A diagnosis and treatment simulation method, system, medium and program product based on digital human

Abstract

A diagnosis and treatment simulation method, system, medium and program product based on digital human, which relates to the field of educational or demonstration tools, and includes: calling the current digital human from the digital human waiting queue, outputting symptom description and physical sign image; receiving the target user's inquiry instruction, and outputting the symptom feedback of the corresponding pathological state based on the clinical conversation model; in response to the inspection instruction, marking the current digital human as a waiting state, and receiving the inspection priority set by the target user; determining the inspection waiting time of the current digital human, and generating the inspection result of the current digital human; obtaining the target user's diagnosis result and disposal plan, and determining the diagnosis and treatment effect score; based on the inspection priority and diagnosis and treatment effect score of all digital humans in the digital human waiting queue, determining the diagnosis and treatment score of the target user and updating the physician ability rating of the target user. The implementation of this application can simulate the diagnosis and treatment scenarios of multiple cases and provide more learning paths.

Description

Technical Field

[0001] The present application relates to the field of educational or demonstration tools, and in particular to a diagnosis and treatment simulation method, system, medium and program product based on digital human. Background Art

[0002] With the rapid growth of demand for medical education and clinical training, the traditional medical education model faces huge challenges. Medical knowledge is updated and iterated faster, and clinical cases are becoming increasingly complex and diverse. Medical students and medical staff need more efficient and intelligent training methods to improve their diagnosis and treatment level.

[0003] The relevant technology simulates diagnosis and treatment through digital humans. This technology constructs virtual patients through deep learning models to simulate the changes in patients' physiological parameters, the progression of their illness, and the response to diagnosis and treatment. The simulation system uses multimodal data processing methods to integrate patients' vital signs, medical images, test reports and other information, and uses natural language processing technology to achieve doctor-patient dialogue interaction. The system generates dynamic responses from patients based on the preset case database and disease progression model, providing trainees with a close-to-real diagnosis and treatment training environment.

[0004] The relevant technologies mainly provide training for single case scenarios. However, in the actual medical environment, mature physicians often need to manage the examination and treatment processes of multiple patients at the same time. The examination results of different patients are often returned at different time points, and the urgency of the results varies. The relevant technologies are difficult to simulate such staggered work scenarios, and it is difficult for trainees to accumulate resource allocation experience and time management capabilities in complex scenarios. Summary of the invention

[0005] The present application provides a diagnosis and treatment simulation method, system, medium and program product based on digital humans, which are used to simulate diagnosis and treatment scenarios of multiple cases and provide more learning paths.

[0006] In the first aspect, the present application provides a diagnosis and treatment simulation method based on digital humans, which is applied to a diagnosis and treatment simulation system, and the method includes: retrieving the current digital human from the digital human waiting queue according to the order of diagnosis, and outputting symptom descriptions and physical sign images according to the pathological state of the current digital human; receiving the target user's inquiry instruction, and outputting symptom feedback of the corresponding pathological state based on the clinical conversation model; in response to the target user's examination instruction, marking the current digital human as a waiting state, and receiving the examination priority set by the target user; while the target user continues to diagnose and treat other digital humans, determining the examination waiting time of the current digital human according to the preset examination time and examination priority, and generating the examination result of the current digital human after the examination waiting time; obtaining the target user's diagnosis result and treatment plan, and determining the diagnosis and treatment effect score of the current digital human according to the diagnosis result and treatment plan; determining the diagnosis and treatment score of the target user based on the examination priority and diagnosis and treatment effect score of all digital humans in the digital human waiting queue, and updating the target user's physician ability rating according to the diagnosis and treatment score.

[0007] In the above embodiment, the diagnosis and treatment simulation system simulates the scenario of managing multiple patients at the same time in a real clinical environment, allowing users to reasonably allocate time and resources during the diagnosis and treatment process, and evaluate the physician's ability based on the overall diagnosis and treatment effect, so that trainees can accumulate clinical experience in complex scenarios in an environment close to reality and improve the ability to process multiple cases in parallel.

[0008] In combination with some embodiments of the first aspect, in some embodiments, before the step of retrieving the current visiting digital human from the digital human waiting queue according to the order of consultation, and outputting symptom descriptions and physical sign images according to the pathological state of the current visiting digital human, the method also includes: determining the physician ability rating of the target user according to the user login information, and determining the case difficulty index and the number of digital humans for diagnosis and treatment simulation according to the physician ability rating; selecting multiple case data from a preset case library according to the case difficulty index, and determining multiple pathological states based on the multiple case data; generating a waiting queue of digital humans corresponding to the number of digital humans and in multiple pathological states.

[0009] In the above embodiment, the diagnosis and treatment simulation system dynamically adjusts the difficulty and number of cases according to the user's physician ability rating to ensure that the training content matches the level of the trainees, avoids poor learning effects due to excessively high or low difficulty, and realizes personalized step-by-step training.

[0010] In combination with some embodiments of the first aspect, in some embodiments, before the step of generating a waiting queue of digital humans corresponding to the number of digital humans and in multiple pathological states, the method also includes: collecting basic training data including photos, videos and three-dimensional scanning data; extracting facial feature points, expression feature mapping and clothing features in the basic training data, performing three-dimensional reconstruction on the digital human model to obtain an initial digital human; inputting different pathological states into the physiological parameter model, adjusting the complexion, posture and expression features of the initial digital human, and obtaining multiple different pathological digital humans.

[0011] In the above embodiment, the diagnosis and treatment simulation system collects multimodal training data and performs feature extraction and three-dimensional reconstruction, and adjusts the digital human's complexion, posture and other expressions in combination with the physiological parameter model, so that the appearance and symptoms of the virtual patient are more realistic and natural, thereby improving the user experience of the simulation training.

[0012] In combination with some embodiments of the first aspect, in some embodiments, the steps of receiving the target user's inquiry instructions and outputting symptom feedback corresponding to the pathological state based on the clinical conversation model specifically include: obtaining a set of inquiry instructions input by the target user; the inquiry instruction set includes chief complaint inquiry instructions, medical history inquiry instructions and physical examination instructions; based on the clinical conversation model and the pathological state database of the current digital person being treated, outputting symptom feedback corresponding to the inquiry instruction set; calculating the standardization score of the inquiry instruction set based on a preset standard inquiry process; and generating inquiry guidance prompts for the target user based on the standardization score.

[0013] In the above embodiment, the diagnosis and treatment simulation system helps users master standardized diagnosis methods through standardized diagnosis process scoring and guidance prompts, and at the same time realizes natural doctor-patient dialogue interaction based on the clinical conversation model, which can improve users' diagnosis skills and communication abilities.

[0014] In combination with some embodiments of the first aspect, in some embodiments, the steps of obtaining the diagnosis results and treatment plans of the target user, and determining the diagnosis and treatment effect score of the digital person currently visiting the hospital based on the diagnosis results and treatment plans, specifically include: obtaining the diagnosis results and treatment plans input by the target user, and determining the logical association score of the diagnosis results and the treatment plans; performing similarity matching on the diagnosis results with the standard diagnosis results of the preset pathological state to obtain a diagnosis score; determining the treatment score of the treatment plan based on multiple treatment standards based on drug information data and clinical knowledge graphs; the treatment standards include indications, dosage, course of treatment and contraindications for medication; and determining the diagnosis and treatment effect score of the digital person currently visiting the hospital based on the logical association score, diagnosis score and treatment score.

[0015] In the above embodiment, the diagnosis and treatment simulation system uses a multi-dimensional scoring system to conduct a comprehensive assessment from aspects such as the accuracy of the diagnosis results, the rationality of the treatment plan, and the logical correlation between the two, which can objectively reflect the user's clinical decision-making level and provide precise guidance for capacity improvement.

