Intelligent pediatric nursing teaching AI inquiry system and method based on digital background

Abstract

The invention discloses an intelligent pediatric nursing teaching AI inquiry system and method based on a digital background, and relates to the crossing field of education informatization and artificial intelligence. The system comprises a front-end interaction module, a voice processing module, a teaching knowledge base module, an intelligent question and answer engine module, a teaching evaluation and feedback module, a background management module, a data analysis module and a virtual simulation operation linkage module, and the teaching knowledge base integrates on-duty competition certificates and traditional Chinese medicine pediatric characteristic knowledge and supports knowledge graph and vectorization retrieval; the intelligent question and answer engine adopts a retrieval enhancement generation technology and is combined with a standardized inquiry dialogue state machine to guarantee the answer accuracy. The method comprises the steps of case selection, multi-modal interaction, intelligent question and answer, process evaluation, virtual simulation linkage, adaptive recommendation and the like. According to the pediatric nursing teaching system, simulated inquiry training, refined process evaluation and teaching ecological closed-loop optimization are realized, the professionality, safety and individuation of pediatric nursing teaching are improved, and core ability cultivation of nursing talents is assisted.

Description

Technical Field

[0001] This invention relates to the field of interdisciplinary application of educational informatization and artificial intelligence technologies, specifically to an intelligent teaching system and method, and more particularly to an intelligent pediatric nursing teaching AI consultation system and method that integrates digital human technology, intelligent voice interaction, knowledge enhancement generation, teaching evaluation and data analysis, specifically for pediatric nursing courses in higher vocational education under the background of digital education. Background Technology

[0002] With the deepening of the digital education strategy and the rapid development of artificial intelligence technology, blended learning models and intelligent teaching tools have been widely used in vocational education, especially in the fields of medical and nursing education. Currently, utilizing online platforms, virtual simulations, and intelligent question-and-answer systems to assist teaching has become an important means to improve teaching effectiveness and alleviate the pressure on practical training resources.

[0003] Currently available online medical consultation or teaching support systems mostly use text chatbots or pre-recorded videos for knowledge transfer. While "digital humans" integrating voice interaction and simple animated avatars have emerged, their core functions remain limited to information broadcasting or fixed-process question-and-answer sessions. These systems lack a structured knowledge system deeply integrated with professional nursing education. Their dialogue logic is generic and cannot simulate the complex and dynamic nursing assessment and communication scenarios in real clinical settings (such as pediatric consultations). When faced with open-ended, unstructured student questions, systems based on generic language models are prone to content illusions, providing answers that do not conform to medical standards or the syllabus, posing a risk of misleading students and failing to meet the stringent accuracy and safety requirements of high-standard nursing education.

[0004] Many so-called teaching software programs are merely digital displays of knowledge points, failing to effectively integrate "on-the-job training, curriculum, competition, and certification." They cannot organically incorporate the core competency requirements of clinical positions, the scoring criteria for nursing skills competitions, and the assessment points for professional skill level certificates into an interactive training and assessment system. When students practice, they lack clear professional competency benchmarks and standardized operational procedures, resulting in a disconnect between the training process and actual clinical needs and authoritative evaluation systems. This leads to weak training relevance and makes it difficult to quantify teaching effectiveness.

[0005] Traditional online learning systems rely heavily on post-class multiple-choice tests or subjective teacher evaluations for assessment, failing to provide formative and process-oriented assessments of students' core competencies such as clinical reasoning, communication skills, and on-the-spot decision-making abilities. Current technologies struggle to capture and analyze in real-time whether students' logic is complete, their vocabulary professional, and their communication empathetic during simulated consultations, thus hindering the provision of immediate, specific, and data-driven personalized feedback and improvement suggestions. Teaching remains at a superficial level of "practice-answer checking," lacking a deep "assessment-feedback-optimization" closed loop.

[0006] Most teaching platforms only record superficial data such as login frequency and video viewing time, lacking the ability to effectively mine and analyze deeper interactive data generated during the teaching process (such as dialogue text, operation sequences, and decision paths). Teachers cannot quickly and accurately identify weak points in teaching content or common cognitive misconceptions among students from group data, making teaching optimization rely on experience rather than evidence, thus hindering the continuous iteration and upgrading of "data-driven" precision teaching and blended learning models.

[0007] Current training systems often separate clinical consultation training from practical skills training. The consultation system only focuses on "asking," while the virtual simulation software only focuses on "doing," lacking organic integration between the two. Students cannot experience the natural flow from communication and assessment to execution of actions during a complete nursing care process. Furthermore, the system's interaction mode is simplistic, failing to effectively integrate multimodal inputs such as voice, text, and images (e.g., photos of children's rashes), limiting the realism and complexity of the training scenarios.

[0008] The information disclosed in this background section is intended only to enhance the understanding of the overall background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0009] The purpose of this invention is to provide an AI-based pediatric nursing teaching system and method with a digital background.

