Patient service method and device based on large medical model, equipment and storage medium

Abstract

The invention discloses a patient service method and device based on a large medical model, equipment and a storage medium, and the method comprises the steps: obtaining the service request data of a target patient in a target service scene; wherein the target service scene is any one of the following medical service scenes: pre-diagnosis guidance, in-diagnosis guidance, post-diagnosis rehabilitation, health questions and answers, medication guidance, disease prevention and convenience service; and inputting the service request data into the medical large model for analysis, and obtaining medical service suggestions provided by the medical large model for the target patient in the target service scene. According to the scheme, medical service suggestions can be automatically provided for the patient, and the problem that medical service cannot be provided for the patient in time due to shortage of medical staff is avoided.

Description

Technical Field

[0001] The present application relates to the field of medical technology, and in particular to a patient service method and apparatus, device and storage medium based on a medical big model. Background Art

[0002] At present, in the medical field, patients generally consult doctors and nurses to get medical service suggestions, such as what medicine to use, how to use the medicine, how to prevent diseases, etc. However, due to the limited number of medical staff, people usually cannot get medical service suggestions in time. In view of this, how to enable people to get medical service suggestions in time has become an urgent problem to be solved. Summary of the invention

[0003] The main technical problem solved by this application is to provide a patient service method and device, equipment and storage medium based on a medical big model, which can automatically provide medical service suggestions to patients, avoid the tension of medical staff and the inability to provide medical services to patients in a timely manner.

[0004] In order to solve the above-mentioned technical problems, the first aspect of the present application provides a patient service method based on a medical big model, including: obtaining service request data of a target patient in a target service scenario; wherein the target service scenario is any one of the following medical service scenarios: pre-diagnosis guidance, in-diagnosis guidance, post-diagnosis rehabilitation, health Q&A, medication guidance, disease prevention, and convenience services; inputting the service request data into the medical big model for analysis, and obtaining the medical service recommendations provided by the medical big model to the target patient in the target service scenario.

[0005] In order to solve the above-mentioned technical problems, the second aspect of the present application provides a patient service device based on a medical big model, including: an acquisition module and an analysis module, the acquisition module is used to obtain the service request data of the target patient in the target service scenario; wherein the target service scenario is any one of the following medical service scenarios: pre-diagnosis guidance, in-diagnosis guidance, post-diagnosis rehabilitation, health Q&A, medication guidance, disease prevention, and convenient services; the analysis module is used to input the service request data into the medical big model for analysis, and obtain the medical service recommendations provided by the medical big model to the target patient in the target service scenario.

[0006] In order to solve the above technical problems, the third aspect of the present application provides an electronic device, including a memory and a processor coupled to each other, the memory storing program instructions, and the processor being used to execute the program instructions to implement the patient service method based on the medical big model in the above first aspect.

[0007] In order to solve the above technical problems, the fourth aspect of the present application provides a computer-readable storage medium storing program instructions that can be executed by a processor, and the program instructions are used to implement the patient service method based on the medical big model of the above first aspect.

[0008] The above scheme obtains the service request data of the target patient in the target service scenario; the target service scenario is any of the following medical service scenarios: pre-diagnosis guidance, in-diagnosis guidance, post-diagnosis rehabilitation, health Q&A, medication guidance, disease prevention, and convenience services; the service request data is input into the medical big model for analysis, and the medical service recommendations provided by the medical big model for the target patient in the target service scenario are obtained. As long as the service request data is obtained, the medical big model can automatically provide the patient with the corresponding medical service recommendations, which can solve the problem of the shortage of medical staff and the inability to provide medical services to patients in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 It is a flowchart of an embodiment of a patient service method based on a medical big model provided by the present application;

[0010] Figure 2 It is a schematic diagram of a framework of an embodiment of a patient service device based on a medical macro model provided by the present application;

[0011] Figure 3 It is a schematic diagram of a framework of an embodiment of an electronic device provided by the present application;

[0012] Figure 4 It is a schematic diagram of a framework of an embodiment of a computer-readable storage medium provided by the present application. DETAILED DESCRIPTION

[0013] The scheme of the embodiment of the present application is described in detail below in conjunction with the drawings of the specification.

[0014] In the following description, for the purpose of explanation rather than limitation, specific details such as specific system structures, interfaces, and technologies are provided to facilitate a thorough understanding of the present application.

[0015] The terms "system" and "network" are often used interchangeably in this article. The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the fragment " / " in this article generally indicates that the associated objects before and after are in an "or" relationship. In addition, "many" in this article means two or more than two.

[0016] See also Figure 1 , Figure 1This is a flow chart of an embodiment of a patient service method based on a medical big model provided by the present application. Specifically, it may include the following steps:

[0017] Step S11: Obtain service request data of the target patient in the target service scenario.

[0018] In the disclosed embodiment, the target service scenario can be any of the following medical service scenarios: pre-diagnosis guidance, in-diagnosis guidance, post-diagnosis rehabilitation, health Q&A, medication guidance, disease prevention, and convenience services; it can be understood that the above examples are only a few possible examples in the actual application process, and do not limit the specific content of the target service scenario. The target service scenario can be selected by the target patient in the human-computer interaction interface.

[0019] Among them, the pre-diagnosis guidance scene is used to provide pre-diagnosis preparation guidelines, recommend medical hospitals, medical departments, medical doctors and other services for target patients. The in-diagnosis guidance scene is used to provide the target patient with the next consultation operation, and can also provide the target patient with a medical report, and can analyze the medical report to find the cause of the target patient's disease, analyze the severity of the disease, etc.; in one embodiment, when the target patient needs surgery, the in-diagnosis guidance scene can also provide the target patient with at least one of surgical planning information, preoperative contraindication prompts, and postoperative contraindication prompts. The post-diagnosis rehabilitation scene is used to provide rehabilitation program guidance, health education, active exercise rehabilitation plan and medical evaluation, and assist the target patient to reasonably complete the rehabilitation plan. The health question and answer scene is used to provide online consultation functions such as medical knowledge query, online consultation communication, and chronic disease management. The medication guidance scene is used to assist the target patient to understand the mechanism of drug action, strictly follow the doctor's advice on medication, understand the contraindications of drug use, remind and urge the target patient to complete the medication plan, and support pharmaceutical research. The disease prevention scenario is used to provide personalized health examinations, disease prevention plans, targeted popular science education and sports health guidance recommendations, and assist target patients to independently choose vaccine types, physical examination items and disease prevention plans. The convenience service scenario is used to provide convenient service functions such as medication for special groups and fertility guidance for pregnant women.

[0020] Among them, the service request data of each target service scenario may be the same or different. In one embodiment, the target service scenario is pre-diagnosis guidance, and the service request data may include at least one of the target patient's personal information, the target patient's description of the cause of the disease and the condition, the willingness to see a doctor, the target patient's personalized situation, the severity of the disease, the difficulty of treatment, etc. The target patient's personal information may include the target patient's name, age, gender, residential address, etc. The target patient's description of the cause of the disease and the condition may include the target patient's current condition information, past medical history, etc. The willingness to see a doctor may include the hospital, clinic, doctor, etc. that the target patient expects to visit.

[0021] In one embodiment, the target service scenario is guidance during diagnosis, and the service request data may include at least one of the target patient's visit process, visit report, current visit process, target patient's diagnosis and treatment analysis intention, target patient's surgical analysis intention, and current visit process, and the visit process is, for example, registration-report-diagnosis-examination-re-diagnosis-medication. If the target patient needs surgery, the service request data may also include the target patient's surgery-related information, such as surgery planning information, preoperative contraindication reminders, and postoperative contraindication reminders.

[0022] In one embodiment, the target service scenario is post-diagnosis rehabilitation, and the service request data may include at least one of the target patient's rehabilitation willingness information, physical condition information, medical history information, medication information, the target patient's rehabilitation plan, rehabilitation education information, the target patient's personalized situation, the target patient's request information for chronic disease management, and the target patient's evaluation information on at least one stage of medical treatment and rehabilitation.

[0023] In one embodiment, the target service scenario is health Q&A, and the service request data may include at least one of the target patient's medical questions, the target patient's knowledge search requests, the target patient's personal symptoms, the target patient's disease conditions, and the target patient's medical history.