[0016] In combination with some embodiments of the first aspect, in some embodiments, after determining the diagnosis and treatment score of the target user based on the examination priority and diagnosis and treatment effect score of all digital humans in the digital human waiting queue, and updating the physician ability rating of the target user according to the diagnosis and treatment score, the method also includes: constructing a diagnosis and treatment ability development prediction model according to the changing trend of the diagnosis and treatment score of the target user; generating learning suggestions and learning resource links for different types of cases according to the diagnosis and treatment ability development prediction model; and displaying a learning plan including learning suggestions, learning resource links and recommended simulation time.

[0017] In the above embodiment, the diagnosis and treatment simulation system builds a personalized ability development prediction model by analyzing the changing trend of the user's diagnosis and treatment scores, and generates targeted learning suggestions and resource recommendations to help users improve their weak links more effectively.

[0018] In combination with some embodiments of the first aspect, in some embodiments, before receiving the target user's consultation instruction and outputting symptom feedback corresponding to the pathological state based on the clinical conversation model, the method also includes: obtaining the system load rate of the diagnosis and treatment simulation system; when the system load rate is higher than a preset load threshold, adjusting the clinical conversation model to a lightweight model.

[0019] In the above embodiment, the diagnosis and treatment simulation system monitors the system load and switches the lightweight model in time to ensure that the system can still run smoothly when there are many concurrent users and provide a stable training experience.

[0020] In the second aspect, an embodiment of the present application provides a diagnosis and treatment simulation system, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the diagnosis and treatment simulation system to perform the method described in the first aspect and any possible implementation method of the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer program product comprising instructions, which, when the computer program product is run on a diagnosis and treatment simulation system, enables the diagnosis and treatment simulation system to perform the method described in the first aspect and any possible implementation method of the first aspect.

[0022] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, comprising instructions, which, when executed on a diagnosis and treatment simulation system, enables the diagnosis and treatment simulation system to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0023] It is understandable that the diagnosis and treatment simulation system provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the method provided in the embodiment of the present application. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method, which will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. Due to the adoption of digital human waiting queue management and inspection priority setting mechanism, users can handle multiple cases at the same time and determine the processing priority. Therefore, it can truly simulate the working scenario in the clinical environment where doctors need to pay attention to multiple patients at the same time and reasonably allocate medical priorities. It effectively solves the problem that single case training in related technologies cannot cultivate the ability to cope with complex scenarios, thereby improving the target users' multi-tasking ability and time management ability in a high-pressure working environment, so that they can better cope with various challenges in actual clinical work.

[0025] 2. Due to the use of standardized consultation process scoring and intelligent dialogue interaction based on clinical conversation models, the system can evaluate the user's consultation standardization in real time and provide targeted guidance, so it can help users master scientific consultation methods and establish correct clinical thinking patterns. It effectively solves the problem of slow improvement in consultation skills due to the lack of effective consultation guidance and instant feedback mechanism in related technologies, and then helps users quickly master standardized consultation skills, which helps to improve the effectiveness of doctor-patient communication.

[0026] 3. Due to the adoption of a diagnosis and treatment capability development prediction model and a personalized learning suggestion generation mechanism, the system can intelligently recommend suitable learning content and practice directions based on the user's learning curve, thereby providing users with a more targeted capability improvement path, effectively solving the problem of low training efficiency caused by the lack of personalized learning guidance in related technologies, and thus helping users find their own shortcomings more quickly and improve them effectively, thereby improving learning efficiency and training effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a flow chart of a diagnosis and treatment simulation method based on a digital human in an embodiment of the present application; Figure 2 is another flow chart of the diagnosis and treatment simulation method based on digital human in the embodiment of the present application; Figure 3It is a schematic diagram of the structure of a physical device of the diagnosis and treatment simulation system in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification of the present application, the singular expressions "one", "a kind of", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations comprising one or more of the listed items.

[0029] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0030] For ease of understanding, the application scenarios of the embodiments of the present application are introduced below.

[0031] With the expansion of medical education and the growth of clinical training demand, medical schools and teaching hospitals are facing tremendous teaching pressure. On the one hand, each supervising doctor needs to teach multiple interns and residents at the same time, making it difficult to provide adequate clinical guidance to each trainee; on the other hand, due to the limited number of actual patients and the uneven distribution of diseases, it is difficult for trainees to systematically and completely contact various typical cases. Especially for some rare diseases and critical and severe diseases, trainees often lack practical opportunities. In addition, in a real clinical environment, doctors need to manage the diagnosis and treatment processes of multiple patients at the same time, and the examination results of different patients are returned at different time points. The work rhythm and resource allocation capabilities in this complex scenario require long-term accumulation to master. These problems restrict the training effect of medical students' clinical skills.

[0032] In the related art, clinical skills training for medical students can be achieved by adopting a single case training, instant feedback of examination results and fixed scoring standards. The following introduces the scenario of using the diagnosis and treatment simulation method based on digital human in the related art.

[0033] In the existing diagnosis and treatment simulation teaching, the main method used is to train one case at a time. For example, the diagnosis and treatment simulation system used by a tertiary hospital can only allow trainees to treat a typical virtual patient at a time. The system will return the examination results immediately, and there is no waiting time for examinations in the real environment. Only after a case is completed can the next case training begin. Although this method can help trainees master the basic diagnosis and treatment process, it cannot simulate the scenario in which doctors need to pay attention to multiple patients and coordinate various medical resources at the same time in clinical work. Trainees lack adaptation training for complex working environments, which leads to difficulties in effectively managing the diagnosis and treatment process of multiple patients after entering actual work, and is prone to work confusion and inefficiency.

[0034] The diagnosis and treatment simulation method based on digital humans in the embodiment of the present application is adopted to achieve multi-case parallel training by establishing a digital human waiting queue management mechanism, an inspection priority scheduling system and a multi-dimensional evaluation system, which not only allows trainees to receive the inspection results of other cases while processing one case, but also allows trainees to dynamically adjust the diagnosis and treatment priority according to the severity of the disease and the status of medical resources. The following introduces the scenario of using the diagnosis and treatment simulation method based on digital humans in the present application.

[0035] After adopting the diagnosis and treatment simulation system of the present application, interns at a medical school can manage multiple virtual patients at the same time in a simulation environment. The system will dynamically assign 3-5 virtual patients with different diseases according to the ability level of the interns. Interns need to reasonably set priorities for examinations of different patients. The system will simulate the real waiting time according to the type and priority of the examination, and return the examination results one after another while the interns are dealing with other patients. This training model allows interns to gradually adapt to the multi-threaded working mode in a low-pressure environment, learn how to balance various diagnosis and treatment tasks, and improve resource allocation capabilities. The system will also score the overall performance of interns to help them gradually improve their overall capabilities.

[0036] It can be seen that the use of the digital human-based diagnosis and treatment simulation method in the embodiment of the present application can not only realize basic clinical skills training, but also effectively solve the problem of lack of complex scenario training in related technologies, thereby achieving a comprehensive improvement in the multi-threaded work ability and resource allocation ability of medical students.

[0037] For ease of understanding, the following describes the process of the method provided by this implementation in combination with the above scenario. Figure 1 , which is a flow chart of a diagnosis and treatment simulation method based on a digital human in an embodiment of the present application.

[0038] S101. Retrieve the current digital human from the digital human waiting queue according to the order of consultation, and output symptom description and physical sign image according to the pathological state of the current digital human.

[0039] Among them, the digital human waiting queue represents the queue structure formed by the virtual patients to be diagnosed in the order of their visits; the digital human currently visiting represents the virtual patient who is receiving diagnosis and treatment; the pathological state refers to the current disease condition of the virtual patient, including the main symptoms, physical signs and the stage of disease development; the symptom description is used to represent the patient's subjective feeling of discomfort, which is taken from some typical case descriptions in the hospital; the physical sign image refers to the medical imaging data that can intuitively reflect the pathological state.

