Multi-role collaborative information processing method and device based on large language model

Abstract

The present invention relates to the field of computer application technology, and discloses a multi-role collaborative information processing method and device based on a large language model, the method comprising: obtaining basic information and question information of a target object; analyzing and processing the basic information and question information to obtain a plurality of corresponding prompt words; analyzing the prompt words through a pre-established guidance agent, matching a plurality of pre-established information processing agents, wherein the plurality of information processing agents are configured to adapt to different information processing needs; analyzing the basic information and question information through each information processing agent to obtain the speech results corresponding to each information processing agent; summarizing the speech results of each information processing agent through a pre-established consultation agent to obtain the final processing result, wherein the guidance agent, each information processing agent, and the consultation agent are established through a large language model. The problem analysis results and suggestions obtained by the present invention are more comprehensive and reliable.

Description

Technical Field

[0001] The present invention relates to the field of computer application technology, and in particular to a multi-role collaborative information processing method and device based on a large language model. Background Art

[0002] In intelligent diagnosis and treatment systems, most existing methods are based on traditional methods or a single large language model (LLM) for auxiliary diagnosis and treatment. However, when a single LLM agent faces relevant medical questions raised by users, the answers obtained may appear to be relatively simple in some aspects, and the analysis results obtained may be far from the actual situation. Summary of the invention

[0003] In view of this, the present invention provides a multi-role collaborative information processing method and device based on a large language model to solve the problem that the results obtained by a single LLM agent analysis are far from the actual situation.

[0004] In the first aspect, the present invention provides a multi-role collaborative information processing method, the method comprising: acquiring basic information and question information of a target object; analyzing and processing the basic information and question information to obtain a corresponding plurality of prompt words; analyzing the prompt words through a pre-established guidance agent, matching a plurality of pre-established information processing agents, wherein the plurality of information processing agents are configured to adapt to different information processing needs; analyzing the basic information and question information through each information processing agent to obtain a speech result corresponding to each information processing agent; summarizing the speech results of each information processing agent through a pre-established consultation agent to obtain a final processing result, wherein the guidance agent, each information processing agent, and the consultation agent are established through a large language model.

[0005] The method provided by the embodiment of the present invention, after obtaining the basic information and question information of the target object, matches multiple information processing agents according to the basic information and question information, and the multiple information processing agents respectively obtain their own collaborative processing results, and summarize the collaborative processing results to obtain the final processing result. In the embodiment of the present invention, the collaborative processing results are obtained by multiple information processing agents. On the one hand, through division of labor, the ability of information processing agents to analyze problems is increasingly refined. On the other hand, multiple agents who are good at different information processing needs debate and discuss, which can obtain better results than when there is no specialization, thereby improving the efficiency and output quality of the entire problem processing, effectively reducing the problem of incomplete analysis caused by a single information processing agent, and effectively alleviating the hallucination problem that a single information processing agent is prone to, making the problem analysis results and suggestions more comprehensive and reliable.

[0006] In an optional implementation, the basic information and question information are analyzed and processed to obtain a corresponding plurality of prompt words, including: based on the types of the basic information and question information, calling a corresponding recognition algorithm to recognize the basic information and question information to obtain a recognition result; the recognition algorithm includes: at least one of a text recognition algorithm, a speech recognition algorithm, and an image recognition algorithm; based on the recognition result, determining the information processing requirements of the target object; based on the information processing requirements, generating a plurality of prompt words.

[0007] In an optional implementation, the guide agent, each information processing agent, and the consultation agent are generated by inputting different prompt projects into a large language model, and different information processing agents correspond to different prompt projects; the prompt words, basic information, and question information are analyzed by a pre-established guide agent, and multiple pre-established information processing agents are matched, including: comparing the prompt words with the prompt projects corresponding to each information processing agent, and selecting multiple information processing agents based on the matching results.

[0008] In an optional embodiment, each information processing agent analyzes the basic information and question information to obtain the speech results corresponding to each information processing agent, including: inputting the basic information and question information into the ith information processing agent in a preset order to obtain the speech result of the ith information processing agent, where i is 1; inputting the speech result, basic information and question information of the ith information processing agent into the (i+1)th information processing agent to obtain the speech result of the (i+1)th information processing agent; if the value of (i+1) is less than n, adding 1 to the value of i, and returning to the step of inputting the speech result, basic information and question information of the ith information processing agent into the (i+1)th information processing agent to obtain the speech result of the (i+1)th information processing agent, until the value of (i+1) is equal to n, and obtaining the speech results corresponding to each information processing agent.

[0009] In an optional implementation, each information processing agent analyzes basic information and question information to obtain the speech results corresponding to each information processing agent, including: inputting the basic information and question information into the ith information processing agent to obtain the speech result of the ith information processing agent, where i is 1; the ith information processing agent randomly selects an information processing agent that has not spoken from each information processing agent as the (i+1)th agent; inputting the speech result, basic information and question information of the ith information processing agent into the (i+1)th information processing agent to obtain the speech result of the (i+1)th information processing agent; if the value of (i+1) is less than n, adding 1 to the value of i, and returning to the step of the ith information processing agent randomly selecting an information processing agent that has not spoken from each information processing agent as the (i+1)th agent, until the value of (i+1) is equal to n, to obtain the speech results corresponding to each information processing agent.

[0010] In an optional implementation, each information processing agent analyzes basic information and question information to obtain the speech results corresponding to each information processing agent, including: inputting basic information and question information into the ith information processing agent to obtain the speech result of the ith information processing agent, where i is 1; comparing the speech result of the ith information processing agent with the prompting project of the information processing agents that have not spoken among the information processing agents through the consultation agent, and selecting the information processing agent with the highest correlation with the speech result of the ith information processing agent as the (i+1)th agent; and selecting the ith information processing agent as the (i+1)th agent. The speech results, basic information and question information of the agent are input into the (i+1)th information processing agent to obtain the speech results of the (i+1)th information processing agent; if the value of (i+1) is less than n, the value of i is increased by 1, and the consultation agent is used to compare the speech results of the i-th information processing agent with the prompt project of the information processing agents that have not spoken among the information processing agents, and the information processing agent with the highest correlation with the speech results of the i-th information processing agent is selected as the step of the (i+1)th agent, until the value of (i+1) is equal to n, and the speech results corresponding to each information processing agent are obtained.