[0024] In one embodiment, the target service scenario is medication guidance, and the service request data may include at least one of the target patient's prescription, the target patient's current condition information, the target patient's physical condition information, medical order and medical record information, the target patient's personal information, and feedback information about adverse drug reactions.

[0025] In one embodiment, the target service scenario is disease prevention, and the service request data may include at least one of the target patient's personal wish information, the target patient's physical condition information, the target patient's medical history and medication history information, the physical examination report, request information about disease prevention, and the target patient's own situation.

[0026] In one embodiment, the target service scenario is a convenience service, and the service request data may include at least one of medical insurance service request information, health record data, personal information of the target patient, fertility information, etc. It can be understood that the above service request data are only several possible examples in the actual application process, and the specific content of the service request data is not limited thereto.

[0027] The above implementation can be performed by the medical big model, that is, the medical big model obtains the service request data of the target patient in the target service scenario. The service request data includes at least one modality data of text, picture, voice, and video. Specifically, the medical big model can provide a human-computer interaction interface, and the user can input text and pictures in the human-computer interaction interface; the medical big model can also be provided with a camera device and a recording device to obtain the voice or video of the target patient.

[0028] S12: Input the service request data into the medical big model for analysis, and obtain the medical service recommendations provided by the medical big model for the target patient in the target service scenario.

[0029] Specifically, in different target service scenarios, different service request data can be obtained, and the service request data can be input into the medical big model, which analyzes the service request data and outputs medical service recommendations for target patients. Among them, the analysis process of the medical big model can be at least one of speech recognition, image recognition, natural speech reasoning, text classification, sentiment analysis, translation, relationship extraction, semantic annotation, entity recognition and other processing. The medical big model is trained in advance and has the capabilities of complex language understanding, medical professional text generation, multimodal recognition and retrieval, evidence-based logical reasoning, multi-round active interaction and ethical safety assurance in the medical and health industry. Please refer to the following for an introduction to the medical big model, which will not be described here for the time being.

[0030] In one embodiment, the target service scenario is pre-diagnosis guidance, and the medical big model can provide medical service recommendations for the target patient based on the service request data in the pre-diagnosis guidance scenario. In a specific embodiment, after the medical big model obtains the target patient's past medical history based on the service request data analysis, the medical big model can output medical service recommendations for past doctor options in response to the target patient's past medical history. Among them, the past doctor options are related to the past medical history. In one embodiment, based on the target patient's past medical history, past doctors who have treated the target patient can be output, or based on the target patient's past medical history, doctors who are good at treating the diseases described in the past medical history can be recommended to the target patient. It can be understood that the past doctors output by the medical big model can be one or more, and are not limited here.

[0031] Among them, in one embodiment, the service request data in the pre-diagnosis guidance scenario may include the target patient's personal information, and the medical big model may search whether the target patient has historical medical information based on the target patient's personal information. If so, the target patient's historical medical information is retrieved, and the historical medical information is analyzed to obtain the target patient's past medical history. The historical medical information may include the target patient's historical medical number, historical medical time, historical medical report, etc. In another specific embodiment, the service request data may directly include the target patient's past medical history. For example, the target patient may input his past medical history through the human-computer interaction interface. In other embodiments, the target patient's past medical history may also be inferred from the service request data by the medical big model based on its own understanding ability, so the way in which the medical big model obtains the target patient's past medical history is not specifically limited. That is to say, the target patient's past medical history may be directly extracted from the service request data by the medical big model, or the medical big model may analyze the service request data, or the medical big model may extract and analyze the service request data.

[0032] In a specific embodiment, the medical big model obtains at least one of the current condition information and the medical willingness information of the target patient based on the service request data analysis. The medical big model responds to at least one of the current condition information and the medical willingness information of the target patient and outputs a medical service suggestion including a recommended medical hospital. In one embodiment, the service request data includes the medical willingness information, and the medical willingness information may include at least one medical hospital. Then, a medical hospital can be directly selected from the at least one medical hospital as the recommended medical hospital. In another embodiment, the service request data includes the current condition information of the target patient, and the current condition information of the target patient is analyzed, and the hospital with the best effect of treating the current condition of the target patient is used as the recommended medical hospital. In other embodiments, the service request data may also include the medical willingness information and the current condition information of the target patient, and the current condition information of the target patient is analyzed to obtain the hospital with the best effect of treating the condition of the target patient, and determine whether the hospital is consistent with the hospital in the medical willingness information. If not, the hospital with the best effect is used as the recommended medical hospital.

[0033] Similarly, the target patient's current condition information and medical treatment willingness information can be directly extracted from the service request data by the medical big model, or can be inferred based on the service request data according to the medical big model's own understanding ability, or can be obtained after the medical big model extracts and analyzes the service request data. For example, the service request data contains the target patient's symptoms, and the medical big model infers the type of disease the target patient suffers from based on the target patient's symptoms. The type of disease the target patient suffers from is, for example, gastroenteritis, and the type of disease the target patient suffers from is used as the target patient's current condition information. For another example, the service request data is the target patient's voice, and the target patient's voice contains the name of the hospital he or she wishes to visit, but the result of the voice recognition does not contain a corresponding hospital, then the hospital in the database that is closest to the voice recognition result is used as the medical treatment willingness information.

[0034] In a specific implementation, after the medical big model obtains the current condition information of the target patient based on the service request data analysis, the medical big model can also output a medical service suggestion including a recommended medical department in response to the current condition information of the target patient. For example, if the medical big model obtains that the target patient has gastroenteritis based on the service request data analysis, the gastroenterology department can be recommended to the target patient.

[0035] In a specific embodiment, after the medical big model obtains the personalized situation of the target patient based on the service request data analysis, it can respond to the personalized situation of the target patient and output a medical service suggestion including a pre-diagnosis preparation guide; wherein the pre-diagnosis preparation guide matches the target patient. In one embodiment, the service request data may include the symptoms of the target patient, and the personalized situation of the target patient may include the examination that the target patient needs to undergo. The medical big model obtains the examination that the target patient needs to undergo based on the symptom analysis of the target patient, and responds to the examination that the target patient needs to undergo, and provides the target patient with a pre-diagnosis preparation guide, which may include consultation materials, pre-diagnosis physiological preparation, and pre-diagnosis psychological counseling. For example, if the medical big model analyzes that the target patient needs to undergo a B-ultrasound examination, it can output a pre-diagnosis preparation guide that allows the user to drink more water. For another example, if the medical big model analyzes that the target patient needs to undergo a blood test, it can output a pre-diagnosis preparation guide that allows the user to fast. In other embodiments, the service request data may also directly include the personalized situation of the target patient. The method for obtaining the personalized situation of the target patient may be the same as the method for obtaining the current condition information of the target patient, which will not be repeated here.

[0036] In a specific embodiment, after the medical big model obtains at least one of the severity of the target patient's condition and the degree of difficulty of treatment based on the analysis of the service request data, it outputs a medical service recommendation containing hierarchical diagnosis and treatment information in response to at least one of the severity of the target patient's condition and the degree of difficulty of treatment. Among them, the hierarchical diagnosis and treatment information may include at least one of first visit to a primary hospital, two-way referral, separate diagnosis and treatment of acute and chronic diseases, and joint diagnosis and treatment of superior and subordinate hospitals. For example, in one embodiment, the medical big model obtains that the condition of the target patient is relatively mild based on the analysis of the service request data, and then it can output a medical service recommendation to go to a primary hospital for diagnosis and treatment first, and if it does not improve within one month, go to a higher-level hospital for diagnosis and treatment. Similarly, the method for obtaining the severity of the target patient's condition and the degree of difficulty of treatment can be the same as the method for obtaining the target patient's current condition information, which will not be repeated here.

[0037] In a specific embodiment, the medical big model obtains at least one of the current condition information and the medical willingness information of the target patient based on the service request data analysis, and the medical big model also inputs the background resume information of each candidate doctor, and in response to the current condition information of the target patient, at least one of the medical willingness information and the background resume information of each candidate doctor, outputs a medical service suggestion including a recommended doctor. For example, in one embodiment, the medical big model stores the background resume information of each candidate doctor. After the medical big model obtains the current condition information of the target patient based on the service request data analysis, the candidate doctor who is good at treating the current condition of the target patient can be selected as the recommended doctor according to the current condition information. In another embodiment, the medical willingness information can be extracted from the service request data, and the medical willingness information can include the doctor who is expected to be treated. Then the recommended doctor output can be directly the doctor who is expected to be treated by the target patient.

