a purchased non-interfering drug group, and a purchased synergistic drug group.

[0255] The purchased interfering drug group consisted of a drug combination in which at least one drug interfered with the purchased effective drug, while the remaining drugs were either non-interfering or synergistic. The purchased non-interfering drug group consisted of a drug combination in which the efficacy of any drug in the group did not interfere with the efficacy of the purchased effective drug. The synergistic drug group consisted of a drug combination in which at least one drug synergized with the purchased effective drug, while the remaining drugs did not interfere with each other. The analysis method was to determine the drug interactions between the drugs in the non-interfering drug group and the purchased effective drug in sequence.

[0256] Step 802: If a purchased synergistic drug group exists, use the purchased synergistic drug group as a recommended drug.

[0257] When a purchased synergistic drug group exists, it means that this combination with the purchased effective drug can also promote synergistic effects, and it will be output as a recommended drug first.

[0258] Step 803: When the purchased synergistic drug group does not exist, the purchased non-interfering drug group is used as the recommended drug.

[0259] When the purchased synergistic drug group does not exist, it means that there are only the purchased interfering drug group and the purchased non-interfering drug group. In this case, only the purchased non-interfering drug group can be output as the recommended drug.

[0260] Step 804: Determine recommended interfering drugs for the purchased interfering drug group based on the purchased interactions when neither the purchased non-interfering drug group nor the purchased synergistic drug group exists.

[0261] Recommended interfering drugs are those in the purchased interfering drug group that interfere with the purchased effective drug. This can be determined by directly searching for the corresponding purchased interactions.

[0262] If neither the purchased non-interfering drug group nor the purchased synergistic drug group exists, then there is only the purchased interfering drug group. Once used, they will interfere with each other and cannot be directly recommended.

[0263] Step 805: A non-recommendation signal is generated based on the recommended interfering drugs, the purchased effective drugs and the purchased interactions.

[0264] The "no-recommendation" signal marks an interfering drug, indicating the reason for the recommendation and requiring the physician to choose between the interfering drug and an already purchased, effective medication. For example, the interfering drug is highlighted in red, while the already purchased, effective medication is also displayed. This allows the physician to simultaneously recommend the interfering drug and remind the patient to remove the already purchased, effective medication, or to not recommend the interfering drug.

[0265] Step 806: Output the purchased interfering drug group as a recommended drug and a non-recommendation signal at the same time.

[0266] Reference Figure 9 The method of simultaneously outputting the purchased interfering drug group as a recommended drug and a non-recommended signal includes:

[0267] Step 900: Based on the recommended interfering drugs and the purchased effective drugs, the corresponding elimination drugs are searched from the preset elimination database.

[0268] Elimination drugs are drugs that counteract the interactions between recommended interfering drugs and purchased effective drugs. Here, the elimination drugs themselves do not have any medicinal efficacy or only have some health-related efficacy. The database stores the mapping relationship between recommended interfering drugs, purchased effective drugs, and elimination drugs. When the conclusion data obtained after the three drugs are tested by staff in this field is that the corresponding side effects are offset, it is recorded. When the system receives the corresponding recommended interfering drugs and purchased effective drugs, it automatically searches for the corresponding elimination drugs from the database and outputs them.

[0269] Step 901: Add the eliminated drug to the purchased non-interfering drug group to form a predicted drug group, and analyze whether it is a non-interfering drug group or a synergistic drug group.

[0270] The predicted drug group is the drug group after the elimination drug is added. Although the elimination drug has a positive effect on eliminating side effects of the recommended interfering drug and the purchased effective drug, it is still uncertain whether it will interact with other drugs, so analysis is still required. The analysis step is to add the elimination drug as a beneficial targeted drug to the drug group, and then determine it according to the method of step 505.

[0271] Step 902: Outputting the predicted drug group as a recommended drug when the predicted drug group is a non-interfering drug group or a synergistic drug group.

[0272] When the expected drug group is a non-interfering drug group or a synergistic drug group, it means that no side effects are formed within the drug group, and the side effects caused by the coexistence of the purchased effective drugs and the recommended interfering drugs are offset. In this case, they can be mixed in the expected drug group and output as recommended drugs.

[0273] Step 903: When the predicted drug group is neither a non-interference drug group nor a synergistic drug group, the purchased non-interference drug group is outputted as a recommended drug and a non-recommendation signal at the same time.

[0274] If not, it means that the drug has not been eliminated or the eliminated drug still has side effects, then the purchased non-interfering drug group will be output as a recommended drug and a non-recommended signal at the same time.

[0275] Based on the same inventive concept, an embodiment of the present invention provides a drug recommendation system.