[0040] When starting the simulated diagnosis and treatment training, the diagnosis and treatment simulation system needs to first determine the virtual patient to be diagnosed. Specifically, the diagnosis and treatment simulation system first retrieves a virtual patient from the pre-generated digital human waiting queue in the order of the consultation queue as the current consultation object, and then generates the corresponding symptom description text information or voice information according to the preset pathological state of the virtual patient, and at the same time calls the image generation module to output a physical sign image that can intuitively reflect the current pathological state, so as to facilitate the target user to make a preliminary diagnosis.

[0041] S102: Receive the target user's consultation instruction, and output symptom feedback corresponding to the pathological state based on the clinical conversation model.

[0042] Among them, medical consultation instructions refer to various inquiries and examination requirements issued by the target user during the diagnosis and treatment process; the clinical conversation model refers to an intelligent interactive system that can simulate doctor-patient dialogues; symptom feedback is used to indicate the response to medical consultations generated by the system based on the pathological state; the target users are doctors or medical students undergoing diagnosis and treatment training.

[0043] The diagnosis and treatment simulation system needs to have a natural consultation interaction with the target user. Specifically, the diagnosis and treatment simulation system receives various consultation instructions input by the target user, including inquiries about the chief complaint, current medical history, past medical history, etc., and inputs these instructions into the pre-trained clinical conversation model. The model will generate symptom feedback information that conforms to medical logic based on the pathological status database of the current digital person being treated, thereby realizing a real and smooth doctor-patient dialogue process.

[0044] It should be noted that this clinical conversation model is similar to the GPT large model. It is a doctor-patient dialogue generation model based on the transformer architecture. The input includes the user's medical consultation instructions, current pathological status data and historical dialogue context. It uses a multi-layer bidirectional attention mechanism for semantic understanding, and combines the medical knowledge graph to verify the accuracy of the answer. Finally, it outputs standardized symptom descriptions and physical sign feedback. The model training uses more than 1 million annotated doctor-patient dialogue data, uses the cross entropy loss function and BLEU score as optimization targets, and optimizes the model parameters through gradient descent iteration. During use, the model can understand more than 90% of non-standard medical consultation statements, and the generated answers have an accuracy rate of 95% in professional evaluations. It also supports multi-round dialogue reasoning and can adjust the level of detail of the answer according to the context.

[0045] In some embodiments, the consultation interaction can be achieved in a variety of ways: Optionally, the diagnosis and treatment simulation system can use natural language processing technology based on deep learning, accurately identify the intention of the consultation through semantic understanding, and combine the knowledge graph to perform multiple rounds of dialogue reasoning to generate complete and accurate symptom feedback; Optionally, the diagnosis and treatment simulation system can also be based on a preset question-and-answer rule library and template, match the corresponding symptom description template according to the keywords of the consultation instruction, and generate personalized feedback content through template filling. It is understandable that other dialogue generation methods can also be used to achieve doctor-patient consultation interaction, which is not limited here.

[0046] S103. In response to the target user's inspection instruction, the current digital person being treated is marked as a waiting-for-inspection state, and the inspection priority set by the target user is received.

[0047] Among them, the examination instructions represent various medical examination applications issued by the target user; the pending examination status refers to the mark that the digital person needs to undergo relevant examinations; the examination priority is used to indicate the urgency level of different examination items, including emergency, expedited, routine and other levels.

[0048] After receiving the inspection instructions from the target user, the diagnosis and treatment simulation system needs to manage the inspection process. Specifically, the diagnosis and treatment simulation system first marks the digital human who receives the inspection instructions as pending inspection, and allows the target user to set the priority for each inspection based on clinical judgment. These priority information will affect the return order of subsequent inspection results, simulating the allocation process of inspection resources in a real medical environment.

[0049] In some embodiments, the diagnosis and treatment simulation system can also recommend inspection priorities in a variety of ways, which is helpful for the target user to conduct simulation learning; optionally, the diagnosis and treatment simulation system can establish a dynamic priority scoring system based on the type of inspection, severity of the disease and medical resource conditions, and obtain the final inspection priority value through weighted calculation, and recommend the inspection priority based on this after the target user clicks the corresponding help button; optionally, the diagnosis and treatment simulation system can also adopt a rule-based priority allocation method to specify standardized priority levels for different types of inspection items according to preset clinical guidelines and expert experience. It is understandable that other methods can also be used to determine and manage inspection priorities, which are not limited here.

[0050] S104. While the target user continues to diagnose and treat other digital humans, the examination waiting time of the current digital human being is determined according to the preset examination time and examination priority, and the examination result of the current digital human being is generated after the examination waiting time.

[0051] Among them, the preset examination time represents the standard completion time of different types of examination items; the examination waiting time refers to the simulated time interval from issuing the examination to obtaining the results; the examination results are used to represent the data and reports of various types of medical examinations.

[0052] The diagnosis and treatment simulation system needs to simulate the process of generating examination results in a real medical environment. Specifically, while the target user continues to process other cases, the diagnosis and treatment simulation system will calculate the waiting time for each examination of the current digital person according to the preset completion time of the examination items and the previously set priority, and automatically generate examination results that meet the pathological status at the corresponding time point, realizing asynchronous processing of the examination process.

[0053] The inspection waiting time is calculated based on the dynamic resource scheduling model. The system establishes a resource constraint matrix that includes factors such as the number of inspection equipment, operators, and inspection complexity. The queuing theory model is used to describe the arrival and service process of the inspection task, and the inspection priority is converted into a service priority coefficient. Through the multi-objective planning algorithm, the optimal inspection scheduling plan is calculated by comprehensively considering the equipment utilization, waiting time fairness, and urgency. The system uses exponential smoothing to predict inspection load changes and dynamically adjust the waiting time coefficient. For example, when the system receives a CT examination application, it will calculate the benchmark waiting time based on the current equipment occupancy, the severity of the patient's condition, and the length of the appointment queue, and then combine the priority coefficient to obtain the final inspection waiting time.

[0054] Taking chest CT examination as an example, the current resource status is: 2 devices, 4 technicians, and 8 appointments. The benchmark examination time is 30 minutes, and the equipment utilization rate is 85%. The patient priority coefficient is 1.5 (expedited). The average waiting time of 120 minutes is calculated by the queuing theory model, and the priority adjustment formula is substituted: Final waiting time = benchmark waiting time × (1-priority coefficient × 0.3) = 120 × (1-1.5 × 0.3) = 66 minutes. In addition, the system will continue to optimize these parameters based on actual operation data, analyze the distribution characteristics of historical completion time through machine learning algorithms, and dynamically adjust the calculation parameters to make the preset time more in line with the actual situation; of course, in the actual simulation process, these times will support speed changes such as acceleration and delay, so that the target users can choose to adapt to their own simulation learning state.

[0055] In some embodiments, the diagnosis and treatment simulation system can also use a dynamic resource scheduling model based on queuing theory to calculate the inspection waiting time. The system first calculates the benchmark waiting time based on factors such as the current number of equipment, operators, and inspection queue length, and then introduces a priority adjustment mechanism, which is corrected by the formula "final waiting time = benchmark waiting time × (1-priority coefficient × 0.3)", where the priority coefficient is set to 1, 2, and 3 for routine, expedited, and emergency, respectively, and the adjustment factor 0.3 is used to balance the intensity of the impact of priority on waiting time. For example, for a CT examination under the expedited priority (coefficient 1.5), the original benchmark waiting time of 120 minutes will be adjusted to 66 minutes. The system will also continuously optimize the calculation parameters based on actual operating data to ensure the accuracy of the waiting time estimate.

[0056] S105. Obtain the target user's diagnosis result and treatment plan, and determine the diagnosis and treatment effect score of the current digital person based on the diagnosis result and treatment plan.