[0038] In one embodiment, the target service scenario is in-diagnosis guidance, and the medical big model can provide medical service suggestions for the target patient based on the service request data in the in-diagnosis guidance scenario. In a specific embodiment, the medical big model obtains the current treatment process of the target patient based on the service request data analysis, and can output medical service suggestions containing at least the next treatment operation in response to the current treatment process of the target patient. Specifically, the service request data may include the treatment process of the target patient and the treatment process that has been carried out. The treatment process is, for example, registration-reporting-diagnosis-examination-re-diagnosis-medication. The medical big model can obtain the current treatment process of the target patient based on the target patient's entire treatment process and the treatment process that has been carried out, and then output the next treatment operation according to the current treatment process and the entire treatment process. If the current treatment process is diagnosis, the next treatment operation is examination; or output all treatment operations that have not been carried out, that is, output examination-re-diagnosis-medication.

[0039] In a specific embodiment, in response to the medical big model obtaining the diagnosis and treatment analysis intention of the target patient based on the service request data analysis, a medical service suggestion including at least one of the diagnosis basis, etiology search, condition analysis, similar case analysis, expected hospitalization time, expected cure rate, and expected survival rate is output. In one embodiment, the medical big model can obtain the diagnosis and treatment analysis intention of the target patient based on the diagnosis report, treatment plan, and similar case, and output a medical service suggestion including at least one of the diagnosis basis, etiology search, condition analysis, similar case analysis, expected hospitalization time, expected cure rate, and expected survival rate. In other embodiments, the medical big model can also obtain the diagnosis and treatment analysis intention based on the voice of the target patient. The diagnosis and treatment analysis intention represents that the target patient expresses his willingness to need the medical big model to perform diagnosis and treatment analysis during the interaction process with the medical big model. The diagnosis and treatment analysis intention can be obtained by the medical big model using its own understanding ability to analyze the service request data.

[0040] In a specific embodiment, in response to the medical big model obtaining the surgical analysis intention of the target patient based on the service request data analysis, a medical service suggestion including at least one of surgical planning information, preoperative contraindication prompts, and postoperative contraindication prompts is output. Similarly, the surgical analysis intention represents that the medical big model has analyzed the service request data to find that the target patient has the intention to undergo surgical analysis. For example, the target patient inputs a surgical form, which describes the type of surgery and the time of surgery performed by the target patient.

[0041] In one embodiment, when the target service scenario is post-diagnosis rehabilitation, the medical big model can provide medical service recommendations for the target patient based on the service request data in the post-diagnosis rehabilitation scenario. In a specific embodiment, in response to the medical big model analyzing at least one of the target patient's rehabilitation willingness information, physical condition information, medical history information, and medication information based on the service request data, a medical service recommendation containing a rehabilitation plan is output. Among them, the target patient's rehabilitation willingness information can indicate how long the target patient hopes to recover. The method for obtaining at least one of the target patient's rehabilitation willingness information, physical condition information, medical history information, and medication information can be the same as the method for obtaining the target patient's current condition information, which will not be repeated here.

[0042] In a specific embodiment, in response to the medical big model obtaining the target patient's intention to request guidance on the rehabilitation plan based on the service request data analysis, a medical service suggestion including a rehabilitation guidance prompt is output; wherein the rehabilitation guidance prompt is used to guide the target patient to implement the rehabilitation plan. The intention to request guidance indicates that the target patient expects guidance on the rehabilitation plan. For example, if the service request data includes a rehabilitation plan, a rehabilitation guidance prompt for the rehabilitation plan will be output to guide the target patient on how to implement the rehabilitation plan.

[0043] In a specific implementation, in response to the medical big model obtaining a rehabilitation guidance request from a target patient based on service request data analysis, and the medical big model also inputs at least one of the target patient's diagnosis result and treatment progress, the medical big model outputs a medical service suggestion including rehabilitation education; wherein the rehabilitation education matches the target patient. Rehabilitation education may include dietary guidance, sleep guidance, exercise guidance, psychological guidance, life guidance, etc.

[0044] In a specific embodiment, in response to the evaluation information of the target patient on at least one stage of medical treatment and rehabilitation obtained by the medical big model based on the service request data analysis, a medical service suggestion containing the evaluation feedback of the corresponding stage is output. The evaluation feedback can be a reply to the evaluation information. For example, if the evaluation information points out the problems existing in the medical treatment and rehabilitation process, the evaluation feedback can be improvements made to the relevant problems. Similarly, the method for obtaining the evaluation information of the target patient on at least one stage of medical treatment and rehabilitation can be the same as the method for obtaining the current condition information of the target patient, which will not be repeated here.

[0045] In a specific embodiment, in response to the medical big model obtaining the personalized situation of the target patient based on the service request data analysis, a medical service recommendation including an active rehabilitation plan is output; wherein the active rehabilitation plan is matched with the target patient; the personalized situation of the target patient is the personal needs of the target patient, such as the target patient's eating habits, physical condition, etc.

[0046] In a specific implementation, in response to the target intention of the target patient requesting chronic disease management obtained by the medical big model based on the service request data analysis, a medical service suggestion including at least one of chronic disease screening, online follow-up, health education, intelligent intervention, and daily follow-up is output. For example, in one implementation, the service request data includes "I have eyes that are uncomfortable recently, do I have cataracts?" Since cataracts are chronic eye diseases, the medical big model may believe that the target patient has issued a target intention to request chronic disease management, and will output cataract-related education information.

[0047] In one embodiment, when the target service scenario is a health question and answer, the medical big model can provide medical service recommendations for the target patient based on the service request data in the health question and answer scenario. In a specific embodiment, in response to the medical big model obtaining the medical problem of the target patient based on the analysis of the service request data, a medical service recommendation including a medical answer to the medical problem is output. Specifically, after obtaining the medical problem of the target patient, similar questions similar to the medical problem can be searched in the knowledge base, and the answers to the similar questions can be used as medical answers to the medical problem; or after obtaining the medical problem of the target patient, the medical problem can be sent to an online doctor for answering by an online doctor. The method for obtaining the medical problem of the target patient can be the same as the method for obtaining the current condition information of the target patient, which will not be repeated here.

[0048] In a specific embodiment, in response to the knowledge search request of the target patient obtained by the medical big model based on the service request data analysis, a medical service suggestion including the knowledge search result of the knowledge search request is output. Specifically, the knowledge search request of the target patient may include a keyword, and based on the keyword, information related to the keyword is searched from the database of the medical big model to obtain the knowledge search result.

[0049] In a specific embodiment, in response to obtaining at least one of the target patient's personal symptoms, disease conditions, and medical history based on the service request data analysis by the medical big model, a medical service suggestion including medical prompt information is output; wherein the medical prompt information is used to prompt the target patient to consult the doctor for at least one of the disease diagnosis, treatment plan, guidance advice, medication guide, and surgical plan. When the target patient consults the doctor about the disease according to the medical prompt information, the target patient can communicate with the doctor through text, pictures, voice or video, and obtain disease diagnosis, exercise, diet and other life advice and guidance.

[0050] In a specific implementation, in response to the input of the personal portrait data of the target patient in addition to the service request data by the medical big model, the output includes medical service suggestions provided to the target patient in combination with the personal portrait data. Specifically, knowledge questions and answers, active consultations, health knowledge education, regular follow-up and other functions can be performed based on the personal portrait data.

[0051] In a specific implementation, the medical big model can also provide online family doctor consultation, and provide proactive consultation, diagnosis, treatment, consultation and referral services for common diseases, frequently occurring diseases and chronic diseases.

[0052] In one embodiment, when the target service scenario is medication guidance, the medical big model can provide medical service suggestions for the target patient based on the service request data in the medication guidance scenario. In a specific embodiment, in response to the medical big model analyzing the target patient's medication consultation request based on the service request data and at least one of the target patient's current condition information and physical condition information, a medical service suggestion including a medication prescription is output; for example, if the service request data includes the name of the drug, the effect and precautions of the drug are output; if the service request data includes the target patient's current condition information and physical condition information in addition to the drug name, the medical big model can analyze whether the target patient is currently suitable for the drug, and if not, a suitable drug can be recommended.

