A working method and apparatus for a medical system based on large models and knowledge graphs

Abstract

This disclosure provides a method and apparatus for a medical system based on a large-scale model and knowledge graph, applicable to the field of information-based healthcare. The method includes using a large-scale model to search for content related to a patient's chief complaint from a constructed medical knowledge graph; analyzing the patient's chief complaint based on the searched content to identify the patient's disease and suggest corresponding departments; designating hospitals within a preset range of the patient's location that contain the suggested departments as target hospitals; acquiring appointment data for these target hospitals in real time; generating doctor recommendation results using the large-scale model based on the appointment data and evaluation factors for doctors in each hospital's departments, and performing constraint checks on these recommendations; after passing the constraint checks, using the large-scale model to conduct multiple rounds of inquiries with the patient based on their chief complaint, generating a pre-consultation report and sending it to the patient or one of the recommended doctors. This improves the accuracy of doctor recommendations and the efficiency of medical care, reducing the pressure on hospitals.

Description

Technical Field

[0001] This disclosure relates to the field of information-based healthcare, and in particular to a working method and apparatus for a healthcare system based on a large model and knowledge graph. Background Technology

[0002] In the traditional hospital treatment model, patients face multiple challenges. First, due to a lack of medical knowledge, patients often find it difficult to determine the appropriate department and doctor to consult. For example, chest pain may originate from heart or digestive system diseases, which is difficult for ordinary patients to distinguish. In actual treatment, about 30% of patients choose the wrong department or doctor, leading to delays in treatment and increased time and financial costs. Second, hospital medical resources are generally strained. In addition, the uneven distribution of medical resources causes patients to flock to large hospitals regardless of the severity of their condition, further exacerbating the pressure on medical treatment.

[0003] Therefore, how to accurately recommend suitable departments and doctors to patients, reduce misdiagnosis and delays in treatment, and improve medical efficiency is an urgent problem to be solved. Summary of the Invention

[0004] This disclosure provides a method, apparatus, device, and storage medium for operating a medical system based on a large model and knowledge graph.

[0005] According to a first aspect of this disclosure, a working method for a medical system based on a large model and knowledge graph is provided. The method includes:

[0006] The large model is used to search for content related to the patient's chief complaint from the constructed medical knowledge graph, and the patient's chief complaint is analyzed based on the searched content to identify the patient's disease and suggest the corresponding department for treatment.

[0007] Hospitals located within a preset range of the patient's location and containing the department to be treated are designated as target hospitals, and appointment data of the target hospitals are obtained in real time.

[0008] Based on the appointment data and the evaluation factors of each doctor in the departments of each hospital, a large model is used to generate doctor recommendation results and the doctor recommendation results are subject to constraint checks.

[0009] After the constraint check is passed, the large model is used to conduct multiple rounds of questioning on the patient based on the patient's chief complaint information, and a pre-consultation report is generated and sent to the patient or the doctor in the recommended results.

[0010] In some possible implementations of the first aspect, the medical knowledge graph includes:

[0011] Medical entities and the relationships between them; wherein, the medical entities include departments, doctors, diseases, disease symptoms, treatment methods, examination items, and drugs;

[0012] The relationships between the medical entities are represented in the form of triples.

[0013] In some possible implementations of the first aspect, the step of generating doctor recommendation results using a large model based on the appointment data and the evaluation factors of each doctor in the departments of each hospital includes:

[0014] Based on multiple set recommendation indicators, combined with the appointment data and the evaluation factors of each doctor in the respective departments of each hospital, a large model is used to generate doctor recommendation results; among which,

[0015] The recommended metrics include reinforcement learning composite reward function, each doctor's historical appointment success rate, and patient historical preference matching degree.

[0016] In some possible implementations of the first aspect, the constraint check includes:

[0017] The matching between the doctor's qualifications and the scope of disease treatment in the recommended results is checked from the medical knowledge graph. If a match is found, the KL divergence is used to further constrain the recommended results so that they meet the recommended indicators.

[0018] Among some possible implementations of the first aspect, the method also includes:

[0019] If the qualifications of the doctors in the recommended results do not match the scope of disease treatment, then doctors who meet the qualifications in the department of the target hospital are retrieved from the medical knowledge graph, the retrieved doctors are sorted using a large model, and the doctor recommendation results are regenerated.

[0020] Among some possible implementations of the first aspect, the method also includes:

[0021] For patients without a history of medical visits, obtain the history of medical visits of patients from the same group of patients based on the patient's basic information;

[0022] By using a large model to analyze the historical medical records of similar patient groups, and by obtaining the popularity ranking of the corresponding departments and doctors for the patient's disease from the medical knowledge graph, preliminary doctor recommendation results are generated.

[0023] Among some possible implementations of the first aspect, the method also includes:

[0024] If the constraint check fails, a doctor subgraph that meets the preset conditions is retrieved from the medical knowledge graph, a candidate doctor is selected from the doctor subgraph, the candidate doctor is scored according to the comprehensive scoring function, and the doctor with the highest score among all scores is the final recommended doctor.

[0025] According to a second aspect of this disclosure, a working apparatus for a medical system based on a large model and knowledge graph is provided. The apparatus includes:

[0026] The department identification module is used to use a large model to search for content related to the patient's chief complaint from the constructed medical knowledge graph, and analyze the patient's chief complaint based on the found content to identify the patient's disease and suggest the corresponding department for treatment.

[0027] The real-time appointment data acquisition module is used to select hospitals that are within a preset range from the patient's location and include the department to be treated as target hospitals, and acquire appointment data of the target hospitals in real time.