[0057] Among them, the diagnosis result represents the target user's judgment on the condition; the treatment plan refers to the treatment plan formulated based on the diagnosis result; the diagnosis and treatment effect score is used to represent the accuracy and rationality evaluation index of the diagnosis and treatment process.

[0058] The diagnosis and treatment simulation system needs to evaluate the diagnosis and treatment decisions of the target user. Specifically, the diagnosis and treatment simulation system receives the diagnosis results and treatment plans input by the target user, and through comparative analysis with the standard answers and compliance evaluation with expert rules, comprehensively calculates the effect score of the diagnosis and treatment, which is used to evaluate the clinical decision-making level of the target user.

[0059] It should be noted that the diagnosis and treatment simulation system can determine the diagnosis and treatment effect score through a multi-dimensional weighted calculation method. The system decomposes the diagnosis and treatment process into three dimensions: diagnostic accuracy, treatment rationality, and logical relevance, and assigns weight coefficients of 0.4, 0.4, and 0.2 respectively. The diagnosis score is obtained by calculating the text similarity with the standard answer, the treatment score is determined based on the rule evaluation system of multiple indicators such as medication indications, dosage, and course of treatment, and the logical relevance score evaluates the degree of match between the diagnostic conclusion and the treatment plan. The system calculates the final score through the formula of "total score = 0.4 × diagnosis score + 0.4 × treatment score + 0.2 × logical relevance score". For example, in a certain diagnosis and treatment, the diagnosis score is 85 points, the treatment score is 90 points, and the logical relevance score is 80 points, and the final diagnosis and treatment effect score is 86 points.

[0060] In some embodiments, the evaluation of the diagnosis and treatment effect can be achieved in a variety of ways: Optionally, the diagnosis and treatment simulation system can establish a multi-dimensional scoring system, conduct a comprehensive evaluation from the aspects of diagnostic accuracy, rationality of treatment plan, and drug safety, and calculate the final score through a machine learning algorithm; Optionally, the diagnosis and treatment simulation system can also build a scoring rule library based on expert consensus and clinical guidelines, and perform standardized scoring through rule matching. It is understandable that other methods can also be used to achieve the evaluation and scoring of the diagnosis and treatment effect, which is not limited here.

[0061] S106. Determine the diagnosis and treatment score of the target user based on the examination priority and diagnosis and treatment effect score of all digital humans in the digital human waiting queue, and update the physician ability rating of the target user based on the diagnosis and treatment score.

[0062] Among them, the diagnosis and treatment score represents the comprehensive performance evaluation of the target user in the entire diagnosis and treatment process; the physician ability rating is used to represent the professional skill level grading of the target user.

[0063] The diagnosis and treatment simulation system needs to evaluate the overall performance of the target user. Specifically, the diagnosis and treatment simulation system will comprehensively consider the rationality of the examination arrangements and the diagnosis and treatment effects of all the digital people who visit the patient, calculate the comprehensive diagnosis and treatment score of the target user, and dynamically adjust the physician's ability rating based on the score results, providing a basis for adjusting the difficulty of subsequent training.

[0064] In some embodiments, the ability rating update can be achieved in a variety of ways: Optionally, the diagnosis and treatment simulation system can use a dynamic rating algorithm to establish a user ability development model based on the historical data and change trends of the diagnosis and treatment scores to achieve adaptive adjustment of the rating; Optionally, the diagnosis and treatment simulation system can also trigger the rating update when the cumulative score reaches the corresponding threshold based on the preset grade classification standard. It is understandable that other methods can also be used to achieve the dynamic update of the physician ability rating, which is not limited here.

[0065] For example, the diagnosis and treatment simulation system can use the exponential smoothing method to update the physician's ability rating, and realize dynamic adjustment through the formula of "new rating score = α × historical rating + (1-α) × current diagnosis and treatment score", where α is the smoothing coefficient (value 0.7), which is used to control the influence weight of historical scores. The system divides the rating scores into five levels: 90 points or more for expert level, 80-90 points for advanced level, 70-80 points for intermediate level, 60-70 points for elementary level, and 60 points or less for internship level. For example, a physician's original rating score was 82 points (advanced), and the current diagnosis and treatment score was 86 points. After calculation, the new rating score was 83.2 points, maintaining the advanced rating. This rating mechanism not only maintains the stability of the rating, but also reflects the continuous improvement of the physician's ability.

[0066] In the above embodiment, the parallel management of multiple cases is achieved through the digital human waiting queue, and the reasonable allocation of medical resources is achieved through the inspection priority setting. In actual application, the system will dynamically adjust the training difficulty according to the user's ability level and provide personalized learning suggestions.

[0067] The following is a supplement to the scenario of this embodiment.

[0068] In long-term use, the diagnosis and treatment simulation system of this application has shown strong adaptability and scalability. For example, in the standardized training of residents in a teaching hospital, the system can automatically adjust the training difficulty according to the trainee's learning curve. For trainees who are making rapid progress, the system will increase the number of virtual patients in diagnosis at the same time and introduce more complex and in-depth case combinations. At the same time, the system will collect the performance data of trainees in different types of cases, build a personalized ability development model, and accurately push targeted learning resources. In addition, the system also supports multi-person collaborative training mode, which can simulate the cooperation scenarios of doctors, nurses, and medical technicians to help trainees develop teamwork skills.

[0069] After combining the above scenarios, the following is a more detailed description of the process of the method provided by this implementation. Figure 2 , is another flow chart of the diagnosis and treatment simulation method based on digital human in an embodiment of the present application.

[0070] S201. Determine the physician capability rating of the target user based on the user login information, and determine the case difficulty index and number of digital humans for diagnosis and treatment simulation based on the physician capability rating.

[0071] Among them, user login information represents the identity authentication data of the target user, including professional information such as title, department and years of practice; physician ability rating refers to the classification of physician diagnosis and treatment levels, which is divided into multiple levels from junior to expert; case difficulty index is used to indicate the complexity and challenge of the case; the number of digital people represents the number of virtual patients who need to be diagnosed and treated in a single training session.

[0072] The diagnosis and treatment simulation system needs to be configured for personalized training after the user logs in. Specifically, the diagnosis and treatment simulation system first analyzes the user's login information, combines historical training data to determine the user's current physician ability rating, and then automatically calculates the appropriate case difficulty index based on the rating level, and determines a reasonable number of people who can diagnose at the same time, ensuring that the training difficulty matches the user's ability, providing sufficient challenges without causing excessive pressure.

[0073] In some embodiments, training parameter configuration can be achieved in a variety of ways: Optionally, the diagnosis and treatment simulation system can use a multi-layer rating algorithm, first determine the basic level based on the user's professional title and years of practice, then fine-tune the level based on historical training performance data, and then calculate the corresponding case difficulty coefficient and recommended training volume through an adaptive algorithm; Optionally, the diagnosis and treatment simulation system can also build a rule base based on expert experience, determine training parameters through fuzzy reasoning, and dynamically optimize based on user feedback. It is understandable that other methods can also be used to configure training parameters, which are not limited here.

[0074] S202. Select multiple case data from a preset case library according to the case difficulty index, and determine multiple pathological conditions based on the multiple case data.

[0075] Among them, the case database refers to a pre-stored collection of standardized cases; case data represents detailed medical history, examination and treatment information; and pathological status is used to represent different stages and manifestations of disease development.

[0076] The diagnosis and treatment simulation system needs to select appropriate training cases according to the difficulty of the configuration. Specifically, the diagnosis and treatment simulation system will use the case difficulty index as a screening condition to retrieve matching case data from the preset graded case library, while ensuring that the selected cases cover different types of pathological conditions to form a complete diagnosis and treatment training scenario. The selection process will be based on the correlation and complementarity between cases to ensure the diversity and completeness of the training content.