[0053] In a specific embodiment, in response to the medical big model obtaining the target patient's medication guidance request and medical order and medical record information based on service request data analysis, a medical service recommendation containing medication guidance information is output; wherein the medication guidance information is consistent with the medical order and medical record information, and the medication guidance information includes at least one of medication time, medication route, medication dosage, and contraindication reminders.

[0054] In a specific implementation, in response to the medical big model analyzing the service request data to determine that the target patient belongs to the key protection group, a medical service suggestion on strengthening medication management is output. The key protection group may be the elderly, children, and pregnant women.

[0055] In a specific embodiment, in response to the feedback information about adverse drug reactions obtained by the medical big model based on the service request data analysis, a medical service suggestion for reducing the probability of adverse reactions is output. The feedback information about adverse drug reactions can be obtained based on adverse drug reaction reports and monitoring data.

[0056] In one embodiment, when the target service scenario is disease prevention, the medical big model can provide medical service suggestions for the target patient based on the service request data in the disease prevention scenario. In a specific embodiment, in response to the medical big model analyzing at least one of the target patient's personal willingness information, physical condition information, and medical history and medication history information based on the service request data, a medical service suggestion including a physical examination plan is output. The medical history and medication history information refers to the target patient's historical illness and medication information, which can be obtained by searching the database through the target patient's personal information, or by analyzing the service request data.

[0057] In a specific implementation, after obtaining a physical examination report and a request for interpretation of the physical examination report based on the service request data analysis, the medical big model can respond to the physical examination report and the request for interpretation of the physical examination report and output a medical service suggestion including at least one of abnormal indicator analysis, medical image interpretation, recommended treatment plan, and health promotion plan. The request for interpretation of the physical examination report can be obtained by the medical big model based on its own understanding ability analysis.

[0058] In a specific implementation, in response to a popular science request about disease prevention obtained by the medical big model based on service request data analysis, a medical service suggestion containing a knowledge guide about disease prevention is output. The knowledge guide about disease prevention may include hand hygiene, proper use of personal protective equipment, respiratory hygiene, cough etiquette, portable diagnostic and treatment equipment and environmental cleaning and disinfection, patient placement, safe injection, medical fabric washing, and medical waste disposal.

[0059] In a specific embodiment, in response to the medical big model obtaining the target patient's own condition and the request for interpretation of the target patient's own condition based on the service request data analysis, a medical service recommendation including at least one of popular science education and exercise guidance for the target patient is output.

[0060] In one embodiment, when the target service scenario is a convenience service, the medical big model can provide medical service suggestions for the target patient based on the service request data in the convenience service scenario. In a specific embodiment, in response to the medical big model obtaining a medical insurance service request based on the service request data analysis, a medical service suggestion including medical insurance services is output. Among them, medical insurance services may include medical payment, electronic medical insurance voucher, online medical insurance information query, off-site medical record settlement, medical insurance relationship handling, etc.

[0061] In a specific implementation, medical service recommendations output by a medical big model based on service request data and health record data are obtained; wherein the health record data is obtained based on the interconnection of hospital data and medical insurance data.

[0062] In a specific embodiment, in response to the medical big model analyzing the service request data to determine that the target patient belongs to a special population, a medical service recommendation is outputted with at least one of the special medicines and shared medicines for the special population indicated. The special population may be the elderly, children, and the disabled, and the output of the special medicines and shared medicines for the special population is based on the National Essential Medicines List (2018 Edition). The shared medicines may refer to medicines that can be used by both children and adults, and may also be medicines that can be used by both children and pregnant women, children and the elderly, or pregnant women and the elderly.

[0063] In a specific embodiment, in response to a fertility guidance request obtained by the medical big model based on service request data analysis, medical service recommendations are output that include at least the following items: a guidance plan for at least one stage before, during, and after pregnancy, and at least one guidance on maternal nutrition, psychology, and common diseases, so as to prevent gestational diabetes, maternal iron deficiency anemia, depression, and other common diseases during pregnancy.

[0064] In the above implementation, in each target service scenario, the medical big model can output multiple medical service suggestions based on the service request data. In response to the medical big model analyzing the target patient's past medical history, current condition information, and medical treatment willingness information based on the service request data, it can output medical service suggestions including past doctor options, medical service suggestions including recommended medical hospitals, and medical service suggestions including recommended medical departments. In other words, the medical big model can output multiple medical service suggestions.

[0065] The above scheme obtains the service request data of the target patient in the target service scenario; the target service scenario is any of the following medical service scenarios: pre-diagnosis guidance, in-diagnosis guidance, post-diagnosis rehabilitation, health Q&A, medication guidance, disease prevention, and convenience services; the service request data is input into the medical big model for analysis, and the medical service recommendations provided by the medical big model for the target patient in the target service scenario are obtained. As long as the service request data is obtained, the medical big model can automatically provide the patient with the corresponding medical service recommendations, which can solve the problem of the shortage of medical staff and the inability to provide medical services to patients in a timely manner.

[0066] In one embodiment, the medical big model provided by the present application can be deployed in the cloud and can be quickly integrated with the business system or hardware device through independent API call model. In order to ensure the data privacy of the medical big model and the adaptability of the medical big model in the intranet / no network environment, the deployment on the private server can be realized through two integration methods: API (application programming interface) and SDK (software development kit); further, the model can be packaged into an SDK adapted to smart hardware, adapted to the operating system and deployed on the local device side, so that the medical big model can adapt to mainstream operating systems such as iOS, Android, Linux, and Windows.

[0067] In one embodiment, the medical big model can use API, SDK, SAAS (software as a service), PAAS (platform as a service), MAAS (model as a service), etc. in the application process. And / or, the medical big model can also configure the permissions of the operator, for example, it can review the data permission management strategy and verify the configuration personnel's operation permission classification, specifically, it can add permissions, delete permissions, edit permissions, view permissions, etc. Further, it can also configure the operator access control strategy to verify the permission operation level of different roles in the system, specifically, the system administrator has all operation permissions, the business administrator has all operation permissions under the corresponding section corresponding to the business needs, and the business members need to apply to the administrator for the allocation of permissions.

[0068] In one embodiment, the medical big model can also provide different types of big models, application cases, technical documents, model call introductions, and model compression and deployment process guidance; among them, different types of big models can include big models for computer vision tasks, natural speech processing big models, multimodal big models, scientific computing big models, etc.; the medical big model can also include model fine-tuning tools, model compression tools, model conversion tools, model call tools and interface instructions. Furthermore, the medical big model can also provide a visual interaction component, which can include a first viewing component, a second viewing component and a debugging component. The first viewing component can be used to view the model fine-tuning iteration performance trajectory, specifically, the iteration gap, communication time, calculation time, communication operator training time statistics, etc.; the second viewing component is used to view the data flow direction, model structure, and calculation graph node properties of the operator in the model network; the debugging component can be used to set monitoring points, monitor training anomalies, and view parameter changes.

[0069] In one embodiment, the medical big model has multiple performances after training, and the multiple performances may include at least one of stability, robustness, fairness, explainability, auditability, and maintainability. Specifically, stability means that when the medical big model is running under preset pressure conditions, resources such as TPS, CPU, memory, and disk can be used normally; in the same business scenario, within a period of time determined according to business characteristics, the deviation ratio between the actual prediction accuracy of the medical big model and the expected prediction accuracy is no more than 1%; unified planning and analysis of various links such as network design, selection, installation, and debugging can isolate key business partitions, strengthen business links and security group policies, etc. Robustness means that after training the medical big model, data disturbance, noise, and outliers will not have a significant impact on the performance of the medical big model. Explainability means that the medical big model can explain and understand the model decisions and prediction results, help users understand how the model makes decisions or predicts results, and increase the reliability of the model. Auditability means that the medical big model stores operation files such as operation manuals and process logs, and the operation and maintenance management of databases, platforms, etc. are recorded in detail. Maintainability means that the medical big model can diagnose and repair faults during use.

[0070] The fault diagnosis of the medical big model specifically includes: fault detection of task-related states and / or fault detection of system-related states. Fault detection of task-related states includes: information related to computational graph variables, collective communication-related indicators, data enhancement-related indicators, computing device-related indicators, task scheduling trajectories, etc.; fault detection of system-related states: inter-chip communication bandwidth, CPU three-level cache-related indicators, SDRAM (synchronous dynamic random access memory) read-write bandwidth-related indicators, system and process memory usage, high-bandwidth memory-related indicators, network card rate, error rate, packet loss rate, PCIe (a high-speed serial computer expansion bus standard) read-write bandwidth-related indicators. It can be understood that when a fault occurs, the medical big model can save necessary platform system information, such as software and hardware parameters; and can support fault analysis and problem location when a fault occurs. Furthermore, the medical big model can also perform a variety of health monitoring methods, such as automatic state collection, log analysis, heartbeat monitoring, etc.