[0028] The doctor recommendation result acquisition module is used to generate doctor recommendation results using a large model based on the appointment data and the evaluation factors of each doctor in the department of each hospital, and to perform constraint checks on the doctor recommendation results.

[0029] The pre-consultation report generation module is used to conduct multiple rounds of questioning on patients based on their chief complaint information after the examination is deemed satisfactory, using a large model to generate a pre-consultation report and send it to the patient or the recommended doctor.

[0030] According to a third aspect of this disclosure, an electronic device is provided. The electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method described above.

[0031] According to a fourth aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions for causing a computer to perform the methods described above.

[0032] This disclosure provides a method and apparatus for a medical system based on a large-scale model and knowledge graph. The method specifically involves: using a large-scale model to search for content related to the patient's chief complaint from a constructed medical knowledge graph; analyzing the patient's chief complaint based on the found content to identify the patient's disease and suggest corresponding departments; designating hospitals within a preset range of the patient's location that contain the suggested departments as target hospitals; and acquiring appointment data for these target hospitals in real time; generating doctor recommendation results using the large-scale model based on the appointment data and the evaluation factors of doctors in the suggested departments of each hospital, and performing constraint checks on these recommendations; after passing the constraint checks, using the large-scale model to conduct multiple rounds of inquiries with the patient based on their chief complaint, generating a pre-consultation report and sending it to the patient or the recommended doctors. This approach improves the accuracy of doctor recommendations and the efficiency of medical treatment, reducing the pressure on tertiary hospitals.

[0033] It should be understood that the description in the Summary of the Invention is not intended to limit the key or essential features of the embodiments of this disclosure, nor is it intended to restrict the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0034] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. The drawings are provided for a better understanding of the invention and are not intended to limit the scope of this disclosure. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein:

[0035] Figure 1 A flowchart illustrating a working method for a medical system based on a large model and knowledge graph, as provided in an embodiment of this disclosure, is shown.

[0036] Figure 2 This illustration shows a diagram of doctor recommendation results provided in an embodiment of the present disclosure;

[0037] Figure 3 This illustration shows a schematic diagram of a multi-round query process based on a large model provided in an embodiment of this disclosure;

[0038] Figure 4 The diagram shows the working device structure of a medical system based on a large model and knowledge graph according to an embodiment of this disclosure;

[0039] Figure 5 A structural diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure is shown. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0041] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0042] To address the problems in the background technology, this disclosure provides a method and apparatus for a medical system based on a large model and knowledge graph. The method specifically involves: using a large model to search for content related to the patient's chief complaint from a constructed medical knowledge graph; analyzing the patient's chief complaint based on the found content to identify the patient's disease and suggest corresponding departments; designating hospitals within a preset range of the patient's location that contain the suggested departments as target hospitals; and acquiring appointment data for these target hospitals in real time; generating doctor recommendation results using the large model based on the appointment data and the evaluation factors of doctors in the suggested departments of each hospital, and performing constraint checks on the doctor recommendation results; after the constraint checks are passed, using the large model to conduct multiple rounds of inquiries with the patient based on the patient's chief complaint, generating a pre-consultation report and sending it to the patient or the recommended doctors. This approach improves the accuracy of doctor recommendations and the efficiency of medical treatment, reducing the pressure on tertiary hospitals.

[0043] The working method and apparatus of the medical system based on large models and knowledge graphs provided in this disclosure will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0044] Figure 1 This illustration shows a flowchart of a medical system operation method based on a large model and knowledge graph, according to an embodiment of this disclosure. Method 100 includes the following steps:

[0045] S110 uses a large model to search for content related to the patient's chief complaint from the constructed medical knowledge graph, and analyzes the patient's chief complaint based on the found content to identify the patient's disease and suggest the corresponding department for treatment.

[0046] In some embodiments, the large model constructs a dynamic context by retrieving data from the medical knowledge graph in real time, interprets the patient's intent based on the patient's chief complaint, and searches for content related to the patient's chief complaint from the medical knowledge graph based on the interpreted intent. For example, if the patient's chief complaint is fever and abdominal pain, the relevant content searched by the large model from the medical knowledge graph may include that gastroenteritis can cause fever and abdominal pain, influenza can cause fever, food poisoning can cause abdominal pain, and the appropriate department to visit is internal medicine or emergency medicine.

[0047] In some embodiments, the medical knowledge graph includes:

[0048] Medical entities and the relationships between them; wherein, the medical entities include departments, doctors, diseases, disease symptoms, treatment methods, examination items, and drugs;

[0049] The relationships between the medical entities are represented in triplet form, where the triplet form is represented as <entity 1, relation, entity 2>, for example, <department, doctors contained in the department, doctors>, <disease, symptoms of the disease, symptoms of the disease>;

[0050] Furthermore, Table 1 contains information about the disease entity, Table 2 contains information about the department entity, and Table 3 contains information about the doctor entity:

[0051] Table 1

[0052]

[0053] Table 2

[0054]

[0055] Table 3

[0056]

[0057] In some embodiments, the large model in S110 is a pre-trained large model. During the training phase of the large model, before inputting the training samples into the large model, the training samples need to be cleaned and preprocessed. Preprocessing includes deleting duplicate values, null values, and abnormal data; normalizing or standardizing numerical data; segmenting text fields; and removing stop words to ensure data integrity, improve model generalization ability, and adapt to model training requirements. The large model is then trained based on the training samples so that it can search for content related to the patient's chief complaint information from the constructed medical knowledge graph using knowledge retrieval and context understanding methods. Furthermore, the input of the large model includes the patient's chief complaint information.