[0077] In some embodiments, case selection can be achieved in a variety of ways: Optionally, the diagnosis and treatment simulation system can use a case retrieval method based on knowledge graphs, establish case feature vectors through semantic analysis, calculate the similarity with the target difficulty, and select the most suitable case combination; Optionally, the diagnosis and treatment simulation system can also adopt a dynamic selection strategy based on reinforcement learning to adjust the case selection direction according to the user's real-time performance. It is understandable that other methods can also be used to achieve intelligent case selection, which is not limited here.

[0078] S203: Generate a waiting queue of digital humans corresponding to the number of digital humans and in multiple pathological states.

[0079] Among them, the digital human waiting queue represents the sequence of virtual patients to be diagnosed; the pathological state distribution is used to represent the proportion of different disease types in the queue; the queue generation rule refers to the standard that controls the allocation of digital human features.

[0080] The diagnosis and treatment simulation system needs to build a consultation queue that conforms to actual clinical characteristics. Specifically, the diagnosis and treatment simulation system dynamically generates digital human models with different characteristics based on the number of digital humans and the distribution of pathological conditions determined in advance, and organizes these models into waiting queues according to preset queuing rules. The generation process will integrate the reasonable distribution of demographic characteristics such as gender and age to ensure that the queue structure conforms to the actual medical scene.

[0081] In some embodiments, queue generation can be achieved in a variety of ways: Optionally, the diagnosis and treatment simulation system can use a random generation algorithm based on demographics, simulate the distribution characteristics of the real population through the Monte Carlo method, and assign the pathological state to the digital human model according to the epidemiological law; Optionally, the diagnosis and treatment simulation system can also establish a feature extraction model based on real outpatient data, and build a diverse digital human queue through data enhancement. It is understandable that other methods can also be used to achieve the generation of digital human waiting queues, which are not limited here.

[0082] In some embodiments, the diagnosis and treatment simulation system will perform personalized digital human modeling based on the pathological state, that is, the diagnosis and treatment simulation system will collect basic training data including photos, videos and three-dimensional scanning data; then extract facial feature points, expression feature mapping and clothing features in the basic training data, and perform three-dimensional reconstruction of the digital human model to obtain an initial digital human; finally, different pathological states are input into the physiological parameter model, the complexion, posture and expression characteristics of the initial digital human are adjusted, and multiple different pathological digital humans are obtained.

[0083] Among them, basic training data refers to the original material data set used to construct digital humans, which is reasonable and legal data that has been approved by authorities and avoids sensitive information; facial feature points refer to the coordinate data of key positions of the face; expression feature mapping is used to represent the correspondence between facial muscle movements and emotional states; clothing features represent the visual parameters of digital humans' clothing; physiological parameter models refer to mathematical models that describe changes in the physiological state of the human body; and pathological digital humans are used to represent virtual character models with specific disease manifestations.

[0084] The diagnosis and treatment simulation system needs to build a pathological digital human model library during the training initialization phase. Specifically, the diagnosis and treatment simulation system first obtains the image and spatial information of the real person through multi-source data acquisition equipment, and then uses computer vision algorithms to extract key feature parameters to build the skeletal structure and appearance details of the three-dimensional character model. On this basis, the diagnosis and treatment simulation system inputs the preset pathological state parameters into the physiological parameter model, and adjusts the appearance of the model through biomechanical calculations and rendering engines, so that it presents the typical signs of the corresponding disease, thereby establishing a complete pathological digital human resource library.

[0085] It should be noted that the physiological parameter model is a deep neural network composed of a disease state mapper and a sign generator. The input is standardized pathological state parameters (including vital signs, laboratory tests and other indicators). The multi-layer perceptron simulates the interaction between various physiological systems and finally outputs visual feature parameters such as complexion, posture, and expression. The model is trained based on 50,000 real case data, using mean square error and medical expert scores as loss functions, and using the Adam optimizer for parameter updates. In application, the model can adjust the appearance of the digital human in real time according to the progression of the disease, with an accuracy rate of 92%, and supports the simulation of signs of more than 200 common diseases.

[0086] In some embodiments, the correctness of the digital human modeling is also tested by a medical feature consistency verification model. The system constructs a feature constraint network containing three levels of anatomical structure, physiological function and pathological changes. The topological relationship between features is analyzed using a graph neural network to detect the medical rationality of the feature combination. The feature patterns in real case data are extracted through a deep learning model to establish a feature standard library. The system calculates the similarity score between the generated digital human model and the feature standard, and sets multiple layers of thresholds for screening. For example, the system will detect the consistency between the digital human's appearance features such as complexion and posture and its vital signs and laboratory test results to ensure the accuracy of the model in medical logic. Taking the respiratory system examination as an example, the features extracted by the system include: cyanosis of the face 0.4, dyspnea 0.6, and rales of lung auscultation 0.7. The threshold range in the reference feature standard library is: [cyanosis of the face: 0.3-0.5, dyspnea: 0.5-0.8, rales: 0.6-0.8]. Calculate the feature similarity score: complexion matching degree 0.95, respiratory performance matching degree 0.92, and auscultation result matching degree 0.88. The overall similarity score was 0.92, exceeding the qualified threshold of 0.85, confirming that the modeling results were consistent with medical logic.

[0087] S204, retrieve the current digital human from the digital human waiting queue according to the order of consultation, and output symptom description and physical sign image according to the pathological state of the current digital human.

[0088] Referring to step S101, the diagnosis and treatment simulation system will output the symptom description and physical sign image of the digital person being treated.

[0089] S205: Receive the target user's consultation instruction, and output symptom feedback corresponding to the pathological state based on the clinical conversation model.

[0090] Referring to step S102 , the diagnosis and treatment simulation system will output symptom feedback corresponding to the pathological state upon receiving the diagnosis instruction from the target user.

[0091] Among them, the clinical conversation model adopts a hierarchical semantic understanding and response generation architecture. The system processes the content of the consultation dialogue through a bidirectional LSTM network enhanced by an attention mechanism, and identifies professional terms in combination with a medical terminology word embedding model. The reasoning engine based on the knowledge graph is used to analyze the correlation between symptoms and build a dynamic dialogue context tree. The question-answering strategy is optimized through deep reinforcement learning, and the response content is dynamically adjusted according to the development of the disease. The system uses adversarial training to improve the robustness of the model and ensure that the generated answers meet medical standards. For example, when the doctor asks the patient about the coughing symptoms, the system will combine the known medical history information and physical signs to generate a detailed description that conforms to the current pathological state. For example, when the doctor asks "How is the patient's recent coughing?", the system first identifies the keyword "cough" through the LSTM network with a weight of 0.8. Combined with contextual information (previously mentioned fever for 48 hours), the cough-related symptom description template is extracted from the knowledge graph. According to the current pathological state (cough degree 0.7), the answer is generated: "The patient has had frequent coughs in the past two days, mainly dry coughs, about 4-5 times per minute, aggravated at night, less sputum, and white mucus-like". The normative score of this answer is 0.94.

[0092] In some embodiments, the diagnosis and treatment simulation system can guide the target user to conduct a medical consultation, that is, the diagnosis and treatment simulation system will first obtain a set of medical consultation instructions input by the target user, the medical consultation instruction set including chief complaint inquiry instructions, medical history inquiry instructions and physical examination instructions; then, based on the clinical conversation model and the pathological status database of the current digital person being treated, output symptom feedback corresponding to the medical consultation instruction set; finally, based on the preset standard medical consultation process, calculate the standardization score of the medical consultation instruction set, and generate medical consultation guidance prompts for the target user according to the standardization score.

[0093] Among them, the consultation instruction set represents the various interactive commands of the doctor during the consultation process; the clinical conversation model refers to an artificial intelligence model that simulates doctor-patient dialogue; the pathological state database is used to represent disease-related clinical manifestation data; the standardization score refers to the degree of compliance of the consultation process with the standard process; and the consultation guidance prompts are used to indicate suggestions for improvement of the doctor's consultation behavior.