[0071] In one embodiment, the fault repair of the medical big model specifically includes: detecting and repairing application faults, for example, determining whether the medical big model is invalid through concept drift detection, determining reasoning faults through operator overflow, and also through distributed computing fault detection and disaster recovery. When a node managed by the platform fails, the faulty resources are isolated and the medical big model tasks that were running when the failure occurred are automatically rescheduled. Furthermore, the present application also provides for resetting, repairing and automatic isolation of some specific faults, see Table 1.

[0072] Table 1 Fault list for resetting, repairing and automatic isolation of specific faults

[0073]

[0074] In one embodiment, the medical big model can also provide feedback channels and optimization mechanisms. The feedback channels can support telephone, SMS, APP, public account, web page, email, etc. The feedback content can include typical error examples of the medical big model, incompatibility issues during resource customization, and suggestions for the use of the medical big model. The optimization mechanism can optimize service capabilities based on the feedback content and improve the model effect.

[0075] In one embodiment, the medical big model also includes a communication community module, which is mainly used for target patients to conduct knowledge questions and answers, topic discussions, initiate activities, etc. Furthermore, the medical big model can also provide a technical support mechanism, which can optimize the software installation and configuration backup, build the operating environment, introduce the working principle of the medical big model, train the operation process of the medical big model, upgrade the version, etc.

[0076] In one embodiment, technicians associated with the large medical model can provide remote support services, that is, they can communicate with users in real time through telephone, email or other effective means to assist users in solving related problems and system failures; when remote services cannot solve the problem, technicians need to go to the user's site to solve the problem to ensure the normal operation of the large medical model; and provide regular on-site maintenance services to respond to questions raised by customers in a timely manner.

[0077] In one embodiment, the degree of change in the business effect of the medical model is also evaluated. Specifically, under the same business evaluation index system, the following formula is used to calculate the ratio of the improved business effect value to the original value to determine the degree of change in the business effect of the medical model.

[0078]

[0079]

[0080] Among them, S1 and S2 represent the improvement rate (degree of change) of business effect; M after It indicates the business effect value after applying the medical big model under the evaluation of indicator A; M before It indicates the business effect value before applying the medical big model under the evaluation of indicator A. It should be noted that in the formula corresponding to S1, the business evaluation indicator value is directly proportional to the business effect, and in the formula corresponding to S2, the business evaluation indicator value is inversely proportional to the business effect.

[0081] In one embodiment, the business optimization degree of the medical big model is also evaluated. Specifically, the following formula is used to calculate the labor time reduction ratio brought by the medical big model, and the labor replacement rate of the medical big model is determined to determine the business optimization degree of the medical big model.

[0082]

[0083] Where R represents the labor replacement rate; T after It represents the man-hours required for the business after applying the medical big model; T before Indicates the man-hours required for the business before the application of the medical big model.

[0084] In one embodiment, the resource optimization scheduling of the large medical model will also be evaluated; wherein, the resource optimization scheduling of the large medical model includes the richness of the resource optimization strategies of the large medical model during the training phase, and the optimization strategies include but are not limited to machine combing, cluster merging, machine memory expansion, GPU / CPU switching, task classification and resource grouping.

[0085] Specifically, the following formula is used to calculate the actual resource utilization rate within a specified time based on the training log of the medical big model to determine the resource optimization scheduling capability.

[0086]

[0087] Among them, s1 represents the allocated resources; s2 represents the actual usage of resources; and L1 represents the resource utilization rate.

[0088] In one embodiment, the real-time response of the medical big model is also evaluated. Specifically, the following formula is used to determine the real-time response of the medical big model based on the time from the user initiating the request to the medical big model returning the result.

[0089] ES_T=R_T finish -R_T start

[0090] Among them, ES_T represents the response time; R_T finish Indicates the time when the medical big model returns the result (non-streaming) or the time when the medical big model returns the first byte result (streaming); R_T start Indicates the start time of the user's request.

[0091] In one embodiment, the mean time between failures of the medical model is also evaluated. Specifically, the following formula is used to calculate the reliability of the medical model in the application process by the time the medical model can operate normally and the longer the mean time between failures.

[0092]

[0093] Among them, MTTF stands for mean time to failure; S all represents the uptime in the medical large model; S fail Indicates the number of failures of the medical large model.

[0094] In one embodiment, the medical big model can process modality data such as text, image, voice, video, music, code, 3D model, etc.

[0095] In one embodiment, the medical big model can perform image-text retrieval, image-based text question and answer, image-based text description, text-to-image generation, speech synthesis, speech recognition, music video retrieval, and video description generation.

[0096] In one embodiment, the medical big model can perform sequence labeling. The type of sequence labeling is not limited; for example, sequence labeling is named entity recognition, semantic labeling, part-of-speech labeling, word segmentation, etc. In one embodiment, the medical big model can perform text classification, sentiment analysis, etc. In one embodiment, the medical big model can perform sentence pair relationship processing such as natural language processing, question answering, and text semantic similarity. In one embodiment, the medical big model can perform text generation processing such as machine translation, text summarization, and dialogue system. In one embodiment, the medical big model can perform knowledge extraction processing such as relationship extraction, event extraction, and table extraction.

[0097] In one embodiment, the medical model can perform image classification processing such as spam judgment and image violation judgment. In one embodiment, the medical model can perform image recognition processing such as face recognition, character recognition, and vehicle recognition. In one embodiment, the medical model can perform target detection processing such as face detection, pedestrian detection, text detection, and key point detection. In one embodiment, the medical model can perform image recognition processing such as face recognition, character recognition, and vehicle recognition. In one embodiment, the medical model can perform image segmentation processing such as semantic segmentation, instance segmentation, and panorama segmentation. In one embodiment, the medical model can perform image generation processing such as picture generation, video generation, and image editing.

[0098] In one embodiment, the medical large model can perform speech recognition processing such as Chinese speech recognition, multi-language recognition, multi-dialect recognition, multi-language mixed reading recognition, automatic recognition of meeting records, speaker information recognition, and voice wake-up.

[0099] In one embodiment, the medical large model is capable of performing speech generation processing such as Chinese speech generation, streaming speech generation, multi-language generation, multi-timbre generation, music generation, and intelligent dubbing.

[0100] In one embodiment, the language comprehension ability of the medical big model is also evaluated. Specifically, the medical big model identifies the intent of each data in the test data set and the key information contained therein through multiple rounds of task-based dialogues, and calculates the precision, recall and F1 value respectively using the following formulas.

[0101]

[0102] Among them, P A represents the prediction accuracy; A1 represents the number of positive data predicted correctly; A represents the number of data predicted as positive examples.

[0103]

[0104] Among them, R A represents the prediction recall rate; A1 represents the number of positive examples predicted correctly; A 答 Indicates the actual number of positive examples.

[0105]

[0106] Among them, F A Indicates the predicted F1 value; P A Represents the prediction accuracy; R A Represents the prediction recall.

[0107] In other implementations, the accuracy, recall and F1 value of the question-answering dialogue need to be calculated separately based on single-round question-answering and multi-round question-answering, where the single-round question-answering capability mainly determines the direct recall and accuracy as well as the recall and accuracy of the Top3, and the multi-round question-answering capability determines the context association capability based on the single-round capability.

[0108] In one embodiment, the text generation capability of the medical model is also evaluated.

[0109] In a specific implementation, for a reading comprehension task, the EM value of a text generation result is calculated using the following formula.

[0110]

[0111] Among them, C represents the candidate answer generated by the medical big model; R represents the correct answer marked by humans; Count c Indicates the number of answers generated by the medical big model; Count exactmatch(c,r) Indicates the number of sentences where the answers generated by the medical big model completely match the correct answers.

[0112] In a specific implementation, for the summary task, the objective indicator ROUGE-N is calculated using the following formula.

