Medical teaching and training system based on AI-driven high-simulation SP medical history collection

Abstract

The invention relates to a medical teaching training system based on AI-driven high-simulation SP medical history acquisition, belongs to the field of teaching tools, and is used for extracting corresponding case data according to teaching requirements and determining symptoms and physiological indexes of AI-driven high-simulation SP. Constructing an SP body model; extracting an entity of the medical history acquisition knowledge graph from the current question of the student, and searching a triple corresponding to the entity from the medical history acquisition knowledge graph to obtain a first set; obtaining an SP state entity according to the AI-driven high-simulation SP symptoms, the physiological indexes and the body parameters; searching a triple corresponding to the SP state entity from the first set to obtain a second set; and according to the second set, in combination with the SP character parameters and the emotion parameters, generating an answer of the SP. According to the invention, a large number of medical history collection training opportunities with AI-driven high-simulation SP diversification and low cost can be provided, the teaching cost of standardized patient medical history collection is reduced, and the performance of virtual patients is more stable and objective.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of teaching tools, in particular to a medical teaching training system based on AI-driven high-simulation SP medical history collection. BACKGROUND

[0002] In medical education, it is very important to cultivate the medical history collection ability of students. The traditional teaching method is usually to let students directly face real patients for medical history collection training. However, the conditions of real patients are relatively complex, and the number is limited, so it is difficult to meet the training needs of a large number of students. Later, standardized patients (Standardized Patients, SP) simulation, also known as simulate patients (Simulate Patients), is to have trained actors play the role of patients. However, the performance of the actors may be affected by subjective factors, and the cost is also relatively high. SUMMARY

[0003] In order to solve the above problems, the present application provides an AI-driven high-simulation SP medical history collection medical teaching training system, which comprises: An AI-driven high-simulation SP generation module is used for extracting corresponding case data from a case database according to teaching needs, determining symptoms and physiological indexes of AI-driven high-simulation SP according to the case data, and determining or updating AI-driven high-simulation SP body model parameters according to preset body parameters. An AI-driven high-simulation SP display module is used for real-time acquisition of the AI-driven high-simulation SP body model parameters and construction of an AI-driven high-simulation SP body model. An AI-driven high-simulation SP medical history collection interaction module is used for obtaining current questions of students, extracting entities of a medical history collection knowledge graph from the current questions of the students to obtain first question entities, constructing the medical history collection knowledge graph based on verified medical literature and case data, finding corresponding triples from the medical history collection knowledge graph according to the first question entities to obtain a first triple set, obtaining AI-driven high-simulation SP state entities according to the symptoms, physiological indexes and body parameters of the AI-driven high-simulation SP, finding triples corresponding to the AI-driven high-simulation SP state entities from the first triple set to obtain a second triple set, and generating answers of the AI-driven high-simulation SP based on the second triple set, the AI-driven high-simulation SP personality parameters and the emotion parameters.

[0004] Further, the AI-driven high-simulation SP medical history collection medical teaching training system further comprises: A case database construction module is used for establishing a case database according to a plurality of verified case data. The Medical History Collection Knowledge Graph Construction Module is used to construct a medical history collection knowledge graph based on validated medical literature and case data.

[0005] Furthermore, the case database construction module is specifically used for: The collected and verified case data are preprocessed to remove erroneous, redundant, and private data. Based on the preprocessed case data, the data is classified by symptoms, diseases, and treatment plans, and corresponding symptom indexes, disease indexes, and treatment plan indexes are established to obtain a case database.

[0006] Furthermore, the medical history collection knowledge graph construction module is specifically used for: Collect and validate medical literature and case data, clean and structure the medical literature and case data to obtain a medical dataset; perform entity recognition on the medical dataset to obtain multiple entity libraries, each of which includes multiple entities; merge synonyms for entities whose similarity exceeds a threshold; extract relations from the medical dataset based on preset relation templates and entities to obtain multiple triples reflecting the relations between entities.