[0094] The diagnosis and treatment simulation system needs to provide intelligent interactive support during the consultation training stage. Specifically, the diagnosis and treatment simulation system receives the user's consultation instructions in real time, parses the instruction intent through the semantic understanding module, and queries the pathological state database to generate symptom descriptions and physical signs that match the condition. At the same time, the system will compare the user's consultation behavior with the preset standard process in real time, evaluate the completeness, logic and efficiency of the consultation, and provide personalized guidance and suggestions at key nodes to help users improve their consultation skills.

[0095] In some embodiments, intelligent consultation guidance can be achieved in a variety of ways: Optionally, the diagnosis and treatment simulation system can use a sequence model based on the attention mechanism to predict the best consultation path through the historical dialogue context, dynamically generate feedback content that conforms to clinical thinking, and evaluate the consultation quality based on the rule engine; Optionally, the diagnosis and treatment simulation system can also use reinforcement learning methods to model the consultation process as a Markov decision process, optimize the consultation strategy through a dynamic programming algorithm, and achieve intelligent consultation guidance. It is understandable that other methods can also be used to implement the consultation guidance function, which is not limited here.

[0096] In some embodiments, the diagnosis and treatment simulation system will adjust the system load to reduce the data processing delay of the digital human, that is, the diagnosis and treatment simulation system will obtain the system load rate of the diagnosis and treatment simulation system; and when the system load rate is higher than the preset load threshold, the clinical conversation model will be adjusted to a lightweight model.

[0097] Among them, the system load rate indicates the resource usage of the diagnosis and treatment simulation system; the preset load threshold refers to the warning line value of the system performance; the data processing delay is used to indicate the time difference of the system response; the lightweight model refers to a simplified version of the clinical session processing model.

[0098] The diagnosis and treatment simulation system needs to maintain stable service performance during operation. Specifically, the diagnosis and treatment simulation system continuously monitors system resource indicators such as CPU usage, memory usage, network bandwidth, and calculates the comprehensive load rate. When the load rate exceeds the preset threshold, the system automatically triggers the performance optimization mechanism to reduce computing pressure by downgrading the model, ensuring that the interactive response time of the digital human is maintained within an acceptable range, while ensuring that the basic diagnosis and treatment simulation functions are not affected.

[0099] In some embodiments, system load regulation can be achieved in a variety of ways: Optionally, the diagnosis and treatment simulation system can adopt a dynamic resource scheduling strategy, through task priority sorting and resource pool management, to reasonably allocate computing resources while ensuring core functions; Optionally, the diagnosis and treatment simulation system can also implement a multi-level cache mechanism, through pre-calculation and result caching to reduce repeated operations and improve system response speed. It is understandable that other methods can also be used to achieve dynamic optimization of system performance, which is not limited here.

[0100] It should be noted that the lightweight model here is implemented using model compression and knowledge distillation technology. First, the structure of the original clinical conversation model is simplified, and the neuronal connections with less contribution are removed through network pruning to reduce the number of model parameters. Then the teacher-student network framework is used for knowledge transfer, and the training process of the lightweight model is guided by the complete model. In the quantization stage, the model weights are converted from 32-bit floating point numbers to 8-bit integer representations to reduce storage and computing overhead. At the same time, the sparse processing of the attention mechanism is used to retain only the key semantic feature calculations. In the end, the lightweight model reduces the amount of calculation to less than 30% of the original model while maintaining a response accuracy of more than 80%. The system will also dynamically adjust the calculation accuracy of the model according to the load conditions, and in extreme cases it can be further downgraded to a simple rule-based question-and-answer mode.

[0101] S206. In response to the target user's inspection instruction, the current digital person being treated is marked as a waiting-for-inspection state, and the inspection priority set by the target user is received.

[0102] Referring to step S103, after receiving the examination instruction from the target user, the diagnosis and treatment simulation system will mark the current digital person being treated and record the priority set by the user.

[0103] S207. While the target user continues to diagnose and treat other digital humans, the examination waiting time of the current digital human being is determined according to the preset examination time and examination priority, and the examination result of the current digital human being is generated after the examination waiting time.

[0104] Referring to step S104, the diagnosis and treatment simulation system will generate the examination results of the current digital person being treated.

[0105] S208, obtaining the target user's diagnosis results and treatment plan, and determining the diagnosis and treatment effect score of the current digital person according to the diagnosis results and treatment plan.

[0106] Referring to step S105, the diagnosis and treatment simulation system will determine the diagnosis and treatment effect score of the current digital person being treated.

[0107] In some embodiments, the diagnosis and treatment simulation system will conduct a multi-dimensional analysis based on the diagnosis results and treatment plans of the target user to determine the score of the diagnosis and treatment effect. That is, the diagnosis and treatment simulation system will obtain the diagnosis results and treatment plans input by the target user, and determine the logical association score of the diagnosis results and the treatment plan; then match the diagnosis results with the standard diagnosis results of the preset pathological state for similarity to obtain the diagnosis score; and determine the treatment score of the treatment plan based on multiple treatment standards based on drug information data and clinical knowledge graphs; the treatment standards include indications, dosage, course of treatment and contraindications; finally, the diagnosis and treatment effect score of the current digital person is determined based on the logical association score, diagnosis score and treatment score.

[0108] Among them, the logical association score represents the rationality score between diagnosis and treatment; similarity matching refers to the degree of matching between the diagnosis result and the standard answer; the clinical knowledge graph is used to represent the association network of medical knowledge; and the treatment standard refers to multiple dimensional indicators for evaluating treatment plans.

[0109] The diagnosis and treatment simulation system needs to conduct a comprehensive evaluation of the user's diagnosis and treatment process. Specifically, the diagnosis and treatment simulation system first analyzes the internal logical relationship between the diagnosis results and the treatment plan input by the user, and evaluates the matching degree between the disease diagnosis and the treatment measures. Then the user's diagnosis results are semantically compared with the preset standard answers to calculate the accuracy of the diagnosis. At the same time, the system will verify the rationality of the treatment plan based on the clinical knowledge graph, and evaluate the standardization and safety of the treatment plan from multiple dimensions. Finally, the system integrates the scores of each item according to the preset weights to obtain a comprehensive diagnosis and treatment effect score.

[0110] In some embodiments, the diagnosis and treatment effect evaluation can be achieved in a variety of ways: Optionally, the diagnosis and treatment simulation system can use an ontology-based reasoning engine to verify the integrity and correctness of the diagnosis and treatment process through formal medical knowledge representation and logical reasoning; Optionally, the diagnosis and treatment simulation system can also use a multi-layer perceptron model, take the feature vectors of each evaluation dimension as input, learn the expert scoring model, and achieve automated effect evaluation. It is understandable that other methods can also be used to achieve multi-dimensional evaluation of diagnosis and treatment effects, which are not limited here.

[0111] S209. Determine the diagnosis and treatment score of the target user based on the examination priority and diagnosis and treatment effect score of all digital humans in the digital human waiting queue, and update the physician ability rating of the target user based on the diagnosis and treatment score.

[0112] Referring to step S106 , the diagnosis and treatment simulation system determines the diagnosis and treatment score and updates the physician capability rating.

[0113] S210. Construct a diagnosis and treatment capability development prediction model based on the changing trend of the diagnosis and treatment scores of the target users.

[0114] Among them, the diagnosis and treatment score represents the performance score of the target user during the diagnosis and treatment process; the change trend refers to the dynamic change characteristics of the score over time; the diagnosis and treatment ability development prediction model is used to represent the mathematical model for the future improvement of the doctor's ability; the prediction period represents the time span of the model.