[0113]

[0114] Where N is n-gram, which means the number of bytes in the text content sliding window, and the reference value is 2; Count match (gram n ) represents the number of n-grams in common between the reference summary and the machine-generated summary; Count(gram n ) represents the number of n-grams in the reference summary.

[0115] In a specific implementation, the text generation results of medical professional documents such as doctor's orders and electronic medical records can also be scored according to fluency, accuracy, and interpretability, and the average score of the generated text scores is used as the acceptability score of the generated text.

[0116] In one embodiment, the multi-round active interaction capability of the medical large model is also evaluated.

[0117] In a specific implementation, the following formula is used to calculate the multi-round dialogue task completion rate of the dialogue system.

[0118]

[0119] Among them, P G represents the completion rate of multi-round dialogue tasks; G1 represents the number of multi-round dialogues that actually complete the task; G represents the total number of multi-round dialogue data.

[0120] In a specific implementation, the following formula is used to calculate the accuracy of the following text recognition of the dialogue system, that is, if a certain round of multiple rounds of dialogue is correct, the following text recognition is considered to be correct.

[0121]

[0122] Among them, P D represents the completion rate of multi-round dialogue tasks; G1 represents the number of multi-round dialogues that actually complete the task; D represents the total number of multi-round dialogue data.

[0123] In a specific implementation, the multi-round active interaction capability of the medical large model can also be evaluated based on the average scores of fluency scores, coherence scores, and interestingness scores of the multi-round dialogue tasks.

[0124] In a specific implementation, the multi-round active interaction capability of the medical large model can also be evaluated from three dimensions: the fun, sensitivity, and clarity of the multi-round dialogue tasks.

[0125] See also Figure 2 , Figure 2 It is a framework diagram of an embodiment of a patient service device based on a medical big model provided by the present application.

[0126] The patient service device 20 includes an acquisition module 21 and an analysis module 22. The acquisition module 21 is used to obtain the service request data of the target patient in the target service scenario; wherein the target service scenario is any one of the following medical service scenarios: pre-diagnosis guidance, in-diagnosis guidance, post-diagnosis rehabilitation, health Q&A, medication guidance, disease prevention, and convenience services; the analysis module 22 is used to input the service request data into the medical big model for analysis, and obtain the medical service recommendations provided by the medical big model for the target patient in the target service scenario.

[0127] In one embodiment, when the target service scenario is pre-diagnosis guidance, the analysis module 22 can be used to respond to the medical big model obtaining the target patient's past medical history based on the service request data analysis, and output medical service suggestions including past doctor options; wherein the past doctor options are related to the past medical history; respond to the medical big model obtaining at least one of the current condition information and medical willingness information of the target patient based on the service request data analysis, and output medical service suggestions including recommended medical hospitals; respond to the medical big model obtaining the current condition information of the target patient based on the service request data analysis, and output medical service suggestions including recommended medical departments; The medical big model obtains the personalized situation of the target patient based on the service request data analysis, and outputs medical service recommendations including a pre-diagnosis preparation guide; wherein the pre-diagnosis preparation guide matches the target patient; in response to the medical big model obtaining at least one of the severity of the target patient's condition and the difficulty of treatment based on the service request data analysis, the medical service recommendation including graded diagnosis and treatment information is output; in response to the medical big model obtaining at least one of the current condition information and medical treatment willingness information of the target patient based on the service request data analysis, and the medical big model also inputs the background resume information of each candidate doctor, the medical service recommendation including the recommended doctor is output.

[0128] In one embodiment, when the target service scenario is in-diagnosis guidance, the analysis module 22 can also be used to respond to the current medical treatment process of the target patient obtained by the medical big model based on the analysis of the service request data, and output a medical service suggestion including at least the next medical treatment operation; respond to the diagnosis and treatment analysis intention of the target patient obtained by the medical big model based on the analysis of the service request data, and output a medical service suggestion including at least one of the diagnosis basis, cause finding, condition analysis, similar case analysis, expected hospitalization time, expected cure rate, and expected survival rate; respond to the surgical analysis intention of the target patient obtained by the medical big model based on the analysis of the service request data, and output a medical service suggestion including at least one of the surgical planning information, preoperative contraindication prompts, and postoperative contraindication prompts.

[0129] In one embodiment, when the target service scenario is post-diagnosis rehabilitation, the analysis module 22 can also be used to respond to at least one of the target patient's rehabilitation willingness information, physical condition information, medical history information, and medication information obtained by the medical big model based on the service request data analysis, and output a medical service suggestion including a rehabilitation plan; respond to the target patient's request for guidance on the rehabilitation plan obtained by the medical big model based on the service request data analysis, and output a medical service suggestion including rehabilitation guidance prompts; wherein the rehabilitation guidance prompts are used to guide the target patient to implement the rehabilitation plan; respond to the target patient's rehabilitation guidance request obtained by the medical big model based on the service request data analysis, and the medical big model also inputs the target patient's diagnosis results, treatment process to At least one of the following is output: medical service recommendations including rehabilitation education, wherein the rehabilitation education is matched with the target patient; in response to the evaluation information of the target patient on at least one stage of medical treatment and rehabilitation obtained by the medical big model based on the service request data analysis, medical service recommendations including evaluation feedback of the corresponding stage are output; in response to the personalized situation of the target patient obtained by the medical big model based on the service request data analysis, medical service recommendations including active rehabilitation plans are output; wherein the active rehabilitation plans are matched with the target patient; in response to the target intention of the target patient to request chronic disease management obtained by the medical big model based on the service request data analysis, medical service recommendations including at least one of chronic disease screening, online follow-up, health education, intelligent intervention, and daily follow-up are output.

[0130] In one embodiment, when the target service scenario is a health question and answer, the analysis module 22 can also be used to respond to the medical problem of the target patient obtained by the medical big model based on the service request data analysis, and output a medical service suggestion including a medical answer to the medical problem; respond to the knowledge search request of the target patient obtained by the medical big model based on the service request data analysis, and output a medical service suggestion including the knowledge search result of the knowledge search request; respond to the personal symptoms, disease condition, and medical history of the target patient obtained by the medical big model based on the service request data analysis, and output a medical service suggestion including medical prompt information; wherein the medical prompt information is used to prompt the target patient to consult a doctor for at least one of a disease diagnosis, treatment plan, guidance suggestion, medication guide, and surgical plan; respond to the medical big model inputting the personal portrait data of the target patient in addition to the service request data, and output a medical service suggestion provided to the target patient in combination with the personal portrait data.

[0131] In one embodiment, when the target service scenario is medication guidance, the analysis module 22 can also be used to respond to the target patient's medication consultation request and at least one of the target patient's current condition information and physical condition information obtained by the medical big model based on the service request data analysis, and output a medical service recommendation containing a medication prescription; respond to the target patient's medication guidance request and medical order and medical record information obtained by the medical big model based on the service request data analysis, and output a medical service recommendation containing medication guidance information; wherein the medication guidance information is consistent with the medical order and medical record information, and the medication guidance information includes at least one of medication time, medication route, medication dosage, and contraindication reminders; respond to the target patient belonging to the key protection population obtained by the medical big model based on the service request data analysis, and output a medical service recommendation on strengthening medication management; respond to the feedback information on adverse drug reactions obtained by the medical big model based on the service request data analysis, and output a medical service recommendation for reducing the probability of adverse reactions.

[0132] In one embodiment, when the target service scenario is disease prevention, the analysis module 22 can also be used to respond to at least one of the personal willingness information, physical condition information, and medical history and medication history information of the target patient obtained by the medical big model based on the service request data analysis, and output a medical service suggestion including a physical examination plan; respond to the physical examination report and the request for interpretation of the physical examination report obtained by the medical big model based on the service request data analysis, and output a medical service suggestion including at least one of abnormal indicator analysis, medical image interpretation, recommended treatment plan, and health promotion plan; respond to the popular science request about disease prevention obtained by the medical big model based on the service request data analysis, and output a medical service suggestion including a knowledge guide about disease prevention; respond to the target patient's own situation and the request for interpretation of the own situation obtained by the medical big model based on the service request data analysis, and output a medical service suggestion including at least one of popular science education and exercise guidance for the target patient.