[0276] Reference Figure 10 , a drug recommendation system, comprising:

[0277] The acquisition module is used to obtain patient diagnosis-related data, the latest scientific research data, the latest test data, adoption probability, recommended expert category, user IP address, patient category, patient information, and purchased drug information;

[0278] A memory for storing a program for a control method of a precise personalized intelligent auxiliary decision-making method for anti-rheumatic drugs;

[0279] The program in the processor memory can be loaded and executed by the processor to realize a control method for a precise personalized intelligent auxiliary decision-making method for anti-rheumatic drugs.

[0280] Those skilled in the art will clearly understand that for the sake of convenience and brevity, the division of the above-mentioned functional modules is only used as an example for illustration. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working processes of the above-mentioned systems, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0281] The above are all preferred embodiments of the present application and are not intended to limit the scope of protection of this application. Unless otherwise stated, any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features. In other words, unless otherwise stated, each feature is merely an example of a series of equivalent or similar features.

Claims

1. A precise and personalized intelligent decision-making method for anti-rheumatic drugs, characterized by: include: Obtain patient diagnosis-related data, including rheumatology-related laboratory test results, joint imaging features, disease activity scores, and comorbidity information; Performing language analysis on patient diagnosis-related data to obtain current keyword information, wherein the current keyword information includes rheumatic disease phenotype keywords and pathological characteristic keywords; Based on the current keyword information, the medicines stored in the preset recommendation database are range-screened to obtain recommended medicines, wherein the recommendation database stores a mapping relationship between keyword information, medicines, and drug efficacy that matches the current keyword information; Based on the current keyword information and recommended drugs, the corresponding recommended drug efficacy is found from the preset recommendation database; A recommendation plan is formed based on the recommended drugs and the efficacy of the recommended drugs and displayed on the preset doctor's end.

2. The method for precise and personalized intelligent decision-making for anti-rheumatic drugs according to claim 1, characterized in that: Also included is a method for updating a recommendation database, the method comprising: Obtain the latest scientific research data, authoritative guidelines on recommended medications for related diseases, and the latest trial data, including non-human trial data and virtual human trial data; Analyze the latest scientific research data and the latest experimental data to extract the latest keyword information, corresponding new drugs and new drug efficacy; Form the latest mapping relationship based on the latest keyword information, new drugs and new drug efficacy; When the latest keyword information and new drugs do not match the keyword information or new drugs in the mapping relationship in the recommendation database, the latest mapping relationship is stored in the recommendation database; When the latest keyword information matches the keyword information in the mapping relationship in the recommendation database, the latest mapping relationship and the mapping relationship are selected and updated based on the efficacy of the new drug and the efficacy of the drug; When a new drug matches a drug in a mapping relationship in a recommendation database, the latest mapping relationship is stored in the recommendation database.

3. The method for precise and personalized intelligent decision-making for anti-rheumatic drugs according to claim 2, characterized in that: Methods for forming a recommendation plan and displaying it on a preset doctor's terminal include: Get the adoption probability of the recommended solution; When the adoption probability is less than the preset critical appropriate probability, the recommended solution is disassembled to obtain a separate recommended solution; Based on the recommended drugs corresponding to the individual recommendation plan, the corresponding drug expert category is searched from the preset doctor database; Get recommended expert categories; When the recommended expert category and the drug expert category are consistent, a recommendation plan is formed and displayed on the doctor's side; No recommendation plan will be formed when the recommended expert category and the drug expert category are inconsistent; When the adoption probability is greater than the critical appropriate probability, a recommendation plan is formed and displayed on the doctor's side.

4. The method for precise and personalized intelligent decision-making for anti-rheumatic drugs according to claim 2, characterized in that: The method for updating the recommendation database further includes: Obtain the user's IP address and patient category when the recommended solution is adopted; Reversely search for the corresponding mapping relationship based on the adopted recommendation solution, and define the mapping relationship as the mapping relationship to be updated; Receive feedback information from the user's IP input; Disassemble feedback information to get actual drug efficacy; Based on the patient category and the mapping relationship to be updated, the corresponding reference degree is found from the preset reference database; The mapping relationship to be updated is updated based on the actual drug efficacy, user category, and reference degree to obtain an updated mapping relationship, and is stored in the recommendation database as a mapping relationship.

5. The method for precise and personalized intelligent decision-making for anti-rheumatic drugs according to claim 1, characterized in that: The method of sequentially screening the drugs stored in the preset recommendation database based on the current keyword information to obtain the recommended drugs includes: Analyze the current keyword information to obtain the current symptom words and the past symptom words; Screening drugs based on the current symptom word to obtain drugs that can treat the current symptom word, and defining the drugs as targeted drugs; Screening targeted drugs based on previous disease words to obtain targeted drugs that are beneficial to the previous disease words, and defining the targeted drugs as beneficial targeted drugs; When the number of beneficial targeted drugs required by the current symptom word is equal to 1, the beneficial targeted drug is used as the recommended drug; When the number of beneficial targeted drugs required by the current symptom word is greater than 1, the corresponding interaction is searched from a preset interaction database based on each beneficial targeted drug corresponding to each current symptom word; Based on the interaction, each beneficial targeted drug corresponding to each current symptom word is self-combined to obtain an interfering drug group, a non-interfering drug group and a synergistic drug group; When a synergistic drug group exists, the synergistic drug group will be used as the recommended drug; When the synergistic drug group does not exist, the non-interfering drug group will be recommended as the drug; When only the interfering drug group exists, a preset alarm signal is output.