[0115] The diagnosis and treatment simulation system needs to conduct predictive analysis on the user's learning progress. Specifically, the diagnosis and treatment simulation system first collects the historical diagnosis and treatment score data of the target user, extracts the changing rules of the score through time series analysis, and establishes a dual prediction model that includes short-term fluctuations and long-term trends. This model is not only used to determine the improvement trajectory of the overall diagnosis and treatment capabilities, but also to subdivide the changes in the diagnosis and treatment levels of different types of diseases, thereby forming a multi-dimensional capability development prediction.

[0116] It should be noted that the diagnosis and treatment ability development prediction model here is a sequence prediction network that combines LSTM and attention mechanisms. The input includes user historical training data (rating records, learning time, error types, etc.) and ability evaluation indicators. Through time series feature extraction and trend analysis, the user's ability development trajectory on different types of cases is predicted. The model is trained using 100,000 user learning records, with prediction accuracy and time series consistency as optimization goals, and an early stopping strategy is used to prevent overfitting. In actual applications, the model can predict the user's ability bottleneck in advance, with a prediction accuracy of 85%, and provide data support for the formulation of personalized learning plans, and can dynamically adjust the training difficulty and content recommendations.

[0117] Specifically, the diagnosis and treatment simulation system uses a time series-based deep learning model to predict the user's ability development trend. The system first collects historical performance data of users in different types of cases, including feature vectors such as diagnosis and treatment scores, learning time, and error types. Then, a time series prediction model is established through the LSTM network. The model adopts the prediction function of "prediction score = W1×historical score sequence+W2×learning input+W3×error improvement rate", where W1, W2, and W3 are dynamic weight coefficients, which are adaptively adjusted by the model according to data characteristics. The system uses a cross-validation method to evaluate the prediction accuracy, and the model is considered valid when the prediction error is less than 15%. For example, the historical score sequence of a user in the diagnosis and treatment of respiratory diseases is [75, 78, 82, 85]. Combined with his learning input of 0.8 and error improvement rate of 0.6, the system predicts that his ability level in the next stage will be 88 points, and the prediction accuracy reaches 85%.

[0118] In addition, the diagnosis and treatment simulation system can adopt a three-layer fuzzy comprehensive evaluation structure when performing diagnosis and treatment scoring. First, an evaluation factor set U and a comment set V are established, where U includes multiple dimensions such as clinical thinking, professional knowledge, and operating specifications, and V is set to four levels of {excellent, good, qualified, and unqualified}. A membership function is established for each evaluation factor, and the membership parameter is determined by combining expert scoring and machine learning. In the weight determination stage, the hierarchical analysis method is used to construct a judgment matrix and calculate the weight vector W of each factor. The evaluation process first performs a first-level fuzzy evaluation on each specific indicator to obtain a single-factor evaluation matrix R, and then obtains a comprehensive evaluation result through a fuzzy synthesis operation B=WxR. Finally, the center of gravity method is used for clarification processing to convert the fuzzy evaluation result into a specific score. For example, when judging the factor of "clinical thinking", the system will analyze the user's questioning logic, diagnostic reasoning process, and treatment decision-making basis, calculate the membership of each comment level through semantic analysis, and then combine it with the weight of the factor to obtain the sub-item score.

[0119] The diagnosis and treatment simulation system can also use a multi-dimensional scoring system to quantify the diagnosis and treatment scores. The scoring calculation is based on the decision tree model, which decomposes the diagnosis process into four key links: symptom identification, examination selection, diagnostic reasoning and treatment plan. Multiple evaluation indicators are set for each link, and the single score is calculated by the fuzzy comprehensive evaluation method. Among them, the accuracy of symptom identification accounts for 30%, the rationality of examination selection accounts for 25%, the logic of diagnostic reasoning accounts for 25%, and the appropriateness of the treatment plan accounts for 20%. The score of each indicator is obtained by calculating the similarity with the standard answer, using the cosine similarity algorithm for text matching, and the edit distance algorithm to evaluate the accuracy of the diagnostic statement. The final diagnosis and treatment score is the weighted sum of the scores of each link, and then normalized to a score range of 0-100 by the Sigmoid function.

[0120] S211. Develop a prediction model based on diagnosis and treatment capabilities to generate learning suggestions and learning resource links for different types of cases.

[0121] Among them, learning suggestions refer to improvement plans generated based on prediction results; learning resource links refer to the index of teaching materials related to the suggestions; resource matching degree is used to indicate the degree of fit between learning resources and user needs; and priority recommendation index indicates the priority of resource recommendations.

[0122] The diagnosis and treatment simulation system needs to provide personalized learning guidance based on the prediction results. Specifically, the diagnosis and treatment simulation system identifies the user's strengths and weaknesses in the diagnosis and treatment of different diseases based on the ability development prediction model, generates targeted learning suggestions based on the medical knowledge system, and retrieves relevant teaching materials from the resource library to establish a collection of learning resource links to provide users with comprehensive ability improvement support.

[0123] It should be noted that the diagnosis and treatment simulation system can adopt a hierarchical resource link generation strategy. First, an association network of medical knowledge points and learning resources is constructed, and each knowledge point is mapped to multiple learning resources of different types and difficulties. The resource link generation process first identifies the user's knowledge shortcomings, and then conducts a breadth-first search in the association network to construct a candidate resource set. The candidate resources are scored, and the scoring criteria include multiple dimensions such as content relevance, difficulty matching, and learning effect feedback. The PageRank algorithm is used to calculate the importance weight of the resources, and the optimal resource combination is selected through a multi-objective optimization algorithm combined with the user's learning style preferences. The system also maintains the timeliness indicators of the resources, regularly updates expired or low-quality links, and ensures the effectiveness of the recommended resources.

[0124] In some embodiments, learning guidance generation can be achieved in a variety of ways: Optionally, the diagnosis and treatment simulation system can use knowledge graph technology to establish a mapping relationship between the user's ability shortcomings and standardized medical knowledge points, plan the optimal learning path through the path analysis algorithm, and associate the corresponding learning resources; Optionally, the diagnosis and treatment simulation system can also use a collaborative filtering recommendation algorithm to predict the most effective learning plan and resource combination based on the learning trajectories of similar users. It is understandable that other methods can also be used to achieve the generation of personalized learning guidance, which is not limited here.

[0125] S212: Display a learning plan including learning suggestions, learning resource links, and recommended simulation time.

[0126] Among them, the recommended simulation time refers to the recommended training duration; the learning plan refers to the training plan including specific content and time arrangement.

[0127] The diagnosis and treatment simulation system needs to convert learning suggestions into executable plans. Specifically, the diagnosis and treatment simulation system integrates learning suggestions, resource links and time suggestions to generate a structured learning plan, which is clearly presented to users through a visual interface for easy understanding and execution. The displayed content will be dynamically updated according to the user's learning progress to ensure the timeliness and pertinence of the plan.

[0128] In some embodiments, the plan display can be implemented in a variety of ways: Optionally, the diagnosis and treatment simulation system can use an adaptive layout algorithm to automatically adjust the display structure according to the importance and relevance of the content to highlight the display effect of key information; Optionally, the diagnosis and treatment simulation system can also optimize the presentation of the learning plan based on user interaction data to provide a personalized browsing and operation experience. It is understandable that other methods can also be used to implement the display of the learning plan, which is not limited here.

[0129] In the embodiment of the present application, due to the adoption of a multi-case parallel management mechanism based on a digital human waiting queue, an intelligent examination priority scheduling system, and a multi-dimensional diagnosis and treatment effect evaluation system, it is possible to truly simulate the complex scenario in which doctors in a clinical environment need to manage multiple patients and coordinate various types of medical resources at the same time. It effectively solves the problems in related technologies such as the inability to cultivate complex scenario resource allocation capabilities through single case training, insufficient ability to control work rhythm due to the lack of a real examination waiting mechanism, and difficulty in comprehensively improving the diagnosis and treatment level due to a single evaluation system. It thereby helps medical students gradually adapt to the multi-threaded work mode in a low-pressure environment, master examination resource allocation skills, and establish a standardized clinical thinking model, ultimately achieving the goal of improving overall clinical diagnosis and treatment capabilities.