[0007] Furthermore, the AI-driven high-fidelity SP generation module is specifically used for: Based on the diseases or symptoms involved in the teaching needs, case data related to the diseases or symptoms are screened from the case database. The case data includes the patient's symptoms, physiological indicators, and body parameters, including gender, age, height, and weight. Based on the symptoms in the case data, multiple identical or similar symptoms are obtained, and the symptoms of the AI-driven high-fidelity simulation SP are selected from these. Based on the physiological indicators and body parameters in the case data, physiological indicator ranges for different genders and age groups are determined. Based on the teaching needs, the gender and age of the AI-driven high-fidelity simulation SP are determined. Based on the gender and age of the AI-driven high-fidelity simulation SP, the physiological indicators, height, and weight of the AI-driven high-fidelity simulation SP are selected from the corresponding physiological indicator ranges for the gender and age group. Based on the height and weight of the AI-driven high-fidelity simulation SP, the body model size of the AI-driven high-fidelity simulation SP is determined or updated. Based on the symptoms and physiological indicators of the AI-driven high-fidelity simulation SP, the body model morphology of the AI-driven high-fidelity simulation SP is determined or updated.

[0008] Furthermore, the AI-driven high-fidelity SP display module is specifically used for: The parameters of the AI-driven high-fidelity SP body model are acquired in real time, including the size and shape of the AI-driven high-fidelity SP body model. The AI-driven high-fidelity SP body model is established based on the gender, age, and size of the AI-driven high-fidelity SP body model. The AI-driven high-fidelity SP body model is then adjusted based on its shape.

[0009] Furthermore, based on AI-driven highly realistic SP symptoms, physiological indicators, and body parameters, an AI-driven highly realistic SP state entity is obtained, including: The entity with the highest correlation to the symptoms of the AI-driven highly realistic SP is found in the medical history collection knowledge graph. The entity with the highest correlation to the physiological indicators of the AI-driven highly realistic SP is found in the medical history collection knowledge graph. The entity with the highest correlation to the body parameters of the AI-driven highly realistic SP is found in the medical history collection knowledge graph. The above three types of entities are identified as the state entities of the AI-driven highly realistic SP.

[0010] Furthermore, the AI-driven highly realistic SP medical history collection and interaction module is also used for: if there is no triple corresponding to the AI-driven highly realistic SP state entity in the first set of triples, then combining the AI-driven highly realistic SP symptoms, physiological indicators, and body parameters to obtain multiple AI-driven highly realistic SP state entities; searching for triples corresponding to each AI-driven highly realistic SP state entity from the first set of triples to obtain a third set of triples; and generating the AI-driven highly realistic SP's response based on the third set of triples, combined with the AI-driven highly realistic SP personality parameters and emotional parameters.

[0011] Furthermore, based on the second set of the triples, and combined with the AI-driven highly realistic SP personality parameters and emotion parameters, the AI-driven highly realistic SP's answers are generated, including: Calculate the relevance of each triplet in the second set of triplets to the student's current question; calculate the frequency of occurrence of the entity and relation of each triplet in the second set of triplets; determine the score of each triplet in the second set of triplets based on the relevance and the frequency of occurrence of the entity and relation; obtain AI-driven high-fidelity SP personality parameters and emotional parameters through the AI-driven high-fidelity SP medical history collection interaction module; establish an AI-driven high-fidelity SP intelligent agent connected to the large language model data based on the AI-driven high-fidelity SP symptoms, physiological indicators, body parameters, personality parameters, and emotional parameters. The language model is trained based on validated medical literature and case data, a personality language database, and an emotion database. The personality language database includes structured statements corresponding to different personalities, and the emotion database includes structured statements corresponding to different emotions. The highest-scoring triplet, the student's and AI-driven high-fidelity SP's historical question-and-answer sessions, the AI-driven high-fidelity SP's personality parameters, and emotion parameters are used as context. The student's current question is used as a prompt word and input into the AI-driven high-fidelity SP agent. The AI-driven high-fidelity SP agent obtains the answer based on the context, prompt word, personality parameters, and emotion parameters from the large language model and feeds the answer back to the AI-driven high-fidelity SP's medical history collection and interaction module.