[0011] In a second aspect, the present invention provides an information processing device for multi-role collaboration based on a large language model, the device comprising: an information acquisition module for acquiring basic information and question information of a target object; a data analysis module for analyzing and processing the basic information and question information to obtain a corresponding plurality of prompt words; an information processing agent matching module for analyzing the prompt words through a pre-established guidance agent and matching a plurality of pre-established information processing agents, wherein the plurality of information processing agents are configured to adapt to different information processing needs; a processing result acquisition module for analyzing the basic information and question information through each information processing agent to obtain the speech results corresponding to each information processing agent; a processing result integration module for summarizing the speech results of each information processing agent through a pre-established consultation agent to obtain the final processing result, wherein the guidance agent, each information processing agent and the consultation agent are established through a large language model.

[0012] In an optional implementation, the guidance agent, each information processing agent, and the consultation agent are generated by inputting different prompt projects into a large language model respectively. Different information processing agents correspond to different prompt projects. The data analysis module is specifically used to compare the prompt words with the prompt projects corresponding to each information processing agent, and select multiple information processing agents based on the matching results.

[0013] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to execute the multi-role collaborative information processing method based on a large language model according to the first aspect or any corresponding embodiment thereof.

[0014] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the information processing method based on multi-role collaboration of a large language model according to the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 This is a schematic diagram of the auxiliary diagnosis and treatment process based on a single LLM;

[0017] Figure 2 is a flowchart of a multi-role collaborative information processing method based on a large language model according to an embodiment of the present invention;

[0018] Figure 3 is a structural block diagram of an information processing device based on multi-role collaboration of a large language model according to an embodiment of the present invention;

[0019] Figure 4 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0021] To assist patients in medical consultation, patients can ask health questions or describe their symptoms to the intelligent diagnosis and treatment system, and the system will provide medical diagnosis and advice, which will help improve the convenience and timeliness of patients' access to medical information and advice.

[0022] In the intelligent diagnosis and treatment system, most of them are based on traditional methods or single LLM for auxiliary diagnosis and treatment. The auxiliary diagnosis and treatment process based on a single LLM is as follows: Figure 1 As shown, after the user inputs the patient's symptoms and related information, a single LLMagent replies with an answer, and a consultation is completed.

[0023] Although LLM-based methods can flexibly provide explanations and suggestions by combining multiple data such as medical images, electronic health records, and genomics, and have advanced medical reasoning capabilities, the analysis results obtained by a single LLM are not comprehensive enough and are prone to hallucinations.

[0024] According to an embodiment of the present invention, an embodiment of an information processing method based on multi-role collaboration of a large language model is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.

[0025] In this embodiment, a multi-role collaborative information processing method based on a large language model is provided. Figure 2 is a flowchart of an information processing method based on multi-role collaboration of a large language model according to an embodiment of the present invention. Figure 2As shown, the process includes the following steps:

[0026] Step S101, obtaining basic information and question information of the target object.

[0027] In an optional embodiment, when the method provided by the embodiment of the present invention is applied to an intelligent diagnosis and treatment system, the target object may be a patient, and the basic information of the target object includes patient information and symptom information, wherein the patient information includes name, age, gender, etc., and the symptom information includes one or more of symptom description, medical history, drug use, allergy history, medical records, medical records, examination results, etc. The question information includes relevant information such as diagnosis and treatment suggestions and questions about symptoms.

[0028] Step S102: Analyze and process the basic information and question information to obtain a plurality of corresponding prompt words.

[0029] In an optional embodiment, analyzing and processing the basic information and question information refers to preprocessing the basic information and question information input by the user so that the preprocessed prompt words can be better understood by the LLM.

[0030] In an optional embodiment, each agent used in the embodiment of the present invention is generated by inputting a prompt project into a large language model, and the prompt words obtained by analyzing and processing the basic information and question information are in the same form as the prompt project, so that the prompt words can be compared with the prompt project of each information processing agent, and the information processing agent suitable for answering the current question can be selected.

[0031] After obtaining the basic information and question information, they are analyzed and processed in order to expand the agent's perception space from text-only input to multimodal input, including text, auditory and visual modalities.

[0032] Step S103, analyzing the prompt words through the pre-established guidance agent, matching multiple pre-established information processing agents, and the multiple information processing agents are configured to adapt to different information processing requirements.

[0033] In the embodiment of the present invention, multiple information processing agents are pre-established, each of which is played by LLM, and different information processing agents are obtained by inputting different prompt projects into the large language model, so that different information processing agents can play different roles, have different functions, and perform different tasks. Therefore, different information processing agents can adapt to different information processing needs. Since different information processing agents are adapted to different information processing needs, each information processing agent is not suitable for all questions. After obtaining the user's basic information and question information, the appropriate information processing agent can be matched based on the prompt words, without using all information processing agents to answer the question, which not only saves computing resources but also improves the accuracy of the answer.

[0034] In an optional embodiment, when the method provided by the present invention is implemented to establish corresponding information processing agents for different departments, a prompt project corresponding to the input to the LLM is designed according to the different attributes of these multiple information processing agents, so that the LLM can more fully understand the attributes and capabilities that information processing agents with different functions and departments should have, so as to better play the role of an information processing agent.

[0035] In an optional embodiment, multiple information processing agents are matched based on prompt words and prompt projects. This process is also implemented by the agent played by LLM (guiding agent). This agent has different functions from other information processing agents. The function of this agent is to select information processing agents representing different departments who need to see a patient based on prompt words and prompt projects, while the role of each information processing agent is to answer the patient's questions.

[0036] In an optional embodiment, the list of information processing agent members is as follows:

[0037] (1) Internal Medicine: Cardiovascular Medicine, Respiratory Medicine, Gastroenterology, Hematology, Nephrology, Rheumatology and Immunology, Neurology, Oncology, 8 agents.

[0038] (2) Surgery: General surgery, cardiothoracic surgery, hepatobiliary surgery, urology, orthopedics, and neurosurgery, 6 agents.

[0039] (3) Department of Obstetrics and Gynecology: 1 agent.