[0058] S120, select hospitals that are within a preset range from the patient's location and include the department to be treated as target hospitals, and obtain the appointment data of the target hospitals in real time.

[0059] In some embodiments, considering the patient's real-time location, the addresses of hospitals around the patient, and real-time traffic conditions, hospitals within a preset range of the patient's location and containing the department to be treated are selected as target hospitals. Doctors in the corresponding departments of the target hospitals are recommended to the patient. This can reduce the pressure on tertiary hospitals and allow patients with mild symptoms to receive treatment more quickly, reducing the chance of delaying treatment.

[0060] In some embodiments, the appointment data of the target hospital includes the available appointment time slots, the number of remaining appointment slots, and the status of emergency appointments for each doctor in the department where the patient needs to see the patient. Combining the appointment data obtained in real time can reduce information delays, improve the efficiency of patient appointments, and reduce the time patients spend making appointments.

[0061] S130, Based on the appointment data and the evaluation factors of each doctor in the department of each hospital, a large model is used to generate doctor recommendation results and the doctor recommendation results are subjected to constraint checks.

[0062] In some embodiments, evaluation factors for each doctor are determined based on historical patient evaluations of doctors in the departments that patients need to visit within the target hospital. Evaluation factors can be comprehensively determined based on factors such as the doctor's professionalism, recent patient volume, and treatment effectiveness. A higher evaluation factor indicates that the doctor is more popular. Based on the appointment data of the target hospital and the evaluation factors of each doctor in the departments that patients need to visit within the target hospital, a large model is used to generate doctor recommendation results. Comprehensive consideration of doctors' evaluation factors can effectively reduce the chances of excellent doctors being overlooked.

[0063] In some embodiments, generating doctor recommendation results using a large model based on the appointment data and the evaluation factors of each doctor in the respective departments of each hospital includes:

[0064] Based on multiple set recommendation indicators, combined with the appointment data and the evaluation factors of each doctor in the respective departments of each hospital, a large model is used to generate doctor recommendation results; among which,

[0065] The recommended metrics include reinforcement learning composite reward function, each doctor's historical appointment success rate, and patient historical preference matching degree.

[0066] For example, in target hospital A, there are three doctors in the department the patient needs to see: Doctor 1, Doctor 2, and Doctor 3. In target hospital B, there are two doctors in the department the patient needs to see: Doctor 4 and Doctor 5. The evaluation factors of these five doctors, ranked from highest to lowest, are Doctor 4, Doctor 1, Doctor 2, Doctor 3, and Doctor 5. However, the appointment data of target hospital B shows that Doctor 4 has no appointments available that day, while Doctor 1 has appointments available, but appointments are scarce and Doctor 1 has the lowest historical appointment success rate. Doctor 2 also has appointments available that day, and has a high historical appointment success rate, which better matches the patient's historical preferences. Therefore, the reinforcement learning composite reward function value is optimal. In this case, the large model will recommend Doctor 2 as the doctor recommendation result and suggest that the patient go to target hospital A to book an appointment with Doctor 2. Here, the reinforcement learning composite reward function integrates factors such as the distance between the doctor's hospital and the patient, and the matching degree between the doctor's specialty and the patient's disease.

[0067] In some embodiments, method 100 further includes:

[0068] For patients without a history of medical visits, obtain the history of medical visits of patients from the same group of patients based on the patient's basic information;

[0069] By using a large model to analyze the historical medical records of patients in the same group, and by obtaining the popularity ranking of the corresponding departments and doctors for patients' diseases from the medical knowledge graph, preliminary doctor recommendation results are generated.

[0070] For example, if patient A has no medical history at target hospital A, and patient A is the same age and gender as patient B, while patient B has a medical history at target hospital A, then patient A and patient B are grouped into the same category. Based on patient B's medical history at target hospital A, it can be seen that patient B consulted doctor 6 in the same department that patient A needed to see, and doctor 6 appears frequently in this category, or in other words, doctor 6 is very popular in this category. In this case, doctor 6 can be recommended to patient A.

[0071] In some embodiments, doctor recommendation results need to be associated with at least one entity in a medical knowledge graph, which can improve the compliance of medical treatment.

[0072] The doctor recommendation results include the recommended target hospital, the department the patient needs to visit and its corresponding percentage, and the doctor's ranking; such as Figure 2 As shown, the patient's chief complaint mainly describes heart-related problems, so the large model provides... Figure 2 The doctor recommendation results shown in the figure indicate that the recommended departments include cardiology (73.7% recommendation rate), endocrinology (15.8% recommendation rate), and psychiatry (10.5% recommendation rate). The recommended cardiologist is Dr. Zhang, who belongs to the Heart Center; the recommended endocrinologist is Dr. Xu, who belongs to the Comprehensive Internal Medicine Center; and the recommended psychiatrist is Dr. Ding, who also belongs to the Comprehensive Internal Medicine Center. This suggests that Dr. Zhang in cardiology is the best choice for the patient.

[0073] In some embodiments, the constraint check includes:

[0074] The matching between the doctor's qualifications and the scope of disease treatment in the recommended results is checked from the medical knowledge graph. If a match is found, the KL divergence is used to further constrain the recommended results so that they meet the recommended indicators.

[0075] In some embodiments, if the qualifications of the doctors in the recommended results do not match the scope of disease treatment, doctors who meet the qualifications in the department of the target hospital are retrieved from the medical knowledge graph, and the retrieved doctors are sorted using a large model to regenerate the doctor recommendation results. That is, the doctor ranked first in the ranking results is the doctor finally recommended to the patient.