[0012] Furthermore, the AI-driven high-fidelity SP medical history acquisition medical teaching and training system also includes: an AI-driven high-fidelity SP physiological indicator display module, used to acquire and display multiple physiological indicators of the AI-driven high-fidelity SP in real time.

[0013] The beneficial effects of the above-mentioned technical solutions provided in this application include at least the following: This application generates AI-driven, highly realistic patient (SP) symptoms and physiological indicators based on real case data. Using this data, a body model is constructed for physical examination, facilitating more realistic history-taking experience for trainees. Based on trainee questions, an initial range of triplets is obtained from the history-taking knowledge graph. This range is further narrowed down using the AI-driven SP's symptoms and physiological indicators, generating responses from the AI-driven SP. Trainees can then verify their diagnostic judgments and gain diagnostic experience based on these responses. This application can simulate various patient types, including their symptoms, personality, and emotions. Trainees interacting with the AI-driven SP during history-taking is like communicating with a real patient. Furthermore, the AI-driven SP can automatically provide reasonable answers to trainee questions, guiding them to a deeper understanding of the condition. The virtual patient's performance is more stable and objective. The medical teaching and training system provided in this application offers medical learners access to medical history taking practice resources that are not limited by the number or time of real or standardized patients. It provides trainees with a large number of diverse medical history taking training opportunities at a relatively low cost, and can be adjusted and modified at any time according to teaching needs. Furthermore, the performance of virtual patients is more stable and objective, providing trainees with a more standardized training environment, which helps improve their medical history taking skills and clinical reasoning abilities.

[0014] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings.

[0015] The technical solution of this application will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0016] The accompanying drawings are provided to further illustrate the present application and form part of the specification. They are used together with the embodiments of the present application to explain the application and do not constitute a limitation thereof. In the drawings: Figure 1 This is a block diagram of the AI-driven, highly realistic SP (Specialist) medical history collection system in this application embodiment. Detailed Implementation

[0017] Exemplary embodiments of the present application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this application will be thorough and complete, and will fully convey the scope of the present application to those skilled in the art.

[0018] To address the problems existing in the prior art, this application provides an AI-driven, highly realistic SP (Specialist) medical history acquisition system for medical teaching and training, referring to... Figure 1 As shown, this AI-driven, highly realistic SP (Specialist) medical history collection medical teaching and training system includes: The AI-driven high-fidelity SP (Special Person Actor) generation module is used to extract relevant case data from a case database according to teaching needs. Based on the case data, it determines the symptoms and physiological indicators of the AI-driven high-fidelity SP. It also determines or updates the body model parameters of the AI-driven high-fidelity SP based on preset body parameters. For example, if teaching requires learning about cardiac history taking, it extracts cardiac case data from the case database. Based on this data, it determines the range of "symptoms" and the range of changes in various "physiological indicators" for the AI-driven high-fidelity SP. From these ranges, it selects the determined "symptoms" and "physiological indicators" as the symptoms and physiological indicators for the AI-driven high-fidelity SP. Body parameters include gender, age, height, and weight. Based on these parameters, a preset conversion ratio is used to obtain the body model parameters for the AI-driven high-fidelity SP.

[0019] The AI-driven high-fidelity SP display module is used to acquire the parameters of the AI-driven high-fidelity SP body model in real time and construct the AI-driven high-fidelity SP body model. Based on the parameters of the AI-driven high-fidelity SP body model, modeling tools such as Blender, Maya, or MetaHuman Creator are used to construct the AI-driven high-fidelity SP body model. Trainees obtain basic physical information of the AI-driven high-fidelity SP by observing the AI-driven high-fidelity SP body model. For example, by collecting authorized portraits of real standardized patients (including photos, videos, audio, movements, expressions, symptoms, and vital signs), a highly realistic digital human is constructed.