[0040] (4) Ophthalmology and dentistry: 2 agents.

[0041] (5) Others: Emergency Department, Dermatology and Venereology Department, Traditional Chinese Medicine Department, Anesthesiology Department, Nutrition Department, Radiotherapy Department, Nuclear Medicine Department, 7 agents.

[0042] In summary, there are 24 information processing agents in total.

[0043] Step S104: each information processing agent analyzes the basic information and question information to obtain the speech results corresponding to each information processing agent.

[0044] In an optional embodiment, the basic information and question information are input into the corresponding information processing agent, and the information processing agent can obtain the respective collaborative processing results. In addition, in the embodiment of the present invention, each information processing agent speaks in turn, and the speech result of the previous information processing agent will affect the speech result of the next information processing agent, that is, the next information processing agent will speak in combination with the speech result of the previous information processing agent.

[0045] Step S105, the speech results of each information processing agent are summarized through the pre-established consultation agent to obtain the final processing result. The guidance agent, each information processing agent, and the consultation agent are established through a large language model.

[0046] In an optional embodiment, different information processing agents obtain different collaborative processing results. By summarizing the collaborative processing results, the patient's condition and treatment plan can be discussed and improved from various aspects, and the final processing result obtained is more accurate.

[0047] In an optional embodiment, the aggregation of collaborative processing results is also achieved by the agent played by LLM (consultation agent). The consultation agent is different from the agents used in the above steps S103 and S104. The guidance agent, various information processing agents, and the consultation agent are all established through the large language model. However, when establishing different agents, the prompt projects used are different, resulting in different agents having different functions and playing different roles.

[0048] The method provided by the embodiment of the present invention, after obtaining the basic information and question information of the target object, matches multiple information processing agents according to the basic information and question information, and the multiple information processing agents respectively obtain their own collaborative processing results, and summarize the collaborative processing results to obtain the final processing result. In the embodiment of the present invention, the collaborative processing results are obtained by multiple information processing agents. On the one hand, through division of labor, the ability of information processing agents to analyze problems is increasingly refined. On the other hand, multiple debates and discussions with different information processing needs can obtain better results than when there is no specialization, thereby improving the efficiency and output quality of the entire problem processing, effectively reducing the problem of incomplete analysis caused by a single information processing agent, and effectively alleviating the hallucination problem that is prone to occur in a single information processing agent, making the problem analysis results and suggestions more comprehensive and reliable.

[0049] In an optional embodiment, the above step S102 specifically includes:

[0050] Step a1, based on the types of basic information and question information, calling corresponding recognition algorithms to recognize the basic information and question information to obtain recognition results; the recognition algorithms include: at least one of text recognition algorithms, speech recognition algorithms and image recognition algorithms.

[0051] In an optional embodiment, the basic information and question information may be input via text, audio, image, etc. Therefore, when identifying the basic information and question information, at least one of a text recognition algorithm, a speech recognition algorithm, and an image recognition algorithm may be used for identification.

[0052] Step a2: determine the information processing requirements of the target object based on the recognition results.

[0053] Step a3: Generate multiple prompt words based on information processing requirements.

[0054] In an optional embodiment, the guidance agent, each information processing agent, and the consultation agent are generated by inputting different prompt projects into a large language model respectively. Different information processing agents correspond to different prompt projects. The prompt words are compared with the prompt projects corresponding to each information processing agent, and multiple information processing agents are selected based on the matching results.

[0055] In the embodiment of the present invention, because different prompt projects are used, different information processing agents are used to meet different processing needs. It can be seen that the role played by the information processing agent and the capabilities it possesses are closely related to the prompt project. Therefore, the present invention is private, and the prompt words are compared with the prompt projects corresponding to each information processing agent. The higher the similarity between the prompt project and the prompt word, the more suitable the information processing agent corresponding to the prompt project is for processing the basic information and question information corresponding to the prompt word.

[0056] In an optional embodiment, the above step S104 specifically includes:

[0057] Input the basic information and question information into the i-th information processing agent in a preset order, and obtain the speech result of the i-th information processing agent, where i is 1;

[0058] Input the speech result, basic information and question information of the i-th information processing agent into the (i+1)-th information processing agent to obtain the speech result of the (i+1)-th information processing agent;

[0059] If the value of (i+1) is less than n, add 1 to the value of i, and return to the step of inputting the speech result, basic information and question information of the i-th information processing agent into the (i+1)-th information processing agent to obtain the speech result of the (i+1)-th information processing agent, until the value of (i+1) is equal to n, and the speech results corresponding to each information processing agent are obtained.

[0060] That is, in the embodiment of the present invention, the order in which each information processing agent outputs the information processing results is pre-set according to actual needs.

[0061] In an optional embodiment, the above step S104 specifically includes:

[0062] Input the basic information and question information into the i-th information processing agent, and obtain the speech result of the i-th information processing agent, where i is 1;

[0063] The i-th information processing agent randomly selects an unspoken information processing agent from among all the information processing agents as the (i+1)-th agent;

[0064] Input the speech result, basic information and question information of the i-th information processing agent into the (i+1)-th information processing agent to obtain the speech result of the (i+1)-th information processing agent;

[0065] If the value of (i+1) is less than n, add 1 to the value of i and return to the step where the i-th information processing agent randomly selects an information processing agent that has not spoken from among the information processing agents as the (i+1)-th agent until the value of (i+1) is equal to n, and the corresponding speech results of each information processing agent are obtained.

[0066] That is, in the embodiment of the present invention, the order in which each information processing agent outputs information processing results is random. Whenever an information processing agent obtains a speech result, the next information processing agent is randomly designated to output the speech result, and so on, until all the multiple information processing agents determined in the above step S103 output the speech results.

[0067] In an optional embodiment, the above step S104 specifically includes:

[0068] Input the basic information and question information into the i-th information processing agent, and obtain the speech result of the i-th information processing agent, where i is 1;

[0069] The consultation agent compares the speech result of the i-th information processing agent with the prompt engineering of the information processing agent that has not spoken among the information processing agents, and selects the information processing agent with the highest correlation with the speech result of the i-th information processing agent as the (i+1)-th agent;

[0070] Input the speech result, basic information and question information of the i-th information processing agent into the (i+1)-th information processing agent to obtain the speech result of the (i+1)-th information processing agent;

[0071] If the value of (i+1) is less than n, add 1 to the value of i, return to the step of comparing the speech result of the i-th information processing agent with the prompt project of the information processing agent that has not spoken among the information processing agents through the consultation agent, and select the information processing agent with the highest correlation with the speech result of the i-th information processing agent as the (i+1)-th agent, until the value of (i+1) is equal to n, and the corresponding speech results of each information processing agent are obtained.