[0076] In some embodiments, the recommendation results are further constrained using KL divergence, including:

[0077] Based on the recommendation method and reference recommendation strategy given in Method 100, construct the KL divergence regularization term;

[0078] Recommendation results are constrained using KL divergence regularization terms;

[0079] Furthermore, the KL divergence regularization term The calculation formula is as follows:

[0080] ;

[0081] in, , These represent the recommendation method and the reference recommendation strategy given by method 100, respectively. , Let CLIP loss function and KL divergence represent the two values, respectively. For the parameters of the large model, is the regularization coefficient, representing the weight of the KL divergence. By adjusting the regularization coefficient, we can prevent the recommendation method given by Method 100 from over-optimizing short-term rewards (such as user click-through rate) and ignoring medical safety (such as doctor qualifications).

[0082] In some embodiments, Calculated using the following formula:

[0083] ;

[0084] in, , , These represent the dominance function, the pruning hyperparameter, and the ratio of action probabilities between the old and new recommendation strategies, respectively. The dominance function measures the performance of an action relative to the average, the pruning hyperparameter limits the update magnitude of the recommendation strategy, and the ratio of action probabilities between the old and new recommendation strategies indicates the update magnitude of the recommendation strategy. Large models continuously update their recommendation strategies while providing the final recommendation results. Indicates time step t Expectations Indicates will The value is limited to between;

[0085] Furthermore, Calculated using the following formula:

[0086] ;

[0087] in, , These represent the average performance of the action measured by the dominance function and the corresponding recommended strategy, respectively. , These represent the new recommendation strategy and the old recommendation strategy, respectively.

[0088] In some embodiments, the constraint check further includes:

[0089] Based on the similarity between the doctor recommendation results given by method 100 and the doctor recommendation results obtained based on medical knowledge graph, the doctor recommendation results given by method 100 are constrained.

[0090] Specifically, based on this method, if the KL divergence If the value exceeds a preset threshold, it indicates that the recommendation result provided by Method 100 deviates too much from the recommendation result obtained based on the medical knowledge graph. It is necessary to re-search the medical knowledge graph or reduce the ranking weight of authoritative doctors in the medical knowledge graph to maintain consistency between the recommendation result provided by Method 100 and the recommendation result obtained based on the medical knowledge graph. The constraint check method is calculated based on this. The formula is as follows:

[0091] ;

[0092] in, , These represent the probabilities of recommended doctors obtained based on medical knowledge graphs and the probabilities of recommended doctors given by method 100, respectively. , These represent the first [item] obtained based on the medical knowledge graph. i The probability and method of the first doctor are given in 100. i The probability of a doctor.

[0093] In some embodiments, patient preferences and basic medical knowledge can also be balanced using KL divergence, and the balancing method is as follows:

[0094] ;

[0095] in, , These represent basic medical knowledge and patient preferences, respectively. These are the weighting coefficients. , Let represent the loss function for DPO and the loss function considering DPO and reference recommendation strategies, respectively. This balancing method can prevent large models from overfitting patient preferences (e.g., frequently choosing non-authoritative doctors) and violating medical knowledge.

[0096] In some embodiments, The calculation formula is as follows:

[0097] ;

[0098] in, This represents the Sigmoid function. This represents the scaling factor, used to control the alignment strength. , These represent the doctor chosen by the patient and the doctor rejected by the patient, respectively. This represents the input to the large model. Indicates from dataset M Triples collected from Expected value;

[0099] According to this formula, the large model can automatically deduce user preferences from user behavior without manually setting reward functions. When the doctor's practice information in the medical knowledge graph changes, the large model updates synchronously through fine-tuning. This ensures that preference learning is always based on the latest medical knowledge graph, thus enabling the model to self-evolve.

[0100] In some embodiments, the reinforcement learning composite reward function can be derived from the formulas above. The calculation formula is expressed as:

[0101] ;

[0102] in, , , , These represent the reward functions for successful patient registration, the distance between the hospital and the patient, the qualifications of the recommending doctor, and the user preference matching, respectively. , , , These represent the coefficients corresponding to each reward function;

[0103] Furthermore, if the patient completes the registration, then =+1.0, if the distance between the hospital and the patient is d ,but If the recommended doctor is highly qualified, then =+0.5, .

[0104] In some embodiments, generalized advantage estimation is used. The large model is optimized using the following formula:

[0105] ;

[0106] in, , , These are the discount factor (typically between 0.9 and 0.99), the GAE hyperparameter (typically 0.95), and the distance from the current time step to the future time step. For time step t The timing difference error at point A is calculated using the following formula:

[0107] ;

[0108] in, , , They are t Reward value at any moment t Real-time Recommendation Results S Long-term expected returns t Recommendation results at time +1 S The long-term expected return.

[0109] S140, after the constraint check is passed, the large model is used to conduct multiple rounds of questioning on the patient based on the patient's chief complaint information, and a pre-consultation report is generated and sent to the patient or the doctor in the recommended results.

[0110] In some embodiments, method 100 further includes:

[0111] If the constraint check fails, a doctor subgraph that meets the preset conditions is retrieved from the medical knowledge graph, a candidate doctor is selected from the doctor subgraph, the candidate doctor is scored according to the comprehensive scoring function, and the doctor with the highest score among all scores is the final recommended doctor.