[0020] The AI-driven highly realistic SP medical history collection interaction module is used to obtain the student's current question, extract entities from the medical history collection knowledge graph from the student's current question to obtain a first question entity, the medical history collection knowledge graph being constructed based on verified medical literature and case data; based on the first question entity, search for the corresponding triples from the medical history collection knowledge graph to obtain a first set of triples; based on the AI-driven highly realistic SP symptoms, physiological indicators, and body parameters, obtain an AI-driven highly realistic SP state entity; search for the triples corresponding to the AI-driven highly realistic SP state entity from the first set of triples to obtain a second set of triples; based on the second set of triples, combined with the AI-driven highly realistic SP personality parameters and emotional parameters, generate an AI-driven highly realistic SP answer. The aforementioned triples consist of {first entity, relation, and tail entity}. Using a BiLSTM-CRF model, or a BERT model, or a BERT-BiLSTM-CRF model, different types of entities are identified from validated medical literature and case data. Entity types include diseases, symptoms, drugs, genotypes, etc., resulting in multiple entity databases for the medical history collection knowledge graph, including disease entity databases, symptom entity databases, drug entity databases, and genotype entity databases. By defining relation templates, corresponding triples are extracted from the text. For example, based on the relation template "disease-cause-gene," the triple "hypertension-genetic factors-AGT gene" is extracted from validated medical literature; based on the relation template "symptom-disease-drug," the triple "polydipsia-diabetes-insulin" is extracted from validated case data. The BiLSTM-CRF model is used to identify entities included in the text of the student's current question. The entities in the student's current question text are compared with the entities in each entity database of the medical history collection knowledge graph. Entities present in the entity databases of the student's current question text are identified as the first question entity. For example, if the first question entity is "symptoms," then all triples containing the "symptom" entity are searched from the medical history knowledge graph to obtain the first set of triples. Since the symptoms of the AI-driven highly realistic SP include "excessive thirst" and "thirst," triples containing "excessive thirst" and "thirst" are searched from the first set of triples to obtain the second set of triples. Based on the second set of triples, relevant text is searched from the generative AI large-scale language model. For the relevant text, the generative AI large-scale language model, which includes personality and emotion parameters, outputs natural language text reflecting the personality and emotions of the AI-driven highly realistic SP.

[0021] The aforementioned medical teaching and training system generates AI-driven, highly realistic patient (SP) symptoms and physiological indicators based on real cases, resulting in more stable and objective virtual patient performance. Based on this data from the AI-driven, highly realistic SP, a body model is constructed for physical examination, facilitating more realistic history-taking experience for trainees. Based on the trainee's current question, an initial range of triplets is obtained from the history-taking knowledge graph. This range is further narrowed down based on the symptoms and physiological indicators of the AI-driven, highly realistic SP, generating the AI-driven, highly realistic SP's response. This allows trainees to verify their diagnostic judgments and gain diagnostic experience based on the AI-driven, highly realistic SP's responses. This application can simulate various types of patients, including their symptoms, personality, and emotions. When trainees engage in history-taking dialogue with the AI-driven, highly realistic SP, it's like communicating with a real patient. Moreover, the AI-driven, highly realistic SP can automatically provide reasonable answers based on the trainee's questions, guiding the trainee to further understand the condition. The medical teaching and training system provided in this application offers medical learners access to medical history taking practice resources that are not limited by the number or time of real or standardized patients. It provides trainees with a large number of diverse medical history taking training opportunities at a relatively low cost, and can be adjusted and modified at any time according to teaching needs. Furthermore, the performance of virtual patients is more stable and objective, providing trainees with a more standardized training environment, which helps improve their medical history taking skills and clinical reasoning abilities.

[0022] Furthermore, the AI-driven, highly realistic SP (Special Patient) medical history acquisition medical teaching and training system also includes: The case database construction module is used to build a case database based on multiple sets of validated case data; The Medical History Collection Knowledge Graph Construction Module is used to construct a medical history collection knowledge graph based on validated medical literature and case data.

[0023] Furthermore, the case database construction module is specifically used for: The collected and verified case data are preprocessed to remove erroneous, redundant, and private data. Based on the preprocessed case data, the data is classified by symptoms, diseases, and treatment plans, and corresponding symptom indexes, disease indexes, and treatment plan indexes are established to obtain a case database.