[0072] In an embodiment of the present invention, each time an information processing agent outputs a collaborative processing result, the next information processing agent with the highest relevance is determined based on the collaborative processing result to perform analysis to obtain a speech result, until multiple information processing agents determined in the above step S103 all output collaborative processing results.

[0073] In the method provided in the above embodiment, the LLM-based information processing agent can make full use of the powerful text generation ability of LLM in the field of medical question and answer. When a single information processing agent faces relevant medical questions raised by users, the answers obtained will appear to be relatively simple in some aspects, and do not cover the professional opinions of other doctors. In contrast, multiple information processing agents can provide the views of multiple doctors from different professional fields, making the comprehensive treatment plan more comprehensive and multi-dimensional, which can better meet the needs of patients and provide more accurate medical advice. In addition, the collaborative treatment of multiple information processing agents also emphasizes the synergy between different treatment options to provide more individualized treatment plans. When answering, it can more comprehensively consider the patient's general condition, surgical feasibility, anesthesia risks, and suggestions on perioperative care and nutrition, making the answers more comprehensive and practical.

[0074] In the above embodiment, agents played by multiple LLM models work together to solve problems. Compared with the multi-agent reinforcement learning (MARL) algorithm commonly used by multi-agents, this method reduces the training process of agents in traditional multi-agent methods. Compared with the Multi-Agent system trained by the traditional MARL algorithm, the agent generated by LLM greatly saves training time and training costs. Agents played by LLM have unique advantages in medical knowledge retrieval and summary, disease diagnosis and auxiliary decision-making, drug information and interaction, automatic generation of medical documents, medical research and trend analysis, multimodal support, etc.

[0075] The Multi-agent system based on LLM also has its own unique advantages in the medical field:

[0076] 1. Excellent context window length: The large model has an excellent context window length. When processing long sequence data such as medical records, more context information can be considered at the same time. In the question-answering system task, the large model can increase the context length to more comprehensively understand the context information of the medical records, thereby answering questions more accurately.

[0077] 2. Low-cost multi-disciplinary consultation: The LLM-based Multi-agent system can automatically generate agents with different functional characteristics and expertise in different fields. These agents can participate in the diagnosis and treatment process, work together to discuss, and realize multi-angle and multi-departmental automated intelligent disease analysis. This innovation can solve some problems existing in existing technologies in the medical field, such as low medical efficiency and low diagnostic accuracy.

[0078] 3. Cutting-edge medical research: The LLM-based Multi-agent system can automatically extract useful features and rules by learning a large amount of medical data and information, thereby providing a more accurate and reliable basis for medical decision-making. This feature can further improve medical efficiency and diagnostic accuracy, and can also provide better support for personalized treatment and interdisciplinary cooperation.

[0079] 4. Language understanding ability: LLM can understand and interpret natural language text, so it can process large amounts of medical text data, such as case reports, medical literature, etc., to help doctors make more accurate diagnoses and treatments.

[0080] 5. Knowledge reasoning ability: LLM can reason based on existing medical knowledge and experience, thereby providing doctors with more comprehensive and accurate diagnostic suggestions.

[0081] 6. Automated decision support: The Multi-agent system can simulate the collaboration and decision-making process between doctors and experts, thereby providing doctors with more automated and intelligent decision support.

[0082] 7. Personalized treatment recommendations: By analyzing the patient's medical history, symptoms, and test results, the Multi-agent system can provide personalized treatment recommendations for each patient, thereby improving treatment outcomes and patient satisfaction.

[0083] In a specific application scenario, the method and related technology provided in the above embodiments of the present invention are respectively used to analyze the input information and give the speech result. In the following example, the third part is the output result obtained by using a single large language model in the related technology, and the fourth part is the speech result given by using the method provided in the above embodiments of the present invention.

[0084] 1. Patient symptoms

[0085] Gender: Male. Age: 53 years old. Marital status: Married. Chief complaint: Clubbing for 3 months, and a lung mass was found 20 days ago. Admission: 3 months ago, the patient had clubbing without obvious cause, occasional cough, no sputum, no chest tightness, shortness of breath, and hemoptysis, but did not pay attention to it. Later, he went to Zhuozhou City Hospital for a chest X-ray, which showed a right lung mass. A chest enhanced CT scan (2018-12-6) showed a mass nodule (3.6*2.6cm) in the dorsal segment of the right lower lobe, and peripheral lung cancer may be large with obstructive emphysema, right lower lobe cords, and a cyst in the posterior basal segment of the left lower lobe was considered. Further puncture biopsy of the right lower lobe lesion showed: lung adenocarcinoma, alveolar type. Now he is admitted to our department for further treatment. Since the onset of the disease, the patient's appetite and spirit are good, his bowel movements are normal, and his weight has not changed significantly. Physical examination upon admission: T: 36.5℃; P: 78 times / min; R: 20 times / min; BP: 131 / 84mmHg; H: 165cm; W: 69Kg; S: 1.76m2; ECOG: grade 0; Pain level: grade 0. Past medical history: denied history of hepatitis, tuberculosis, hypertension, coronary heart disease, diabetes, surgery, trauma, blood transfusion, allergy, unknown vaccination history. Personal history: born in Laishui County, Hebei Province, lived there for a long time, no history of epidemic water, no history of contact with epidemic sources. Denied history of prostitution. No sexually transmitted diseases. Occasionally drank, smoked for more than 30 years, 1 pack a day, quit for more than a month. Marital and childbearing history: married at the right age, spouse and children are healthy. Family history: denied family history of genetic diseases and tumors.

[0086] 2. Choose the right agent

[0087] The guidance agent replied:

[0088] Based on the patient's condition, I recommend the following types of doctors to conduct a joint consultation and develop a comprehensive treatment plan:

[0089] 1. Respiratory physician - Consult on lung cancer, assess the condition and treatment plan.