6. The method for precise and personalized intelligent decision-making for anti-rheumatic drugs according to claim 5, characterized in that: Methods for recommending a non-interfering group of drugs include: Obtain patient information and purchased medication information; Analyze the purchased drug information to obtain the purchased effective drugs and the purchased quantities; Determine the efficacy of purchased drugs and the efficacy of non-interfering drugs based on drugs in the purchased effective drugs and non-interfering drug groups; Searching for corresponding substitution levels from a preset substitution database based on the purchased effective drugs, the non-interfering drug group, the efficacy of the purchased drugs, and the efficacy of the non-interfering drugs; When the degree of substitution is a preset allowable degree of substitution, the non-interfering drug group is defined as a target drug group, and the replaced drug in the non-interfering drug group is defined as a replacement drug; Determine required quantities based on replacement medication and patient information; When the required quantity is less than or equal to the purchased quantity, the replacement drug is deleted from the target drug group to obtain the first target drug group, which is used as the recommended drug; When the demand quantity is greater than the purchased quantity, the difference quantity is obtained based on the demand quantity and the purchased quantity; updating the target drug group based on the difference quantity to obtain a second target drug group as the recommended drug; When the replacement level reaches the preset negative replacement level, the non-interfering drug group will be used as the recommended drug.

7. The method for precise and personalized intelligent decision-making for anti-rheumatic drugs according to claim 6, characterized in that: Methods for removing the replacement drug from the target drug group to obtain the first target drug group and recommending the first target drug group include: When the degree of substitution is weaker than the preset equal substitution degree, the substituted drug is not deleted from the target drug group, and the target drug group is still used as the recommended drug; When the degree of substitution is stronger than the degree of equal substitution, the substitution drug is deleted from the target drug group to obtain the first target drug group, which is used as the recommended drug.

8. The method for precise and personalized intelligent decision-making for anti-rheumatic drugs according to claim 6, characterized in that: Methods for recommending a non-interfering drug group as a drug when the substitution level is a predetermined negative substitution level include: Based on the purchased effective drugs and non-interfering drug groups, the corresponding purchased interactions are found from the interaction database; Classifying the non-interfering drug group based on the purchased interactions to obtain a purchased interfering drug group, a purchased non-interfering drug group, and a purchased synergistic drug group; When a purchased synergistic drug group exists, the purchased synergistic drug group will be used as the recommended drug; When a purchased synergistic drug group does not exist, the purchased non-interfering drug group will be used as the recommended drug; Determine the recommended interfering drug for the purchased interfering drug group based on the purchased interactions when neither the purchased non-interfering drug group nor the purchased synergistic drug group exists; A non-recommendation signal is formed based on recommended interfering drugs, purchased effective drugs, and purchased interactions; The purchased interfering drug group is outputted simultaneously as a recommended drug and a non-recommended signal.

9. The method for precise and personalized intelligent decision-making for anti-rheumatic drugs according to claim 8, characterized in that: Methods for simultaneously outputting the purchased interfering drug group as a recommended drug and a non-recommended signal include: Based on the recommended interfering drugs and the purchased effective drugs, a corresponding elimination drug is searched from a preset elimination database, and the elimination drug itself has no drug efficacy; Add the eliminated drugs to the purchased non-interfering drug group to form the expected drug group, and analyze whether it is a non-interfering drug group or a synergistic drug group; When the expected drug group is a non-interfering drug group or a synergistic drug group, the expected drug group is output as a recommended drug; When the drug group is not expected to be a non-interfering drug group or a synergistic drug group, the purchased non-interfering drug group is outputted as a recommended drug and a non-recommended signal at the same time.

10. A drug recommendation system, characterized in that: include: The acquisition module is used to obtain patient diagnosis-related data, the latest scientific research data, the latest test data, adoption probability, recommended expert category, user IP address, patient category, patient information, and purchased drug information; A memory for storing a program for a control method of a precise personalized intelligent auxiliary decision-making method for anti-rheumatic drugs according to any one of claims 1 to 9; The processor and the program in the memory can be loaded and executed by the processor to implement a control method for a precise personalized intelligent auxiliary decision-making method for anti-rheumatic drugs as described in any one of claims 1 to 9.

Images