[0024] Furthermore, the medical history collection knowledge graph construction module is specifically used for: The process involves collecting and validating medical literature and case data, cleaning and structuring this data to obtain a medical dataset. Entity recognition is performed on the medical dataset to obtain multiple entity libraries, each containing multiple entities. Entities with similarity exceeding a threshold are merged using synonyms. Based on preset relation templates and entities, relations are extracted from the medical dataset to obtain multiple triples reflecting the relationships between entities. The Neo4j graph database is used to store these triples. The method for calculating entity similarity includes: representing entities as vectors using word embedding models (such as Word2Vec or BERT), calculating the similarity between corresponding vectors of two entities using the cosine similarity formula, and using the similarity between the corresponding vectors of two entities as the similarity between the two entities.

[0025] Furthermore, the AI-driven high-fidelity SP generation module is specifically used for: Based on the diseases or symptoms involved in the teaching needs, case data related to the diseases or symptoms are screened from the case database. The case data includes the patient's symptoms, physiological indicators, and body parameters, including gender, age, height, and weight. Based on the symptoms in the case data, multiple identical or similar symptoms are obtained, and the symptoms of the AI-driven high-fidelity simulation SP are selected from these. Based on the physiological indicators and body parameters in the case data, physiological indicator ranges for different genders and age groups are determined. Based on the teaching needs, the gender and age of the AI-driven high-fidelity simulation SP are determined. Based on the gender and age of the AI-driven high-fidelity simulation SP, the physiological indicators, height, and weight of the AI-driven high-fidelity simulation SP are selected from the corresponding physiological indicator ranges for the gender and age group. Based on the height and weight of the AI-driven high-fidelity simulation SP, the body model size of the AI-driven high-fidelity simulation SP is determined or updated. Based on the symptoms and physiological indicators of the AI-driven high-fidelity simulation SP, the body model morphology of the AI-driven high-fidelity simulation SP is determined or updated. For example, if the symptoms of the AI-driven high-fidelity simulation SP include "dizziness" and the physiological indicator "anemia," then a periodic dizziness action is set for the body model morphology of the AI-driven high-fidelity simulation SP.

[0026] Furthermore, the AI-driven high-fidelity SP display module is specifically used for: The parameters of the AI-driven high-fidelity SP body model are acquired in real time. These parameters include the size and shape of the AI-driven high-fidelity SP body model. An AI-driven high-fidelity SP body model is established based on the SP's gender, age, and the model's size. The model is then adjusted according to its shape. For example, if the model's shape includes dizziness movements, it is controlled to move according to a preset dizziness movement path.

[0027] Furthermore, trainees conduct a physical examination of the AI-driven highly realistic SP body model. For example, if a trainee observes dizziness in the AI-driven highly realistic SP, the trainee will ask further questions about this symptom.

[0028] Furthermore, based on AI-driven highly realistic SP symptoms, physiological indicators, and body parameters, an AI-driven highly realistic SP state entity is obtained, including: The system searches the medical history knowledge graph for entities with the highest correlation to the symptoms of the AI-driven highly realistic SP, entities with the highest correlation to the physiological indicators of the AI-driven highly realistic SP, and entities with the highest correlation to the body parameters of the AI-driven highly realistic SP. These three types of entities are identified as AI-driven highly realistic SP state entities. AI-driven highly realistic SP state entities include the above three types of entities. For example, searching the medical history knowledge graph for AI-driven highly realistic SP symptoms yields multiple related entities. The entity similarity between the AI-driven highly realistic SP symptom entities and these related entities is calculated to obtain the correlation between the related entities and the AI-driven highly realistic SP symptoms.

[0029] Furthermore, the AI-driven highly realistic SP medical history collection and interaction module is also used for: if there is no triple corresponding to the AI-driven highly realistic SP state entity in the first set of triples, then combining the AI-driven highly realistic SP symptoms, physiological indicators, and body parameters to obtain multiple AI-driven highly realistic SP state entities; searching for triples corresponding to each AI-driven highly realistic SP state entity from the first set of triples to obtain a third set of triples; and generating an AI-driven highly realistic SP response based on the third set of triples and the AI-driven highly realistic SP personality parameters and emotional parameters. The specific process of generating an AI-driven highly realistic SP response based on the third set of triples and the AI-driven highly realistic SP personality parameters and emotional parameters can refer to the process of generating an AI-driven highly realistic SP response based on the second set of triples and the AI-driven highly realistic SP personality parameters and emotional parameters.