[0090] 2. Thoracic surgeon - assesses whether surgical resection is possible, surgical risks and prognosis.

[0091] 3. Oncologist - develops systematic treatment plans, such as chemotherapy, targeted therapy, etc.

[0092] 4. Radiation oncologist - assesses whether adjuvant radiotherapy is needed based on the patient's condition.

[0093] 5. Nuclear medicine physician - performs PET-CT to evaluate tumor metabolism and metastasis.

[0094] 6. Cardiothoracic surgeon - assesses if there are contraindications to surgery.

[0095] 7. Anesthesiologist - assesses anesthetic risks.

[0096] 8. Nursing Department Doctor-Develop perioperative nursing plan.

[0097] 9. Nutritionist - formulate dietary standards and improve nutritional status.

[0098] Through joint consultation by a multidisciplinary team, the condition is comprehensively assessed from different aspects and an individualized comprehensive treatment plan is developed.

[0099] 3. Single LLM reply result

[0100] First of all, considering the patient's age and physical condition, we need to conduct a comprehensive assessment, including blood routine, liver function, kidney function, electrocardiogram and other tests to understand the patient's overall health status.

[0101] For the patient's lung cancer, we recommend surgical treatment based on the biopsy results showing alveolar lung cancer. Surgery can include lobectomy or segmentectomy to remove as much tumor tissue as possible. After surgery, we can also consider adjuvant radiotherapy or chemotherapy to reduce the risk of postoperative recurrence.

[0102] In addition, according to the results of chest CT, which showed possible obstructive emphysema, we need to pay attention to the patient's lung function. If the patient's lung function is poor, surgery may bring greater risks. In this case, we can consider minimally invasive surgery such as percutaneous lung puncture or bronchoscopic lung puncture to reduce surgical trauma.

[0103] For patients with clubbing symptoms, we can consider using vasodilators, such as digitalis, to relieve symptoms. In addition, we can also recommend pulmonary rehabilitation training to improve lung function and relieve symptoms.

[0104] Finally, we also need to pay close attention to the patient's cardiovascular and systemic conditions. If the patient has risk factors for cardiovascular disease such as hypertension and coronary heart disease, we need to conduct corresponding assessments and treatments.

[0105] Taking all the above into consideration, we have developed an individualized comprehensive treatment plan, including surgery, radiotherapy or chemotherapy, vasodilator drugs, pulmonary rehabilitation training, etc. The purpose of these treatment plans is to remove the tumor as much as possible, relieve symptoms, improve the quality of life, and comprehensively consider the patient's overall health status and treatment risks.

[0106] 4. Multi-agent response results

[0107] Attending Physician: There is currently a patient whose specific condition is as follows:

[0108] Gender: Male. Age: 53 years old. Marital status: Married. Chief complaint: Clubbing for 3 months, and a lung mass was found 20 days ago. Admission: 3 months ago, the patient had clubbing without obvious cause, occasional cough, no sputum, no chest tightness, shortness of breath, and hemoptysis, but did not pay attention to it. Later, he went to Zhuozhou City Hospital for a chest X-ray, which showed a right lung mass. A chest enhanced CT scan (2018-12-6) showed a mass nodule (3.6*2.6cm) in the dorsal segment of the right lower lobe, and peripheral lung cancer may be large with obstructive emphysema, right lower lobe cords, and a cyst in the posterior basal segment of the left lower lobe was considered. Further puncture biopsy of the right lower lobe lesion showed: lung adenocarcinoma, alveolar type. Now he is admitted to our department for further treatment. Since the onset of the disease, the patient's appetite and spirit are good, his bowel movements are normal, and his weight has not changed significantly. Physical examination upon admission: T: 36.5℃; P: 78 times / min; R: 20 times / min; BP: 131 / 84mmHg; H: 165cm; W: 69Kg; S: 1.76m2; ECOG: grade 0; Pain level: grade 0. Past history: denied history of hepatitis, tuberculosis, hypertension, coronary heart disease, diabetes, surgery, trauma, blood transfusion, allergy, unknown vaccination history. Personal history: born in Laishui County, Hebei Province, lived there for a long time, no history of epidemic water, no history of contact with epidemic sources. Denied history of prostitution. No sexually transmitted diseases. Occasionally drank, smoked for more than 30 years, 1 pack a day, quit for more than a month. Marital and childbearing history: married at the right age, spouse and children are healthy. Family history: denied family history of genetic diseases and tumors. Doctors are required to conduct joint consultation and formulate a comprehensive treatment plan.

[0109] Thoracic surgeon: Based on the patient's condition, I think we need to further evaluate the stage and metastasis of his lung cancer. A PET-CT scan can be considered to evaluate the metabolic activity of the lesions and the presence of lymph node metastasis. In addition, the patient's lung function and surgical feasibility need to be evaluated. If the patient's lung function is poor or surgery is not feasible, chemoradiotherapy combined with targeted therapy can be considered. For patients who are feasible for surgery, lobectomy or segmentectomy can be considered. At the same time, the patient's systemic condition, including heart, liver and kidney function, needs to be evaluated to determine the feasibility and safety of the treatment plan. It is very important to formulate an individualized treatment plan by comprehensively considering the patient's condition and treatment plan.

[0110] Medical oncologist: Based on the patient's condition, I think we can consider surgical resection. First, we need to further evaluate the stage and metastasis of the lung cancer. We can perform a PET-CT scan to evaluate the metabolic activity of the lesion and the presence of lymph node metastasis. This will help determine the feasibility of surgery and the prognosis.

[0111] In addition, we also need to evaluate the patient's lung function to determine the risks of surgery. If the patient's lung function is poor or surgery is not feasible, we can consider chemoradiotherapy combined with targeted therapy as an alternative. For patients who are feasible for surgery, lobectomy or segmentectomy can be considered.

[0112] In addition to local treatment, we also need to comprehensively evaluate the patient's systemic condition, including heart, liver and kidney function, to determine the feasibility and safety of the treatment plan. It is very important to comprehensively consider the patient's condition and treatment plan and develop an individualized treatment plan.