[0010] To achieve the above objectives, the present invention provides the following solution:

[0011] A digital-based intelligent pediatric nursing teaching AI consultation system includes: The front-end interaction module is used to render and display the doctor's digital human image on the user terminal, receive the user's voice or text input, and synchronously play the voice response generated by the system and drive the mouth animation of the digital human image. The voice processing module is connected in communication with the aforementioned front-end interaction module and is used to convert the user's voice input into text and convert the system-generated text questions and answers into a voice stream. The teaching knowledge base module stores structured pediatric nursing teaching knowledge data, which includes at least: pediatric nursing curriculum standards reconstructed based on job competency requirements, typical cases related to nursing skills competition standards, key points of the evaluation standards for the professional skill levels of infant and toddler development guides, and characteristic knowledge and operational norms of traditional Chinese medicine pediatric nursing; the knowledge data is organized into knowledge graphs and / or vectorized representations, supporting semantic retrieval; The intelligent question-answering engine module is communicatively connected to both the speech processing module and the teaching knowledge base module. It receives the converted user query text, performs enhanced retrieval based on the teaching knowledge base module, and generates text responses that conform to pediatric nursing teaching standards and medical accuracy by combining preset dialogue management strategies. The dialogue management strategy includes a dialogue state machine based on a standardized patient consultation process, which guides users through the complete nursing process from nursing assessment and diagnosis to nursing interventions and health guidance. The teaching evaluation and feedback module is communicatively connected to the front-end interaction module and the intelligent question-answering engine module. It is used to record and analyze the user's interaction process data with the system in real time. The interaction process data includes at least: the completeness of the questioning logic, the accuracy of keyword extraction, the standardization of communication language, and the matching degree with the standard answers in the teaching knowledge base. Based on the interaction process data, a multi-dimensional ability evaluation report and personalized learning suggestions are generated. The backend management and data analysis module is used by teachers to manage teaching cases, update the knowledge base, view student learning trajectories, and conduct macro-analysis of teaching effectiveness based on the interaction data of all students, in order to support the continuous optimization of teaching models.

[0012] Optionally, the teaching knowledge base module also includes a dynamically updated typical case library containing multiple standardized pediatric nursing consultation scenarios. Each scenario corresponds to a structured FAQ template, which includes a standard set of questions, allowed variations of similar questions, and standard answers or answer generation logic that conform to nursing standards. During the dialogue, the intelligent question-answering engine module prioritizes matching and calling the FAQ template based on the current dialogue state to provide standardized and assessable interactive training.

[0013] Optionally, the assessment model of the teaching assessment and feedback module introduces a quantitative scoring function to calculate the score of a single consultation exercise. S :

[0014] in, C logic The score represents the completeness of the consultation logic, which is calculated by comparing the coverage of the user's question sequence with the key nodes of the standard consultation process. Ckeyword The score represents the accuracy of keyword extraction, calculated by analyzing the matching degree between key medical terms extracted from user questions and a standard keyword list; C communication The score represents the standardization of communication language, and evaluates the politeness, clarity, and empathy of user statements based on a natural language processing model. C accuracy The score represents the answer matching degree. When a user answers questions in the system while playing the role of a nurse, the similarity between their answer and the standard answer in the knowledge base is calculated. w 1, w 2, w 3, w 4 represents the weighting coefficient for each sub-item score, and w 1+ w 2+ w 3+ w 4=1, and the weighting coefficient can be adjusted by the teacher according to the teaching focus.

[0015] Optionally, the retrieval enhancement generation method adopted by the intelligent question answering and engine module specifically includes: (a) Vectorize the user query text to obtain the query vector. V q ; (b) Calculate in the vector database of the teaching knowledge base module. V q Based on the similarity with all knowledge fragment vectors, retrieve the Top-K most relevant knowledge fragments. D 1, D 2, ..., D K ; (c) Combine the user's query text with the relevant knowledge fragments retrieved. D 1, D 2, ..., D K The prompt words, along with the current dialogue history, together constitute the prompt words, which are then input into the large language model; (d) The large language model generates the final response text based on the prompt words. The response text must reference or follow the content of the retrieved relevant knowledge fragments and conform to the context of nursing communication.

[0016] Optionally, the system further includes a virtual simulation operation linkage module. When the answer generated by the intelligent question-and-answer engine module or the suggestion from the teaching assessment and feedback module involves a specific nursing operation, the virtual simulation operation linkage module is triggered, presenting a corresponding three-dimensional virtual simulation operation interface on the user terminal, guiding the user to practice the simulated operation, and feeding back the operation result data to the teaching assessment and feedback module.

[0017] The front-end interaction module supports multimodal input, allowing users to upload partial photos or simple line drawings of the child in addition to voice and text. The intelligent question-answering engine module integrates a multimodal large model, which can perform comprehensive analysis and answer by combining image information with the text dialogue context.