[0133] In one embodiment, when the target service scenario is a convenience service, the analysis module 22 can also be used to respond to a medical insurance service request obtained by the medical big model based on the analysis of the service request data, and output medical service recommendations including medical insurance services; obtain the medical service recommendations output by the medical big model based on the service request data and the health record data; wherein the health record data is obtained based on the interconnection of hospital data and medical insurance data; in response to the medical big model obtaining, based on the analysis of the service request data, that the target patient belongs to a special population, output a medical service recommendation indicating at least one of special medicines and shared medicines for special populations; in response to the medical big model obtaining, based on the analysis of the service request data, a fertility guidance request, output a medical service recommendation including at least the following items: a guidance plan for at least one stage before, during and after pregnancy, and at least one guidance on maternal nutrition, psychology, and common diseases.

[0134] In one embodiment, the service request data includes at least one modality of data including text, image, voice, and video.

[0135] See also Figure 3 , Figure 3 It is a schematic diagram of the framework of an embodiment of an electronic device 30 provided in the present application.

[0136] The electronic device 30 includes a memory 31 and a processor 32. The memory 31 stores program instructions, and the processor 32 is used to execute the program instructions to implement the steps in any of the above-mentioned patient service method embodiments based on the medical big model. Specifically, the electronic device 30 may include but is not limited to: a mobile phone, a tablet computer, a laptop computer, a server, etc., which are not limited here.

[0137] Specifically, the processor 32 is used to control itself and the memory 31 to implement the steps in any of the above-mentioned patient service method embodiments based on the medical big model. The processor 32 can also be called a CPU (Central Processing Unit). The processor 32 may be an integrated circuit chip with signal processing capabilities. The processor 32 can also be a general-purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field-programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. In addition, the processor 32 can be implemented by an integrated circuit chip.

[0138] In the above scheme, the electronic device 30 obtains the service request data of the target patient in the target service scenario; wherein the target service scenario is any one of the following medical service scenarios: pre-diagnosis guidance, in-diagnosis guidance, post-diagnosis rehabilitation, health Q&A, medication guidance, disease prevention, and convenience services; the service request data is input into the medical big model for analysis, and the medical service suggestions provided by the medical big model for the target patient in the target service scenario are obtained. As long as the service request data is obtained, the medical big model can automatically provide the patient with the corresponding medical service suggestions, which can solve the problem that the medical staff is in a tight situation and cannot provide medical services to the patient in a timely manner.

[0139] In one embodiment, the electronic device 30 is deployed based on at least one of the physical security requirements, functional security requirements, and management security requirements; the physical security requirements include fire protection, lightning protection, waterproofing, disaster recovery, authorization, etc., the functional security requirements include device label security, hardware interface security, firmware security, driver security, etc., and the management security requirements include management mechanisms, management personnel, etc. That is, the electronic device 30 has functions such as fire protection, lightning protection, waterproofing, disaster recovery, authorization, etc., and the labels, hardware interfaces, firmware, drivers, etc. of the electronic device 30 are all safe. Furthermore, the electronic device 30 may also include a management mechanism and be provided with corresponding management personnel.

[0140] In one embodiment, the electronic device 30 is used in at least one stage of training and reasoning of a large medical model. In at least one stage of training and reasoning, the electronic device 30 is deployed based on at least one security requirement of network bandwidth, network latency, network packet loss rate, and network jitter.

[0141] In one embodiment, the electronic device 30 may also be provided with at least one of information theft protection and security vulnerability protection; and / or, the electronic device 30 operates in a chip heterogeneous scenario, that is, the chip in the electronic device 30 can operate stably in a heterogeneous scenario, and the heterogeneous scenario may include a scenario in which a central processing unit and a graphics processing unit are combined.

[0142] In the above implementation, the hardware devices in the electronic device 30 are mainly limited. In other implementations, the software devices in the electronic device 30 may also be limited.

[0143] In one embodiment, the electronic device 30 runs multiple software facilities, and the multiple software facilities include any two or more of the following: dependency library, artificial intelligence framework, vector database, large model middleware, interface; wherein the electronic device 30 is provided with protection means for the multiple software facilities, and the protection means include at least one of vulnerability management and security update. Specifically, the electronic device 30 can scan and repair vulnerabilities at intervals of preset time, and update security patches in time, and the preset time is set by the user and is not limited here.

[0144] Furthermore, the electronic device 30 may also be provided with protection measures related to at least one of data collection, data preprocessing, and data usage, and the protection measures include at least one of storage security, privacy protection, process safety, and destruction security; and / or, the electronic device 30 is provided with protection measures related to at least one of model training security, model reasoning security, and model file security.

[0145] In one embodiment, the electronic device 30 may also be provided with protection means related to at least one of service security and content security; wherein service security includes: at least one of mean time between failures, service compliance, and feedback handling mechanism, and content security includes: at least one of content reliability, content compliance, and value alignment. Service security and content security may indicate that both the service and the content meet the requirements. Taking the example that service security includes mean time between failures, service security may indicate that the mean time between failures is less than the preset time between failures.

[0146] See also Figure 4 , Figure 4 It is a schematic diagram of a framework of an embodiment of a computer-readable storage medium 40 provided in the present application.

[0147] The computer-readable storage medium 40 stores program instructions 41 that can be executed by a processor, and the program instructions 41 are used to implement the steps in any of the above-mentioned patient service method embodiments based on the medical macro model.

[0148] In the above scheme, the computer-readable storage medium 40 obtains the service request data of the target patient in the target service scenario; wherein the target service scenario is any one of the following medical service scenarios: pre-diagnosis guidance, in-diagnosis guidance, post-diagnosis rehabilitation, health Q&A, medication guidance, disease prevention, and convenience services; the service request data is input into the medical big model for analysis, and the medical service suggestions provided by the medical big model for the target patient in the target service scenario are obtained. As long as the service request data is obtained, the medical big model can automatically provide the patient with the corresponding medical service suggestions, which can solve the problem that the medical staff is in a tight situation and cannot provide medical services to the patient in a timely manner.

[0149] In some embodiments, the functions or modules included in the device provided by the embodiments of the present disclosure can be used to execute the method described in the above method embodiments. The specific implementation can refer to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.

[0150] The above description of various embodiments tends to emphasize the differences between the various embodiments. The same or similar aspects can be referenced to each other, and for the sake of brevity, they will not be repeated herein.

[0151] In the several embodiments provided in the present application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device implementation described above is only schematic. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0152] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0153] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0154] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of each implementation method of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code.

[0155] If the technical solution of this application involves personal information, the product using the technical solution of this application has clearly informed the personal information processing rules and obtained the individual's voluntary consent before processing the personal information. If the technical solution of this application involves sensitive personal information, the product using the technical solution of this application has obtained the individual's separate consent before processing the sensitive personal information, and at the same time meets the "explicit consent" requirement. For example, on personal information collection devices such as cameras, clear and prominent signs are set to inform that the personal information collection scope has been entered and personal information will be collected. If the individual voluntarily enters the collection scope, it is deemed that he or she agrees to the collection of his or her personal information; or on the device that processes personal information, the personal information processing rules are notified by obvious signs / information, and the individual's authorization is obtained through pop-up information or by asking the individual to upload his or her personal information; among them, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the type of personal information processed.

Claims

1. A patient service method based on a medical big model, characterized in that: include: Obtain service request data of target patients in target service scenarios; wherein the target service scenario is any one of the following medical service scenarios: pre-diagnosis guidance, in-diagnosis guidance, post-diagnosis rehabilitation, health Q&A, medication guidance, disease prevention, and convenience services; The service request data is input into the medical big model for analysis, and the medical service recommendations provided by the medical big model for the target patient in the target service scenario are obtained.

2. The method according to claim 1, characterized in that When the target service scenario is the pre-diagnosis guidance, the method includes at least one of the following: In response to the medical big model obtaining the past medical history of the target patient based on the service request data analysis, outputting a medical service suggestion including past doctor options; wherein the past doctor options are related to the past medical history; In response to the medical big model obtaining at least one of the current condition information and medical treatment willingness information of the target patient based on the service request data analysis, outputting a medical service suggestion including a recommended medical hospital; In response to the medical big model obtaining the current condition information of the target patient based on the service request data analysis, outputting a medical service suggestion including a recommended medical department; In response to the medical big model obtaining the personalized situation of the target patient based on the service request data analysis, outputting a medical service suggestion including a pre-diagnosis preparation guide; wherein the pre-diagnosis preparation guide matches the target patient; In response to the medical big model analyzing the service request data to obtain at least one of the severity of the target patient's condition and the difficulty of treatment, outputting a medical service suggestion including hierarchical diagnosis and treatment information; In response to the medical big model obtaining at least one of the current condition information and medical treatment willingness information of the target patient based on the analysis of the service request data, and the medical big model also inputs the background resume information of each candidate doctor, the output includes medical service recommendations for recommended doctors.