[0113] It should be noted that the feasibility and prognosis of surgical resection depend not only on the characteristics of the tumor itself, but also on the overall condition of the patient and the experience of the surgical team. Therefore, we need to conduct joint consultations, develop comprehensive treatment plans, and fully communicate and explain with patients and their families.

[0114] Radiation oncologist: According to the patient's pathological classification of alveolar lung cancer, I think we can consider a systematic treatment plan, such as chemotherapy, targeted therapy, etc. For patients who are feasible for surgery, lobectomy or segmentectomy can be considered, combined with postoperative adjuvant chemotherapy to improve the treatment effect. For patients who are not feasible for surgery, chemoradiotherapy combined with targeted therapy can be considered to control tumor progression and alleviate symptoms.

[0115] Chemotherapy is the use of chemical drugs to kill cancer cells or stop them from growing and dividing. For acinar lung cancer, chemotherapy regimens commonly used include regimens containing platinum drugs (such as cisplatin or carboplatin), which may be combined with other drugs (such as paclitaxel or pemetrexed). These drugs can be given intravenously or orally.

[0116] Targeted therapy refers to the use of specific drugs to inhibit the growth and spread of tumor cells by targeting specific molecular targets on the surface of tumor cells. For acinar lung cancer, commonly used targeted therapy drugs include epidermal growth factor receptor (EGFR) inhibitors (such as gefitinib, icotinib, etc.) and tyrosine kinase inhibitors (such as crizotinib). These drugs can be used by oral administration.

[0117] When formulating a systematic treatment plan, we also need to comprehensively consider the patient's overall condition and the feasibility of treatment. Therefore, we need to conduct a joint consultation, including radiation oncologists, thoracic surgeons, radiologists, etc., to jointly formulate an individualized treatment plan and fully communicate and explain it to the patient and his family. At the same time, we also need to closely monitor the patient's treatment effect and adverse reactions, and adjust the treatment plan in a timely manner to improve the treatment effect and the patient's quality of life.

[0118] Nuclear medicine physician: According to the patient's pathological classification of alveolar lung cancer, I think we can consider a systematic treatment plan, such as chemotherapy, targeted therapy, etc. For patients who are feasible for surgery, lobectomy or segmentectomy can be considered, combined with postoperative adjuvant chemotherapy to improve the treatment effect. For patients who are not feasible for surgery, chemoradiotherapy combined with targeted therapy can be considered to control tumor progression and alleviate symptoms.

[0119] Chemotherapy is the use of chemical drugs to kill cancer cells or stop them from growing and dividing. For acinar lung cancer, chemotherapy regimens commonly used include regimens containing platinum drugs (such as cisplatin or carboplatin), which may be combined with other drugs (such as paclitaxel or pemetrexed). These drugs can be given intravenously or orally.

[0120] Targeted therapy refers to the use of specific drugs to inhibit the growth and spread of tumor cells by targeting specific molecular targets on the surface of tumor cells. For acinar lung cancer, commonly used targeted therapy drugs include epidermal growth factor receptor (EGFR) inhibitors (such as gefitinib, icotinib, etc.) and tyrosine kinase inhibitors (such as crizotinib). These drugs can be used by oral administration.

[0121] When formulating a systematic treatment plan, we also need to comprehensively consider the patient's overall condition and the feasibility of treatment. Therefore, we need to conduct a joint consultation, including radiation oncologists, thoracic surgeons, radiologists, etc., to jointly formulate an individualized treatment plan and fully communicate and explain it to the patient and his family. At the same time, we also need to closely monitor the patient's treatment effect and adverse reactions, and adjust the treatment plan in a timely manner to improve the treatment effect and the patient's quality of life.

[0122] Cardiothoracic surgeon: Based on the patient's condition, I think we need to further evaluate his surgical feasibility and surgical risks. First, we need to evaluate the patient's lung function to determine the risks of surgery. Pulmonary function tests can be performed, such as vital capacity, forced expiratory volume in one second, and other indicators. If the patient's lung function is poor, the risk of surgery may increase.

[0123] In addition, we also need to evaluate the patient's overall condition, including heart, liver and kidney function, to determine the feasibility and safety of the operation. Electrocardiogram, cardiac ultrasound, liver and kidney function tests and other related tests can be performed.

[0124] In addition, we need to further evaluate the stage and metastasis of lung cancer, and we can consider PET-CT scanning to evaluate the metabolic activity of the lesion and the presence of lymph node metastasis. This will help determine the feasibility of surgery and prognosis.

[0125] Taking into account the patient's lung function, general condition, and the stage and metastasis of lung cancer, we can more accurately assess the risks of surgery and develop an individualized treatment plan. When developing a treatment plan, we also need to conduct a joint consultation, including radiation oncologists, oncologists, nuclear medicine doctors, etc., to jointly develop the most suitable treatment plan for the patient, and fully communicate and explain it to the patient and his family.

[0126] Anesthesiologist: Based on the patient's condition, I think we need to further evaluate his anesthesia risk. First, we need to evaluate the patient's lung function to determine the anesthesia risk of the surgery. Pulmonary function tests can be performed, such as vital capacity, forced expiratory volume in one second, and other indicators. If the patient's lung function is poor, the risk of anesthesia may increase.

[0127] In addition, we also need to evaluate the patient's heart function to determine the safety of anesthesia. Electrocardiogram, cardiac ultrasound and other related examinations can be performed.

[0128] In addition, we also need to comprehensively consider the patient's overall condition, including liver and kidney function, to determine the feasibility and safety of anesthesia.

[0129] Taking into account the patient's lung function, heart function and overall condition, we can more accurately assess the risk of anesthesia and develop an individualized anesthesia plan. When developing an anesthesia plan, we also need to conduct a joint consultation, including thoracic surgeons, radiation oncologists, nuclear medicine doctors, etc., to jointly develop the most suitable anesthesia plan for the patient, and fully communicate and explain it to the patient and his family.

[0130] Doctor from the Nursing Department: Based on the patient's condition, I think we need to comprehensively consider the patient's lung function, general condition, and the stage and metastasis of lung cancer when formulating a perioperative nursing plan. First, we need to assess the patient's lung function to determine the risk of surgery. Pulmonary function tests can be performed, such as vital capacity, forced expiratory volume in one second, and other indicators. If the patient's lung function is poor, the risk of surgery may increase.