[0030] Furthermore, based on the second set of the triples, and combined with the AI-driven highly realistic SP personality parameters and emotion parameters, the AI-driven highly realistic SP's answers are generated, including: Calculate the relevance of each triple in the second set of triples to the student's current question; calculate the frequency of occurrence of the entity and relation of each triple in the second set of triples; determine the score of each triple in the second set of triples based on the relevance and the frequency of occurrence of the entity and relation; obtain AI-driven high-fidelity SP personality parameters and emotional parameters through the AI-driven high-fidelity SP medical history collection interaction module; establish an AI-driven high-fidelity SP agent connected to a large language model (LLM) data based on the AI-driven high-fidelity SP symptoms, physiological indicators, body parameters, personality parameters, and emotional parameters, where the large language model is a validation-based medical model. The model is trained using literature and case data, a personality language database, and an emotion database. The personality language database includes structured statements corresponding to different personalities indexed by personality parameters, and the emotion database includes structured statements corresponding to different emotions indexed by emotion parameters. The highest-scoring triplet, the student's and AI-driven high-fidelity SP's historical question-and-answer sessions, the AI-driven high-fidelity SP's personality parameters, and emotion parameters are used as context. The student's current question is used as a prompt word. These are input into the AI-driven high-fidelity SP agent. The AI-driven high-fidelity SP agent obtains the answer based on the context, prompt word, personality parameters, and emotion parameters from the large language model and feeds the answer back to the AI-driven high-fidelity SP's medical history collection and interaction module.

[0031] Calculating the relevance of each triple in the second set of triples to the student's current question includes: using an entity recognition model to identify entities in the text of the student's current question to obtain a first entity; using a relation extraction model to extract the relation predicates from the text of the student's current question to obtain a first relation predicate; and through... Determine the relevance of each triple in the second set of triples to the student's current question. The second set of triplets represents the first... i The degree of relevance between each triple and the student's current question. i This indicates the triplet index of the second set of triplets. Represents the entity weight coefficient. The second set of triplets represents the... i The maximum similarity between each entity in the triplet and the first entity. Represents the relation weight coefficient. The second set of triplets represents the... i The maximum similarity between each relation in the triplet and the predicate of the first relation. and The initial value is 0.5, and iterative optimization is performed by setting a loss function. The calculation of relation similarity includes: representing the relation of the triple and the first relation predicate as vectors using a word embedding model (such as Word2Vec or BERT); calculating the similarity between the corresponding vectors of the relation of the triple and the first relation predicate using the cosine similarity formula; and using the similarity between these two vectors as the relation similarity between the relation of the triple and the first relation predicate. This method for determining association has low time and space complexity and is suitable for large-scale knowledge graphs in the medical field.

[0032] The score for each triple in the second set of triples is determined based on the association degree of each triple, the frequency of occurrence of the entity and relation in the second set of triples, and the corresponding formula is as follows: , The second set of triplets represents the first... i The scores of the triples, where n represents the total number of triples in the second set of triples. The second set of triplets represents the first... i The degree of relevance between each triple and the student's current question. i This indicates the triplet index of the second set of triplets. The second set of triplets represents the first... i The number of times each entity in a triple appears in the second set of triples. The second set of triplets represents the first... i The number of times the relation of each triple appears in the second set of triples. This represents the number of entities in the second set of the triple. Denotes the relation number of the second set of triples. express sigmoid The function will Mapped to the interval between 0 and 1.

[0033] Furthermore, the AI-driven high-fidelity SP medical history acquisition medical teaching and training system also includes: an AI-driven high-fidelity SP physiological indicator display module, used to acquire and display multiple physiological indicators of the AI-driven high-fidelity SP in real time. Trainees can directly observe the physiological indicators of the AI-driven high-fidelity SP using this module, facilitating the diagnosis of the AI-driven high-fidelity SP.