3. The method according to claim 1, characterized in that When the target service scenario is the diagnosis guidance, the method further includes at least one of the following: In response to the medical big model obtaining the current medical consultation process of the target patient based on the service request data analysis, outputting a medical service suggestion including at least the next medical consultation operation; In response to the medical big model obtaining the diagnosis and treatment analysis intention of the target patient based on the service request data analysis, outputting a medical service suggestion including at least one of diagnosis basis, cause search, condition analysis, similar case analysis, expected hospitalization time, expected cure rate, and expected survival rate; In response to the medical big model obtaining the surgical analysis intention of the target patient based on the service request data analysis, a medical service recommendation including at least one of surgical planning information, preoperative contraindication prompts, and postoperative contraindication prompts is output.

4. The method according to claim 1, characterized in that When the target service scenario is the post-diagnosis rehabilitation, the method further includes at least one of the following: In response to the medical big model analyzing at least one of the target patient's rehabilitation willingness information, physical condition information, medical history information, and medication information based on the service request data, outputting a medical service suggestion including a rehabilitation plan; In response to the medical big model obtaining the target patient's request guidance intention for the rehabilitation plan based on the service request data analysis, outputting a medical service suggestion including a rehabilitation guidance prompt; wherein the rehabilitation guidance prompt is used to guide the target patient to implement the rehabilitation plan; In response to the medical big model obtaining a rehabilitation guidance request of the target patient based on the service request data analysis, and the medical big model also inputs at least one of the diagnosis result and treatment progress of the target patient, outputting a medical service suggestion including rehabilitation education; wherein the rehabilitation education matches the target patient; In response to the medical big model analyzing the service request data to obtain the target patient's evaluation information on at least one stage of medical treatment and rehabilitation, output a medical service suggestion including the evaluation feedback corresponding to the stage; In response to the medical big model obtaining the personalized situation of the target patient based on the service request data analysis, outputting a medical service suggestion including an active rehabilitation plan; wherein the active rehabilitation plan matches the target patient; In response to the target intention of the target patient requesting chronic disease management obtained by the medical big model based on the service request data analysis, a medical service recommendation including at least one of chronic disease screening, online follow-up, health education, intelligent intervention, and daily follow-up is output.

5. The method according to claim 1, characterized in that: When the target service scenario is the health question and answer, the method further includes at least one of the following: In response to the medical problem of the target patient obtained by the medical big model based on the service request data analysis, outputting a medical service suggestion including a medical answer to the medical problem; In response to the medical macro model obtaining a knowledge search request of the target patient based on the service request data analysis, outputting a medical service suggestion including knowledge search results of the knowledge search request; In response to the medical big model analyzing at least one of the target patient's personal symptoms, disease conditions, and medical history based on the service request data, outputting a medical service suggestion including medical prompt information; wherein the medical prompt information is used to prompt the target patient to consult a doctor for at least one of a disease diagnosis, a treatment plan, a medical guidance suggestion, a medication guide, and a surgical plan; In response to the medical big model receiving the personal portrait data of the target patient in addition to the service request data, the output includes a medical service suggestion provided for the target patient in combination with the personal portrait data.

6. The method according to claim 1, characterized in that When the target service scenario is the medication guidance, the method further includes at least one of the following: In response to the medical big model obtaining the medication consultation request of the target patient and at least one of the current condition information and physical condition information of the target patient based on the service request data analysis, outputting a medical service suggestion including a medication prescription; In response to the medical big model obtaining the medication guidance request and medical order and medical record information of the target patient based on the service request data analysis, a medical service suggestion including medication guidance information is output; wherein the medication guidance information is consistent with the medical order and medical record information, and the medication guidance information includes at least one of medication time, medication route, medication dosage, and contraindication reminder; In response to the medical big model analyzing the service request data to determine that the target patient belongs to a key protection group, outputting a medical service suggestion on strengthening medication management; In response to the feedback information about adverse drug reactions obtained by the medical big model based on the service request data analysis, a medical service suggestion for reducing the probability of adverse reactions is output.

7. The method according to claim 1, characterized in that When the target service scenario is disease prevention, the method further includes at least one of the following: In response to the medical big model analyzing at least one of the personal willingness information, physical condition information, and medical history and medication history information of the target patient based on the service request data, outputting a medical service suggestion including a physical examination plan; In response to the medical big model obtaining a physical examination report based on the service request data analysis and a request for interpretation of the physical examination report, outputting a medical service suggestion including at least one of abnormal indicator analysis, medical image interpretation, recommended treatment plan, and health promotion plan; In response to the medical big model obtaining a popular science request about disease prevention based on the service request data analysis, outputting a medical service suggestion including a knowledge guide about disease prevention; In response to the medical big model obtaining the target patient's own condition and the request for interpretation of the own condition based on the service request data analysis, a medical service recommendation including at least one of popular science education and exercise guidance for the target patient is output.

8. The method according to claim 1, characterized in that When the target service scenario is the convenient service, the method further includes at least one of the following: In response to the medical care big model obtaining a medical care service request based on the service request data analysis, outputting a medical care service suggestion including the medical care service; Obtaining medical service recommendations output by the medical big model based on the service request data and health record data; wherein the health record data is obtained based on the interconnection of hospital data and medical insurance data; In response to the medical big model analyzing the service request data to determine that the target patient belongs to a special population, outputting a medical service suggestion with at least one of a special drug and a shared drug for the special population noted; In response to the fertility guidance request obtained by the medical big model based on the analysis of the service request data, medical service recommendations are output that include at least the following items: a guidance plan for at least one stage before, during and after pregnancy, and at least one guidance on nutrition, psychology and common diseases of pregnant women.

9. The method according to any one of claims 1 to 8, characterized in that: The service request data includes at least one modality of data including text, picture, voice, and video.

10. A patient service device based on a medical big model, characterized in that: include: An acquisition module is used to acquire service request data of a target patient in a target service scenario; wherein the target service scenario is any one of the following medical service scenarios: pre-diagnosis guidance, in-diagnosis guidance, post-diagnosis rehabilitation, health Q&A, medication guidance, disease prevention, and convenience services; The analysis module is used to input the service request data into the medical big model for analysis, and obtain the medical service recommendations provided by the medical big model for the target patient in the target service scenario.

11. An electronic device, characterized in that: The method comprises a memory and a processor coupled to each other, wherein the memory stores program instructions, and the processor is used to execute the program instructions to implement the method according to any one of claims 1 to 9.

12. The device according to claim 11, characterized in that The electronic equipment is deployed based on at least one security requirement among physical security requirements, functional security requirements, and management security requirements; And / or, the electronic device is used in at least one stage of training and reasoning of a large medical model, and in at least one stage of the training and reasoning, the electronic device is deployed based on at least one guarantee requirement of network bandwidth, network delay, network packet loss rate, and network jitter; And / or, the electronic device is provided with at least one of information theft protection and security vulnerability protection; And / or, the electronic device operates in a chip heterogeneous scenario.

13. The device according to claim 11, characterized in that The electronic device runs a variety of software facilities, and the various software facilities include any two or more of the following: dependency libraries, artificial intelligence frameworks, vector databases, large model middleware, and interfaces; The electronic device is provided with protection means for the multiple software facilities, and the protection means include at least one of vulnerability management and security update.

14. The device according to claim 11, characterized in that The electronic device is provided with protection means related to at least one of data collection, data preprocessing, and data use, and the protection means includes at least one of storage security, privacy protection, process security, and destruction security; and / or, The electronic device is provided with protection measures related to at least one of model training security, model reasoning security, and model file security.

15. The device according to claim 11, characterized in that The electronic device is provided with protection means related to at least one of service security and content security; Among them, the service security includes: at least one of mean time between failures, service compliance, and feedback handling mechanism; the content security includes: at least one of content reliability, content compliance, and value alignment.

16. A computer-readable storage medium, characterized in that: Program instructions that can be executed by a processor are stored, and the program instructions are used to implement the method according to any one of claims 1 to 9.

Images