[0131] In addition, we also need to evaluate the patient's overall condition, including heart, liver and kidney function, to determine the feasibility and safety of the operation. Electrocardiogram, cardiac ultrasound, liver and kidney function tests and other related tests can be performed.

[0132] In addition, we need to further evaluate the stage and metastasis of lung cancer, and we can consider PET-CT scanning to evaluate the metabolic activity of the lesion and the presence of lymph node metastasis. This will help determine the feasibility of surgery and prognosis.

[0133] In the perioperative care plan, we need to closely monitor the patient's vital signs, including body temperature, heart rate, respiratory rate, blood pressure, etc., as well as the management of postoperative pain. At the same time, we also need to pay attention to the patient's nutritional support and fluid management to ensure the patient's nutrition and water and electrolyte balance.

[0134] In addition, we also need to prevent and treat postoperative complications, including respiratory tract infection, deep vein thrombosis, postoperative pain, etc. Measures such as early postoperative activities, respiratory rehabilitation training, and preventive antibiotic use can be taken to reduce the incidence of complications.

[0135] During perioperative care, we also need to work closely with other departments, including cardiothoracic surgeons, oncologists, and radiation oncologists, to jointly develop the most suitable perioperative care plan for the patient, and fully communicate and explain it to the patient and his or her family. At the same time, we also need to provide psychological support to the patient and his or her family to help them cope with the stress and anxiety of surgery.

[0136] In short, formulating a perioperative care plan requires comprehensive consideration of the patient's condition and treatment plan, as well as perioperative care needs, in order to provide patients with safe and effective care.

[0137] Nutritionist: According to the patient's condition, I think we need to develop a reasonable diet to improve his nutritional status. For lung cancer patients, a good nutritional status is very important for improving immunity, reducing treatment side effects, and promoting recovery.

[0138] First of all, we should encourage patients to consume enough protein. Protein is an important nutrient for maintaining muscle and immune function. Patients can choose to consume some protein-rich foods, such as chicken, fish, beans, nuts, etc. If the patient has a poor appetite, you can consider increasing the intake of protein supplements.

[0139] Secondly, we should encourage patients to take in enough vitamins and minerals. Vitamins and minerals are very important for maintaining normal body functions and the health of the immune system. Patients can choose to take in some foods rich in vitamins and minerals, such as fresh fruits, vegetables, whole grains, etc.

[0140] In addition, we should also encourage patients to maintain a proper weight. Lung cancer patients often experience weight loss and malnutrition, which can affect treatment outcomes and quality of life. Patients can choose to eat small meals multiple times to avoid indigestion caused by eating large amounts. At the same time, they can increase their intake of high-energy, high-protein foods, such as milk, eggs, and nuts.

[0141] Finally, we should also pay attention to the patient's dietary safety. Lung cancer patients have low immunity and are easily infected by bacteria and viruses in food. Therefore, patients should avoid eating raw or undercooked food and try to choose fresh and well-cooked food. In addition, patients should also pay attention to personal hygiene, wash their hands frequently, and keep eating utensils clean.

[0142] The above is my suggestion on the dietary norms for improving the nutritional status of lung cancer patients. I hope it can help the patients recover.

[0143] Attending Physician: Based on the patient's condition and the advice of various doctors, I think we can develop the following comprehensive treatment plan:

[0144] 1. For patients who are surgically feasible, lobectomy or segmentectomy can be considered, combined with postoperative adjuvant chemotherapy to improve the treatment effect. Surgical resection can effectively control the local progression of the tumor and provide histological diagnosis. Postoperative adjuvant chemotherapy can eliminate residual tumor cells and reduce the risk of recurrence.

[0145] 2. For patients where surgery is not feasible, combined chemotherapy and radiotherapy with targeted therapy can be considered to control tumor progression and relieve symptoms. Radiotherapy can kill cancer cells by destroying the DNA structure of tumor cells, while chemotherapy can kill cancer cells or prevent their growth and division by using chemical drugs. Targeted therapy can inhibit the growth and spread of tumor cells by targeting specific molecular targets on the surface of tumor cells and using specific drugs.

[0146] 3. When formulating a treatment plan, we also need to comprehensively consider the patient's overall condition and the feasibility of treatment. Therefore, we need to conduct a joint consultation, including radiation oncologists, thoracic surgeons, radiologists, etc., to jointly formulate an individualized treatment plan and fully communicate and explain it to the patient and his family.

[0147] 4. During the treatment process, we also need to closely monitor the patient's treatment effects and adverse reactions, and adjust the treatment plan in a timely manner to improve the treatment effect and the patient's quality of life. At the same time, we also need to provide psychological support to patients and their families to help them cope with the stress and anxiety of treatment.

[0148] The reason for this comprehensive treatment plan is based on the patient's pathological classification of alveolar lung cancer, which is a common type of lung cancer. For patients who are feasible for surgery, surgical resection can effectively control the local progression of the tumor and provide a histological diagnosis. Postoperative adjuvant chemotherapy can eliminate residual tumor cells and reduce the risk of recurrence. For patients who are not feasible for surgery, chemoradiotherapy combined with targeted therapy can control tumor progression and relieve symptoms. Radiotherapy kills cancer cells by destroying the DNA structure of tumor cells, while chemotherapy kills cancer cells or prevents their growth and division by using chemical drugs. Targeted therapy inhibits the growth and spread of tumor cells by targeting specific molecular targets on the surface of tumor cells and using specific drugs. Taking into account the patient's systemic condition and the feasibility of treatment, we need to conduct a joint consultation, jointly develop an individualized treatment plan, and fully communicate and explain with the patient and his family. At the same time, we also need to closely monitor the patient's treatment effect and adverse reactions, and adjust the treatment plan in time to improve the effect of treatment and the patient's quality of life.

[0149] In this embodiment, a multi-role collaborative information processing device based on a large language model is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0150] This embodiment provides an information processing device based on multi-role collaboration of a large language model, such as Figure 3 As shown, including:

[0151] The information acquisition module 301 is used to acquire basic information and question information of the target object.

[0152] The data analysis module 302 is used to analyze and process the basic information and question information to obtain corresponding multiple prompt words.