[0112] For example, when the constraint check fails, the system retrieves data from the medical knowledge graph that satisfies (the area of ​​expertise of doctor N). (Patient's inquiry needs) (The distance between the patient and the doctor) The doctor subgraph is used to select candidate doctors. These candidates are then scored according to a comprehensive scoring function, and the doctor with the highest score is selected as the final recommended doctor. The comprehensive scoring function... The calculation formula is as follows:

[0113] ;

[0114] in, , , These are, respectively, the normalization operation of the evaluation factors for doctor N, the distance between patients and doctors, and the availability of appointments for doctor N. , , These are the coefficients. .

[0115] In some embodiments, the large model has multi-round query capabilities, such as Figure 3 As shown, a patient with user ID 123456 inputs "I have a headache today" into the large model. The large model then conducts multiple rounds of questioning based on the patient's input. The question-and-answer process is as follows: Q: When did the headache start today? Was it accompanied by nausea or blurred vision? A: This morning, accompanied by nausea; Q: Did your nausea this morning start suddenly or gradually worsen? Were there any accompanying vomiting or other symptoms? A: It started suddenly, accompanied by vomiting; Q: When exactly did the nausea and vomiting begin? Was it continuous or intermittent? A: 9:00 AM, intermittent; Q: What is the time interval between nausea and vomiting? Did it occur after 9:00 AM? Through the above dialogue, it can be seen that the large model is mimicking a doctor's inquiry into the patient's condition. Based on these inquiries, the large model can generate a detailed pre-diagnosis report, which the patient can review and recommend to a doctor for diagnosis, thus improving treatment efficiency. The pre-diagnosis report includes the patient's chief complaint, medical history, allergy history, family history, etc.

[0116] In some embodiments, the large model can parse images uploaded by patients, extract text information from the images, perform in-depth analysis of the text information, and combine it with a medical knowledge graph to structure the text information and generate corresponding analysis results. For example, the large model can parse the text content of a patient's uploaded report, which includes the parsing process, detailed interpretation of costs, and analysis of examination indicators. For example, the parsing process involves carefully reading and understanding the various information items on the invoice, identifying key data points such as cost items, amounts, and payment methods. The detailed interpretation of costs could be: treatment fee, 9.10 yuan; non-invasive ECG monitoring, 1 hour, 10.00 yuan; intravenous infusion, 1 time, 9.10 yuan; total cost 28.20 yuan. However, the total on the invoice is 2110.55 yuan, which is obviously inconsistent with the calculation and may indicate information omissions or misunderstandings. The payment method is: medical insurance payment, 10.55 yuan. Providing these analysis results to doctors helps them make medical decisions.