[0034] In this embodiment, the medical teaching and training system provided by this application can provide medical learning with medical history collection practice resources that are not limited by the number or time of real or standardized patients. It can provide students with a large number of diverse medical history collection training opportunities, and the cost is relatively low. It can also be adjusted and modified at any time according to teaching needs, and the performance of virtual patients is more stable and objective.

[0035] Those skilled in the art can change the above order without departing from the scope of protection of this application.

[0036] Any modifications, additions, and equivalent substitutions made within the scope of the principles of this application shall still fall within the patent coverage of this application. "First," "Second," etc., do not indicate a sequential order, but merely distinguish between different features.

Claims

1. A medical teaching and training system based on AI-driven, highly realistic SP (spinal patient) medical history acquisition, characterized in that, The AI-driven, highly realistic SP (Specialist) medical history collection system includes: The AI-driven high-fidelity SP generation module is used to extract relevant case data from the case database according to teaching needs, determine the symptoms and physiological indicators of the AI-driven high-fidelity SP based on the case data, and determine or update the body model parameters of the AI-driven high-fidelity SP based on preset body parameters. The AI-driven high-fidelity SP display module is used to acquire the parameters of the AI-driven high-fidelity SP body model in real time and construct the AI-driven high-fidelity SP body model. The AI-driven highly realistic SP medical history collection interaction module is used to obtain the student's current question, extract entities from the medical history collection knowledge graph from the student's current question to obtain a first question entity, the medical history collection knowledge graph being constructed based on verified medical literature and case data; based on the first question entity, search for the corresponding triples from the medical history collection knowledge graph to obtain a first set of triples; based on the AI-driven highly realistic SP symptoms, physiological indicators, and body parameters, obtain an AI-driven highly realistic SP state entity; search for the triples corresponding to the AI-driven highly realistic SP state entity from the first set of triples to obtain a second set of triples; based on the second set of triples, combined with the AI-driven highly realistic SP personality parameters and emotional parameters, generate an AI-driven highly realistic SP answer.

2. The medical teaching and training system based on AI-driven high-fidelity SP medical history acquisition as described in claim 1, characterized in that, The AI-driven, highly realistic SP (spinal patient) medical history acquisition medical teaching and training system also includes: The case database construction module is used to build a case database based on multiple sets of validated case data; The Medical History Collection Knowledge Graph Construction Module is used to construct a medical history collection knowledge graph based on validated medical literature and case data.

3. The medical teaching and training system based on AI-driven high-fidelity SP medical history acquisition as described in claim 2, characterized in that, The case database construction module is specifically used for: The collected and verified case data are preprocessed, including the removal of erroneous, redundant, and private data. Based on the preprocessed case data, the data is classified by symptoms, diseases, and treatment plans, and corresponding symptom indexes, disease indexes, and treatment plan indexes are established to obtain a case database.

4. The medical teaching and training system based on AI-driven high-fidelity SP medical history acquisition as described in claim 2, characterized in that, The medical history collection knowledge graph construction module is specifically used for: Collect and validate medical literature and case data, clean and structure the medical literature and case data to obtain a medical dataset; perform entity recognition on the medical dataset to obtain multiple entity libraries, each of which includes multiple entities; merge synonyms for entities whose similarity exceeds a threshold; extract relations from the medical dataset based on preset relation templates and entities to obtain multiple triples reflecting the relations between entities.

5. The medical teaching and training system based on AI-driven high-fidelity SP medical history acquisition as described in claim 1, characterized in that, The AI-driven high-fidelity SP generation module is specifically used for: Based on the diseases or symptoms involved in the teaching needs, case data related to the diseases or symptoms are screened from the case database. The case data includes the patient's symptoms, physiological indicators, and body parameters, including gender, age, height, and weight. Based on the symptoms in the case data, multiple identical or similar symptoms are obtained, and the symptoms of AI-driven high-simulation SP are selected from these multiple identical or similar symptoms. Based on the physiological indicators and body parameters in the case data, physiological indicator ranges for different genders and age groups are determined. Based on teaching needs, the gender and age of the AI-driven high-fidelity SP are determined. Based on the gender and age of the AI-driven high-fidelity SP, the physiological indicators, height, and weight of the AI-driven high-fidelity SP are selected from the physiological indicator range of the corresponding gender and age group. Based on the height and weight of the AI-driven high-fidelity SP, the body model size of the AI-driven high-fidelity SP is determined or updated. Based on the symptoms and physiological indicators of the AI-driven high-fidelity SP, the body model morphology of the AI-driven high-fidelity SP is determined or updated.