[0153] The information processing agent matching module 303 is used to analyze the prompt words through the pre-established guidance agent and match multiple pre-established information processing agents, where the multiple information processing agents are configured to adapt to different information processing requirements.

[0154] The processing result acquisition module 304 is used to analyze the basic information and question information through each information processing agent to obtain the speech results corresponding to each information processing agent.

[0155] The processing result integration module 305 is used to summarize the speech results of each information processing agent through the pre-established consultation agent to obtain the final processing result. The guidance agent, each information processing agent, and the consultation agent are established through a large language model.

[0156] In some optional implementations, the guidance agent, each information processing agent, and the consultation agent are generated by inputting different prompt projects into a large language model respectively. Different information processing agents correspond to different prompt projects. The data analysis module 302 is specifically used to compare the prompt words with the prompt projects corresponding to each information processing agent, and select multiple information processing agents based on the matching results.

[0157] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0158] The multi-role collaborative information processing device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above-mentioned functions.

[0159] The embodiment of the present invention also provides a computer device having the above Figure 3 An information processing device based on multi-role collaboration of a large language model is shown.

[0160] See also Figure 4 , Figure 4 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 4 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 4 A processor 10 is taken as an example.

[0161] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0162] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0163] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0164] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0165] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 4 The example of connecting through bus is taken in the following.

[0166] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0167] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0168] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A multi-role collaborative information processing method based on a large language model, characterized in that: The method comprises: Obtain basic information and question information of the target object; Analyze and process the basic information and question information to obtain corresponding multiple prompt words; The prompt words are analyzed by a pre-established guidance agent, and a plurality of pre-established information processing agents are matched, wherein different information processing agents are obtained by inputting different prompt projects into a large language model, and each information processing agent is configured to adapt to different information processing requirements; By analyzing the basic information and question information by each of the information processing agents, the speech results corresponding to each of the information processing agents are obtained; The speech results of each of the information processing agents are summarized by a pre-established consultation agent to obtain a final processing result, wherein the guidance agent, each of the information processing agents, and the consultation agent are established through a large language model; The information processing agents analyze the basic information and question information to obtain speech results corresponding to the information processing agents, including: Input the basic information and question information into the i-th information processing agent to obtain the speech result of the i-th information processing agent, where i is 1; The consultation agent compares the speech result of the i-th information processing agent with the prompting project of the information processing agent that has not spoken among the information processing agents, and selects the information processing agent with the highest correlation with the speech result of the i-th information processing agent as the i+1-th agent; Input the speech result, basic information and question information of the i-th information processing agent into the i+1-th information processing agent to obtain the speech result of the i+1-th information processing agent; If the value of i+1 is less than n, add 1 to the value of i, return to the step of comparing the speech result of the i-th information processing agent with the prompt project of the information processing agents that have not spoken among the information processing agents, and select the information processing agent with the highest correlation with the speech result of the i-th information processing agent as the i+1-th agent, until the value of i+1 is equal to n, and obtain the speech results corresponding to each information processing agent.

2. The method according to claim 1, characterized in that: The basic information and question information are analyzed and processed to obtain a plurality of corresponding prompt words, including: Based on the types of the basic information and the question information, calling a corresponding recognition algorithm to recognize the basic information and the question information to obtain a recognition result; the recognition algorithm includes: at least one of a text recognition algorithm, a speech recognition algorithm and an image recognition algorithm; Based on the recognition result, determining the information processing requirements of the target object; Based on the information processing requirement, the plurality of prompt words are generated.

3. The method according to claim 1, characterized in that The guidance agent, each information processing agent, and the consultation agent are generated by inputting different prompt projects into the large language model. Different information processing agents correspond to different prompt projects. The pre-established guidance agent analyzes the prompt words and matches a plurality of pre-established information processing agents, including: The prompt words are compared with the prompt projects corresponding to the information processing agents, and a plurality of information processing agents are selected according to the matching results.

4. An information processing device based on a large language model and multi-role collaboration, characterized in that: The device comprises: Information acquisition module, used to obtain basic information and question information of the target object; A data analysis module, used to analyze and process the basic information and question information to obtain a corresponding plurality of prompt words; An information processing agent matching module is used to analyze the prompt words through a pre-established guidance agent and match multiple pre-established information processing agents, where different information processing agents are obtained by inputting different prompt projects into a large language model, and each information processing agent is configured to adapt to different information processing requirements; A processing result acquisition module, used to analyze the basic information and question information through each of the information processing agents to obtain the speech results corresponding to each of the information processing agents; A processing result integration module is used to summarize the speech results of each information processing agent through a pre-established consultation agent to obtain a final processing result. The guidance agent, each information processing agent, and the consultation agent are established through a large language model; The information processing agents analyze the basic information and question information to obtain speech results corresponding to the information processing agents, including: Input the basic information and question information into the i-th information processing agent to obtain the speech result of the i-th information processing agent, where i is 1; The consultation agent compares the speech result of the i-th information processing agent with the prompting project of the information processing agent that has not spoken among the information processing agents, and selects the information processing agent with the highest correlation with the speech result of the i-th information processing agent as the i+1-th agent; Input the speech result, basic information and question information of the i-th information processing agent into the i+1-th information processing agent to obtain the speech result of the i+1-th information processing agent; If the value of i+1 is less than n, add 1 to the value of i, return to the step of comparing the speech result of the i-th information processing agent with the prompt project of the information processing agents that have not spoken among the information processing agents, and select the information processing agent with the highest correlation with the speech result of the i-th information processing agent as the i+1-th agent, until the value of i+1 is equal to n, and obtain the speech results corresponding to each information processing agent.

5. The device according to claim 4, characterized in that The guidance agent, information processing agents, and consultation agents are generated by inputting different prompt projects into the large language model. Different information processing agents correspond to different prompt projects. The data analysis module is specifically used to compare the prompt word with the prompt project corresponding to each information processing agent, and select multiple information processing agents according to the matching results.

6. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the information processing method based on multi-role collaboration of a large language model according to any one of claims 1 to 3 by executing the computer instructions.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the information processing method based on multi-role collaboration of a large language model according to any one of claims 1 to 3.

Images