[0117] In some embodiments, patients can also interact with the large model as follows: after the patient selects the disease they wish to search for, the large model quickly searches for relevant information about the disease from the medical knowledge graph, including disease introduction, etiology, treatment, prevention, and required examinations. This relevant information is then integrated to generate a comprehensive disease database. For example, the comprehensive disease database may include disease introduction, causes, examination items, disease name, and preventive measures. For instance, the disease introduction could be: Pertussis is an acute respiratory infectious disease caused by Bordetella pertussis, characterized by paroxysmal spasmodic coughing, accompanied by an infectious whooping inspiratory sound at the end of the coughing phase. The course of the disease is relatively long, lasting several weeks or even up to 3 months, hence the name "whooping cough." It is most common in children under 5 years old. Clinically, pertussis can lead to complications such as suffocation, pneumonia, and encephalopathy, resulting in a high mortality rate. In patients infected with Bordetella pertussis, males are most susceptible to infection and carriers. The most infectious period is 2-3 weeks from the onset of paroxysmal spasmodic coughing. The course of pertussis is shorter than that of catarrhal coughing, and the typical course is divided into 3 stages. Stages: (1) Catarrhal stage, coughing begins at the onset of the disease and lasts for 1-2 weeks; (2) Spasmodic stage, generally lasts for 2-4 weeks or longer, characterized by paroxysmal spasmodic coughing.(3) Recovery period, generally 1-2 weeks, the frequency and severity of cough attacks decrease, and paroxysmal spasmodic cough no longer occurs. The peripheral blood white blood cell count is generally significantly increased, with lymphocytes as the main component. When diagnosing this disease, it is important to differentiate it from bronchial foreign bodies and hilar lymph node tuberculosis. In recent years, the incidence of this disease in infants and adults has shown an increasing trend. The cause could be: Pertussis is caused by Bordetella pertussis, a bacterium belonging to the genus Bordetella. There are four known species of Bordetella: Bordetella pertussis, Bordetella parapertussis, Bordetella bronchiseptica, and Bordetella avium. Bordetella avium generally does not cause human disease, only infecting birds. Bordetella pertussis is approximately 1.0-1.5 μm long and 0.3-0.5 μm wide, has a membrane, is non-motile, Gram-negative, aerobic, non-toxic, and non-flagellated. When stained with toluidine blue, the ends stain more deeply. Bacterial culture requires a large amount (15%-25%) of fresh blood for optimal reproduction; therefore, Bordetella-Gold medium (blood, glycerol, and potato) is commonly used to isolate colonies. Bordetella pertussis grows slowly; after 3-7 days of growth in a suitable environment at 35-37℃, small, opaque colonies develop. The initial colonies are raised and smooth, a smooth (S) type, also known as type I. Phase IV bacteria are characterized by uniform morphology, the presence of a capsule, and strong virulence and antigenicity, resulting in high pathogenicity. If isolated colonies are cultured further in a normal culture medium, the colonies change from smooth to rough (R) type, which is called Phase IV bacteria. These bacteria lack a capsule, lose their virulence and antigenicity, and thus lose their pathogenicity. Phase II and Phase III are intermediate transitional types. Bordetella pertussis can produce many toxins. Five toxins are known: (1) Pertussis exotoxin, a protein in Bordetella pertussis cells, formerly known as leukocytosis or lymphocyte promoting factor (LPE), histamine sensitizing factor (HSF), and insulin activating protein (IAP). Pertussis exotoxin has identified its non-promoting subunit (52-55S) and subunit (52-55S) as non-toxic protein components. It can bind to the host cell membrane and mediate toxicity through the enzymatic subunit S1. S1 can catalyze the separation of some ribose from nicotinamide adenine dinucleotide through the activity of adenosine diphosphate-ribonucleotide, and transfer it to the cell membrane to inhibit the binding of guanosine triphosphate to G protein units, leading to cell degeneration. At the same time, it can also promote lymphocyte increase, activate islet cells and enhance immune response; (2) The heat-resistant endotoxin (ET) can only be partially destroyed at 100℃ for 60 minutes and can only be inactivated at 180℃. This toxin can cause fever and immune response in the body.(3) Heat-labile toxin: This toxin can be destroyed by heating at 55°C for 30 minutes. Antibodies against this toxin have no protective effect against Bordetella pertussis infection. (4) Tracheal cell toxin: It can damage the ciliated epithelial cells of the host's respiratory tract, causing them to degenerate and die. (5) Adenosine cyclase toxin: An enzyme present on the surface of Bordetella pertussis cells. After entering blood cells, this enzyme can be activated by caloprotein, catalyzing the production of cyclic adenosine monophosphate, interfering with phagocytosis, and inhibiting the chemotaxis of neutrophils and the bactericidal ability of phagocytes, enabling it to cause continuous infection. Adenosine cyclase toxin is also a hemolysin that can cause hemolysis. The parietal antigen of pertussis consists of two hemolysin active antigens of Bordetella pertussis, one of which is filamentous hemagglutinin (FHA). Because it originates from the pili on the bacterial cell surface, it is also called pili antigen. Filamentous hemagglutinin plays a decisive role in the adhesion of Bordetella pertussis to respiratory epithelial cells and is the main cause of pathogenicity. Experiments have shown that immunization of mice with filamentous hemagglutinin can resist lethal challenges from Bordetella pertussis, therefore filamentous hemagglutinin is a protective antigen. Another agglutinogen (AGG) is a protein component in the outer membrane and pili of Bordetella pertussis, mainly containing three serotypes of clotting factors 1, 2, and 3. AGG-1 is species-specific; AGG-2 and AGG-3 are type-specific. The local prevalence is understood by detecting the type of agglutinogen. Currently, it is believed that the corresponding antibodies to these two hemagglutinin membrane antigens are protective antibodies. Bordetella pertussis is classified into seven agglutinogens based on the different antigenicity of different agglutinogens. Type 1 agglutinogen is present in all Bordetella pertussis, type 7 agglutinogen is common to Bordetella spp. (including Bordetella parapertussis and Bordetella bronchiseptica), and types 2-6... Bordetella pertussis is classified into different serotypes based on different combinations. Serotype determination is mainly used to study the serotype of prevalent strains and to select specific serotype strains for vaccine production. In addition, there is no cross-immunity between Bordetella parapertussis and Bordetella pertussis, and it can also cause epidemics. Bordetella pertussis is highly resistant to external physical and chemical factors. After entering the respiratory tract of susceptible individuals with airborne droplets, the bacterial filamentous hemoglobin adheres to the surface of the ciliated epithelial cells of the pharynx and bronchioles. Subsequently, the bacteria multiply locally and produce various toxins such as pertussis exotoxin and adenylate cyclase, causing epithelial cell ciliary paralysis and cell degeneration, reducing protein synthesis, causing epithelial cell necrosis and shedding, leading to a systemic reaction. Due to the pathological changes in epithelial cells and ciliary paralysis, mucus and necrotic epithelium accumulate and accumulate in the bronchioles. During the recovery period, the respiratory tract is constantly stimulated by nerve afferents, which are transmitted to the cerebral cortex and medullary cough center, reflexively causing spasmodic cough. During the recovery period, crying or other infections can also induce pertussis-like spasmodic cough. The tests may include: white blood cell differential count, complete blood count, ear, nose, and throat swab bacterial culture, and peripheral blood white blood cell count and differential test. The disease name could be "whooping cough" (pertussis).

[0118] According to embodiments of this disclosure, a method and apparatus for operating a medical system based on a large-scale model and knowledge graph are provided. The method specifically involves: using a large-scale model to search for content related to the patient's chief complaint from a constructed medical knowledge graph; analyzing the patient's chief complaint based on the found content to identify the patient's disease and suggest corresponding departments; designating hospitals within a preset range of the patient's location that contain the suggested departments as target hospitals; and acquiring appointment data for the target hospitals in real time; generating doctor recommendation results using the large-scale model based on the appointment data and the evaluation factors of each doctor in the suggested departments of each hospital, and performing constraint checks on the doctor recommendation results; after the constraint checks are passed, using the large-scale model to conduct multiple rounds of inquiries with the patient based on the patient's chief complaint, generating a pre-consultation report and sending it to the patient or the doctor in the recommendation results. This approach improves the accuracy of doctor recommendations and the efficiency of medical treatment, reducing the pressure on tertiary hospitals.

[0119] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this disclosure is not limited to the described order of actions, because according to this disclosure, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this disclosure.

[0120] The above is an introduction to the method embodiments. The following describes the solution described in this disclosure further through device embodiments.