[0130] The following describes the diagnosis and treatment simulation system in the embodiment of the present invention from the perspective of hardware processing. Figure 3 , is a schematic diagram of a physical device structure of a diagnosis and treatment simulation system in an embodiment of the present application.

[0131] It should be noted that Figure 3 The structure of the diagnosis and treatment simulation system shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0132] like Figure 3 As shown, the diagnosis and treatment simulation system includes a CPU (Central Processing Unit) 301, which can perform various appropriate actions and processes according to the program stored in the ROM (Read-Only Memory) 302 or the program loaded from the storage part 308 to the RAM (Random Access Memory) 303, such as executing the method described in the above embodiment. Various programs and data required for system operation are also stored in the RAM 303. The CPU 301, ROM 302 and RAM 303 are connected to each other through a bus 304. The I / O (Input / Output) interface 305 is also connected to the bus 304.

[0133] The following components are connected to the I / O interface 305: an input section 306 including an audio input device, a button switch, etc.; an output section 307 including a liquid crystal display (LCD) and an audio output device, an indicator light, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 310 as needed so that a computer program read therefrom is installed into the storage section 308 as needed.

[0134] In particular, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 309, and / or installed from the removable medium 311. When the computer program is executed by the CPU (central processing unit) 301, various functions defined in the present invention are performed.

[0135] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus, or device.

[0136] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings.

[0137] Specifically, the diagnosis and treatment simulation system of this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, the diagnosis and treatment simulation method based on digital human provided in the above embodiment is implemented.

[0138] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the diagnosis and treatment simulation system described in the above embodiment; or may exist independently without being assembled into the diagnosis and treatment simulation system. The above storage medium carries one or more computer programs, and when the above one or more computer programs are executed by a processor of the diagnosis and treatment simulation system, the diagnosis and treatment simulation system implements the diagnosis and treatment simulation method based on digital human provided in the above embodiment.

[0139] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0140] As used in the above embodiments, the term "when..." may be interpreted to mean "if..." or "after..." or "in response to determining..." or "in response to detecting...", depending on the context. Similarly, the phrases "upon determining..." or "if (the stated condition or event) is detected" may be interpreted to mean "if determining..." or "in response to determining..." or "upon detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)", depending on the context.

[0141] Those skilled in the art can understand that to implement all or part of the processes in the above-mentioned embodiments, the processes can be completed by computer programs to instruct related hardware, and the programs can be stored in computer-readable storage media. When the programs are executed, they can include the processes of the above-mentioned method embodiments. The aforementioned storage media include: ROM or random access memory RAM, magnetic disk or optical disk and other media that can store program codes.

Claims

1. A diagnosis and treatment simulation method based on digital human, characterized in that: Applied to a diagnosis and treatment simulation system, the method comprises: Retrieving the current digital person from the digital person waiting queue according to the order of consultation, and outputting symptom description and physical sign image according to the pathological state of the current digital person; Receive a target user's inquiry instruction, and output symptom feedback corresponding to the pathological state based on the clinical conversation model; In response to the inspection instruction of the target user, marking the current digital person of the patient as a waiting-for-inspection state, and receiving the inspection priority set by the target user; While the target user continues to diagnose and treat other digital people, the examination waiting time of the current digital person is determined according to the preset examination time and the examination priority, and the examination result of the current digital person is generated after the examination waiting time; Obtaining the diagnosis result and treatment plan of the target user, and determining the diagnosis and treatment effect score of the current digital person according to the diagnosis result and the treatment plan; Based on the examination priorities and diagnosis and treatment effect scores of all digital humans in the digital human waiting queue, the diagnosis and treatment score of the target user is determined, and the physician ability rating of the target user is updated according to the diagnosis and treatment score.

2. The method according to claim 1, characterized in that Before the step of retrieving the current digital person from the digital person waiting queue according to the order of consultation, and outputting the symptom description and physical sign image according to the pathological state of the current digital person, the method further includes: Determine the physician capability rating of the target user according to the user login information, and determine the case difficulty index and the number of digital people for diagnosis and treatment simulation according to the physician capability rating; Selecting a plurality of case data from a preset case library according to the case difficulty index, and determining a plurality of pathological conditions based on the plurality of case data; A waiting queue of digital humans corresponding to the number of digital humans and in the multiple pathological states is generated.

3. The method according to claim 2, characterized in that Before the step of generating a waiting queue of digital humans corresponding to the number of digital humans and in the multiple pathological states, the method further includes: Collect basic training data including photos, videos and 3D scan data; Extracting facial feature points, expression feature mapping and clothing features from the basic training data, and performing three-dimensional reconstruction on the digital human model to obtain an initial digital human; Different pathological conditions are input into the physiological parameter model, and the complexion, posture and expression characteristics of the initial digital human are adjusted to obtain a plurality of different pathological digital humans.

4. The method according to claim 1, characterized in that: The step of receiving the target user's consultation instruction and outputting symptom feedback corresponding to the pathological state based on the clinical conversation model specifically includes: Acquire a set of medical inquiry instructions input by the target user; the set of medical inquiry instructions includes a chief complaint inquiry instruction, a medical history inquiry instruction, and a physical examination instruction; Based on the clinical conversation model and the pathological status database of the current digital person being treated, output symptom feedback corresponding to the set of consultation instructions; Calculating the standardization score of the set of medical consultation instructions based on a preset standard medical consultation process; Generate a diagnosis guidance prompt for the target user based on the standardization score.

5. The method according to claim 1, characterized in that The step of obtaining the diagnosis result and treatment plan of the target user, and determining the diagnosis and treatment effect score of the current digital person according to the diagnosis result and the treatment plan, specifically includes: Obtaining the diagnosis result and treatment plan input by the target user, and determining a logical association score between the diagnosis result and the treatment plan; The diagnostic result is matched with the standard diagnostic result of the preset pathological condition for similarity to obtain a diagnostic score; Determine, based on drug information data and clinical knowledge graph, a treatment score of the treatment plan on multiple treatment criteria; the treatment criteria include medication indications, dosage, course of treatment, and contraindications; The diagnosis and treatment effect score of the current visiting digital person is determined according to the logical association score, the diagnosis score and the treatment score.

6. The method according to claim 1, characterized in that After the step of determining the diagnosis and treatment score of the target user based on the examination priorities and diagnosis and treatment effect scores of all digital humans in the digital human waiting queue, and updating the physician capability rating of the target user according to the diagnosis and treatment score, the method further includes: Constructing a diagnosis and treatment capability development prediction model based on the changing trend of the diagnosis and treatment scores of the target users; Generating learning suggestions and learning resource links for different types of cases based on the diagnostic and treatment capability development prediction model; A learning plan including the learning suggestion, the learning resource link and the suggested simulation time is displayed.

7. The method according to claim 1, characterized in that Before the step of receiving the target user's inquiry instruction and outputting symptom feedback corresponding to the pathological state based on the clinical conversation model, the method further includes: Obtaining a system load rate of the diagnosis and treatment simulation system; When the system load rate is higher than a preset load threshold, the clinical conversation model is adjusted to a lightweight model.

8. A diagnosis and treatment simulation system, characterized in that: The diagnosis and treatment simulation system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the diagnosis and treatment simulation system to execute the method described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a diagnosis and treatment simulation system, the diagnosis and treatment simulation system is caused to execute the method as described in any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product runs on a diagnosis and treatment simulation system, the diagnosis and treatment simulation system is enabled to perform the method according to any one of claims 1 to 7.

Images