6. The medical teaching and training system based on AI-driven high-fidelity SP medical history acquisition as described in claim 5, characterized in that, The AI-driven high-fidelity SP display module is specifically used for: The parameters of the AI-driven high-fidelity SP body model are acquired in real time, including the size and shape of the AI-driven high-fidelity SP body model. The AI-driven high-fidelity SP body model is established based on the gender, age, and size of the AI-driven high-fidelity SP body model. The AI-driven high-fidelity SP body model is then adjusted based on its shape.

7. The medical teaching and training system based on AI-driven high-fidelity SP medical history acquisition as described in claim 1, characterized in that, Based on AI-driven highly realistic SP symptoms, physiological indicators, and body parameters, obtain AI-driven highly realistic SP state entities, including: The entity with the highest correlation to the symptoms of the AI-driven highly realistic SP is found in the medical history collection knowledge graph. The entity with the highest correlation to the physiological indicators of the AI-driven highly realistic SP is found in the medical history collection knowledge graph. The entity with the highest correlation to the body parameters of the AI-driven highly realistic SP is found in the medical history collection knowledge graph. The above three types of entities are identified as the state entities of the AI-driven highly realistic SP.

8. The medical teaching and training system based on AI-driven high-fidelity SP medical history acquisition as described in claim 1, characterized in that, The AI-driven highly realistic SP medical history collection and interaction module is further configured to: if there is no triple corresponding to the AI-driven highly realistic SP state entity in the first set of triples, then combine the AI-driven highly realistic SP symptoms, physiological indicators, and body parameters to obtain multiple AI-driven highly realistic SP state entities; search for triples corresponding to each AI-driven highly realistic SP state entity from the first set of triples to obtain a third set of triples; and generate an AI-driven highly realistic SP response based on the third set of triples and the AI-driven highly realistic SP personality parameters and emotional parameters.

9. The medical teaching and training system based on AI-driven high-fidelity SP medical history acquisition as described in claim 1, characterized in that, Based on the second set of the triplet, and combined with the AI-driven highly realistic SP personality parameters and emotion parameters, the AI-driven highly realistic SP's answers are generated, including: Calculate the relevance of each triplet in the second set of triplets to the student's current question; calculate the frequency of occurrence of the entity and relation of each triplet in the second set of triplets; determine the score of each triplet in the second set of triplets based on the relevance and the frequency of occurrence of the entity and relation; obtain AI-driven high-fidelity SP personality parameters and emotional parameters through the AI-driven high-fidelity SP medical history collection interaction module; establish an AI-driven high-fidelity SP intelligent agent connected to the large language model data based on the AI-driven high-fidelity SP symptoms, physiological indicators, body parameters, personality parameters, and emotional parameters. The language model is trained based on validated medical literature and case data, a personality language database, and an emotion database. The personality language database includes structured statements corresponding to different personalities, and the emotion database includes structured statements corresponding to different emotions. The highest-scoring triplet, the student's and AI-driven high-fidelity SP's historical question-and-answer sessions, the AI-driven high-fidelity SP's personality parameters, and emotion parameters are used as context. The student's current question is used as a prompt word and input into the AI-driven high-fidelity SP agent. The AI-driven high-fidelity SP agent obtains the answer based on the context, prompt word, personality parameters, and emotion parameters from the large language model and feeds the answer back to the AI-driven high-fidelity SP's medical history collection and interaction module.

10. The medical teaching and training system based on AI-driven high-fidelity SP medical history acquisition as described in claim 1, characterized in that, The AI-driven, highly realistic SP (spinal patient) medical history acquisition medical teaching and training system also includes: The AI-driven high-simulation SP physiological indicator display module is used to acquire and display multiple physiological indicators of the AI-driven high-simulation SP in real time.

Images