[0121] Figure 4 This illustration shows a structural diagram of a working device for a medical system based on a large model and knowledge graph, according to an embodiment of this disclosure. The device 400 includes:

[0122] The department identification module 410 is used to use a large model to search for content related to the patient's chief complaint information from the constructed medical knowledge graph, and to analyze the patient's chief complaint information based on the found relevant content, identify the patient's disease, and suggest the corresponding department for treatment.

[0123] The real-time appointment data acquisition module 420 is used to select hospitals that are within a preset range from the patient's location and include the department to be treated as target hospitals, and acquire appointment data of the target hospitals in real time.

[0124] The doctor recommendation result acquisition module 430 is used to generate doctor recommendation results using a large model based on the appointment data and the evaluation factors of each doctor in the department of each hospital, and to perform constraint checks on the doctor recommendation results.

[0125] The pre-consultation report generation module 440 is used to conduct multiple rounds of questioning on the patient based on the patient's chief complaint information after the constraint examination is qualified, and generate a pre-consultation report to be sent to the patient or the doctor in the recommended results.

[0126] In some embodiments, the department determination module 410 is specifically used for:

[0127] The medical knowledge graph includes:

[0128] Medical entities and the relationships between them; wherein, the medical entities include departments, doctors, diseases, disease symptoms, treatment methods, examination items, and drugs;

[0129] The relationships between the medical entities are represented in the form of triples.

[0130] In some embodiments, the doctor recommendation result acquisition module 430 is specifically used for:

[0131] The process of generating doctor recommendation results using a large model based on the appointment data and the evaluation factors of each doctor in the respective departments of each hospital includes:

[0132] Based on multiple set recommendation indicators, combined with the appointment data and the evaluation factors of each doctor in the respective departments of each hospital, a large model is used to generate doctor recommendation results; among which,

[0133] The recommended metrics include reinforcement learning composite reward function, each doctor's historical appointment success rate, and patient historical preference matching degree.

[0134] In some embodiments, the doctor recommendation result acquisition module 430 is further used for:

[0135] The constraint check includes:

[0136] The matching between the doctor's qualifications and the scope of disease treatment in the recommended results is checked from the medical knowledge graph. If a match is found, the KL divergence is used to further constrain the recommended results so that they meet the recommended indicators.

[0137] In some embodiments, the device 400 is further configured to:

[0138] If the qualifications of the doctors in the recommended results do not match the scope of disease treatment, then doctors who meet the qualifications in the department of the target hospital are retrieved from the medical knowledge graph, the retrieved doctors are sorted using a large model, and the doctor recommendation results are regenerated.

[0139] In some embodiments, the device 400 is further configured to:

[0140] For patients without a history of medical visits, obtain the history of medical visits of patients from the same group of patients based on the patient's basic information;

[0141] By using a large model to analyze the historical medical records of similar patient groups, and by obtaining the popularity ranking of the corresponding departments and doctors for the patient's disease from the medical knowledge graph, preliminary doctor recommendation results are generated.

[0142] In some embodiments, the device 400 is further configured to:

[0143] If the constraint check fails, a doctor subgraph that meets the preset conditions is retrieved from the medical knowledge graph, a candidate doctor is selected from the doctor subgraph, the candidate doctor is scored according to the comprehensive scoring function, and the doctor with the highest score among all scores is the final recommended doctor.

[0144] Understandable Figure 4 Each module / unit in the apparatus 400 shown has the function of implementing each step in the method 100 provided in the embodiments of this disclosure, and can achieve its corresponding technical effect. For the sake of brevity, it will not be described in detail here.

[0145] Figure 5 A structural diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure is shown. Electronic device 500 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic device 500 may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0146] like Figure 5 As shown, the electronic device 500 includes a computing unit 501, which can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) 502 or a computer program loaded from a storage unit 508 into a random access memory (RAM) 503. The RAM 503 may also store various programs and data required for the operation of the electronic device 500. The computing unit 501, ROM 502, and RAM 503 are interconnected via a bus 504. An I / O interface 505 is also connected to the bus 504.

[0147] Multiple components in electronic device 500 are connected to I / O interface 505, including: input unit 506, such as keyboard, mouse, etc.; output unit 507, such as various types of monitors, speakers, etc.; storage unit 508, such as disk, optical disk, etc.; and communication unit 509, such as network card, modem, wireless transceiver, etc. Communication unit 509 allows electronic device 500 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0148] The computing unit 501 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 501 performs the various methods and processes described above, such as method 100. For example, in some embodiments, method 100 may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 508. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 500 via ROM 502 and / or communication unit 509. When the computer program is loaded into RAM 503 and executed by the computing unit 501, one or more steps of method 100 described above may be performed. Alternatively, in other embodiments, the computing unit 501 may be configured to perform method 100 by any other suitable means (e.g., by means of firmware).

[0149] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0150] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0151] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0152] It should be noted that this disclosure also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute method 100 and achieve the corresponding technical effects achieved by executing the method in the embodiments of this disclosure. For the sake of brevity, they will not be described in detail here.

[0153] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including voice input, speech input, or tactile input).

[0154] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0155] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0156] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0157] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A working method for a medical system based on large models and knowledge graphs, characterized in that, include: The large model is used to search for content related to the patient's chief complaint from the constructed medical knowledge graph, and the patient's chief complaint is analyzed based on the searched content to identify the patient's disease and suggest the corresponding department for treatment. Hospitals located within a preset range of the patient's location and containing the department to be treated are designated as target hospitals, and appointment data of the target hospitals are obtained in real time. Based on the appointment data and the evaluation factors of each doctor in the departments of each hospital, a large model is used to generate doctor recommendation results. This includes: generating doctor recommendation results using a large model based on multiple set recommendation indicators, combined with the appointment data and the evaluation factors of each doctor in the departments of each hospital. The recommendation indicators include a reinforcement learning composite reward function, the historical appointment success rate of each doctor, and the patient's historical preference matching degree. The reinforcement learning composite reward function includes a reward function for successful patient appointments, a reward function for the distance between the hospital and the patient, a reward function for the recommending doctor's qualifications, and a reward function for user preference matching. The doctor recommendation results are subjected to constraint checks. If the constraint checks are satisfactory, a large model is used to conduct multiple rounds of questioning based on the patient's chief complaint information, generating a pre-consultation report which is sent to the patient or the doctor in the recommendation results. The constraint checks include: verifying the matching between the doctor's qualifications and the scope of disease treatment in the recommendation results from a medical knowledge graph. If a match is found, the recommendation results are further constrained using KL divergence, and KL divergence regularization terms are used to constrain the recommendation results to ensure they meet the recommendation criteria. The KL divergence is used to balance patient preferences and basic medical knowledge, and the balancing methods include: , , These represent basic medical knowledge and patient preferences, respectively. These are the weighting coefficients. , Let represent the loss function for DPO, the loss function considering both DPO and the reference recommendation strategy, and the regularization term of the KL divergence, respectively. The calculation formula is expressed as: , , These respectively represent the recommended methods and reference recommendation strategies provided by the medical system's working method. , Let CLIP loss function and KL divergence represent the two values, respectively. For the parameters of the large model, The regularization coefficient represents the weight of the KL divergence, so as to prevent over-optimization of short-term rewards by adjusting the regularization coefficient; wherein, the large model is updated based on the medical knowledge graph to achieve self-evolution function.

2. The method according to claim 1, characterized in that, The medical knowledge graph includes: Medical entities and the relationships between them; wherein, the medical entities include departments, doctors, diseases, disease symptoms, treatment methods, examination items, and drugs; The relationships between the medical entities are represented in the form of triples.

3. The method according to claim 1, characterized in that, The method further includes: If the qualifications of the recommended doctors do not match the scope of disease treatment, then doctors who meet the qualifications in the department of the target hospital are retrieved from the medical knowledge graph, and the retrieved doctors are sorted using a large model to regenerate the doctor recommendation results.

4. The method according to claim 1, characterized in that, The method further includes: For patients without a history of medical visits, obtain the history of medical visits of patients from the same group of patients based on the patient's basic information; By using a large model to analyze the historical medical records of similar patient groups, and by obtaining the popularity ranking of the corresponding departments and doctors for the patient's disease from the medical knowledge graph, preliminary doctor recommendation results are generated.

5. The method according to claim 1, characterized in that, The method further includes: If the constraint check fails, a doctor subgraph that meets the preset conditions is retrieved from the medical knowledge graph, a candidate doctor is selected from the doctor subgraph, the candidate doctor is scored according to the comprehensive scoring function, and the doctor with the highest score among all scores is the final recommended doctor.

6. A working device for a medical system based on a large model and knowledge graph, characterized in that, include: The department identification module is used to use a large model to search for content related to the patient's chief complaint from the constructed medical knowledge graph, and analyze the patient's chief complaint based on the found content to identify the patient's disease and suggest the corresponding department for treatment. The real-time appointment data acquisition module is used to select hospitals that are within a preset range from the patient's location and include the department to be treated as target hospitals, and acquire appointment data of the target hospitals in real time. The doctor recommendation result acquisition module is used to generate doctor recommendation results using a large model based on the appointment data and the evaluation factors of each doctor in the departments of each hospital. This includes: generating doctor recommendation results using a large model based on multiple set recommendation indicators, combined with the appointment data and the evaluation factors of each doctor in the departments of each hospital. The recommendation indicators include a reinforcement learning composite reward function, each doctor's historical appointment success rate, and the patient's historical preference matching degree. The reinforcement learning composite reward function includes a reward function for successful patient appointments, a reward function for the distance between the hospital and the patient, a reward function for the recommending doctor's qualifications, and a reward function for user preference matching. The pre-consultation report generation module is used to perform constraint checks on the doctor recommendation results. After passing the constraint check, a large model is used to conduct multiple rounds of questioning on the patient based on the patient's chief complaint information, generating a pre-consultation report which is sent to the patient or the doctor in the recommendation results. The constraint check includes: verifying the matching between the doctor's qualifications and the scope of disease treatment in the recommendation results from a medical knowledge graph. If a match is found, the recommendation results are further constrained using KL divergence, and further constrained using KL divergence regularization terms to ensure the recommendation results meet the recommendation criteria. The KL divergence is used to balance patient preferences and basic medical knowledge, and the balancing method includes: , , These represent basic medical knowledge and patient preferences, respectively. These are the weighting coefficients. , Let represent the loss function for DPO, the loss function considering both DPO and the reference recommendation strategy, and the regularization term of the KL divergence, respectively. The calculation formula is expressed as: , , These respectively represent the recommended methods and reference recommendation strategies provided by the medical system's working method. , Let CLIP loss function and KL divergence represent the two values, respectively. For the parameters of the large model, The regularization coefficient represents the weight of the KL divergence, so as to prevent over-optimization of short-term rewards by adjusting the regularization coefficient; wherein, the large model is updated based on the medical knowledge graph to achieve self-evolution function.

7. An electronic device, characterized in that, include: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method described in any one of claims 1-5.

8. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-5.

Images