[0018] Optionally, the backend management and data analysis module provides a function for analyzing teaching weaknesses based on group learning data. This function is implemented through the following steps: (a) Collect data on all students' interaction processes and assessment scores on specific teaching cases or knowledge points; (b) For students whose scores are below the threshold, cluster analysis is performed to identify common error types or missing consultation steps; (c) Calculate the overall error rate for a specific knowledge point. E k , ,in N error,k For knowledge points k The number of students who made the error N total,k To access knowledge points k Total number of student visits; (d) The analysis results are presented to teachers in the form of visual charts, and knowledge points with high error rates are automatically marked, indicating that teaching needs to be strengthened or the knowledge base content needs to be optimized.

[0019] A digital-based intelligent pediatric nursing teaching AI-assisted consultation method includes the following steps: S1: The system starts up, the front-end interactive module loads and displays the designated doctor's digital avatar and initial greeting; S2: Receive the user's request for a consultation practice and determine the topic of the pediatric nursing case for the practice; S3: Based on the selected case, initialize the dialogue state machine. The intelligent question-answering engine module generates guiding questions or acts as a patient / family member to make statements based on the current state and in conjunction with the teaching knowledge base module. S4: Receives user's voice or text input, converts it through the voice processing module, processes it through the intelligent question-answering engine module, combines knowledge retrieval and dialogue history to generate answers that conform to nursing standards, and provides feedback through the front-end interaction module in the form of digital human voice and animation. S5: The teaching assessment and feedback module monitors and records the interaction process between steps S3 and S4 in real time, and extracts assessment features; S6: Repeat steps S3 to S5 until the dialogue state machine determines that the consultation process has been completed or the user has actively ended it. S7: The teaching assessment and feedback module generates and pushes a multi-dimensional assessment report and learning suggestions based on the complete interaction record. S8: The backend management and data analysis module summarizes the current and historical interaction data and updates the group learning analysis report.

[0020] Optionally, in step S3, the system supports a training mode that integrates job skills, courses, competitions, and certifications. Teachers can configure the focus of each exercise in the background: when the focus is on job skills, the dialogue process and assessment standards are closely aligned with the actual clinical job requirements; when the focus is on competitions, the system simulates the standardized patient site assessment scenario and scoring rules of nursing skills competitions; when the focus is on certifications, the question-and-answer content and assessment points revolve around the assessment points of professional skill level certificates such as infant and toddler development guides.

[0021] Optionally, the method further includes an adaptive learning path recommendation step: Based on the evaluation results of users' multiple historical practice sessions from the teaching assessment and feedback module, a user ability profile is constructed to identify their strengths and weaknesses. Based on the group's weaknesses information provided by the backend management and data analysis modules, the system intelligently recommends the next most suitable training case or knowledge point topic to the current user, with recommendation priority. P rec The calculation factors include: the user's historical scores for the corresponding knowledge point. S hist The group error rate of this knowledge point E k And the importance and weight of this knowledge point in the overall curriculum system. I k The calculation formula is: ; The recommendation results will be presented to the user through the front-end interaction module.

[0022] This invention addresses the core problems of existing intelligent teaching tools for pediatric nursing, such as insufficient professionalism, superficial evaluation, isolated data, and fragmented training scenarios. It proposes a closed-loop intelligent teaching system and method that deeply integrates educational theory and artificial intelligence technology and is geared towards cultivating professional skills.

[0023] The core concept of this invention lies in constructing an integrated intelligent teaching ecosystem. This ecosystem is built upon a professional and structured teaching knowledge base, utilizes anthropomorphic and multimodal interaction as its front end, employs retrieval-enhanced generation technology as its intelligent core, is driven by process-oriented and multi-dimensional teaching assessment, and is optimized by data-driven group analysis and personalized recommendations. This system is not simply a stacking of technologies, but rather a digital reshaping of the entire teaching process, centered on a student-centered and competency-oriented teaching philosophy.

[0024] Specifically, the system of the present invention achieves the above concept through the following cooperating modules: A highly realistic and configurable front-end interaction layer: This front-end interaction module not only renders the digital human image of the doctor, but more importantly, achieves millimeter-level synchronization of voice, animation, and text. Through emotional speech synthesis technology, it imbues the digital human with a caring and calm tone befitting the nursing profession, greatly enhancing the immersiveness and trustworthiness of the interaction. This is crucial for stimulating students' training interest and simulating a realistic doctor-patient communication atmosphere. This module supports responsive design, ensuring a consistent interactive experience across multiple terminals such as PCs and mobile devices.

[0025] A teaching knowledge base that deeply integrates specialization and structure: The teaching knowledge base module of this invention is the fundamental difference between it and general question-and-answer systems, and its innovation is reflected in: Source integration: The knowledge data organically integrates national teaching standards, industry job specifications, skills competition requirements, vocational skills certificate examination points, and characteristic TCM pediatric nursing knowledge, forming a unified knowledge system that integrates "job, course, competition, and certificate".

[0026] The organization exhibits a dual nature: knowledge is organized in the form of knowledge graphs to depict the complex relationships between entities such as diseases, symptoms, nursing measures, and drugs, supporting reasoning; at the same time, it is vectorized with high precision to convert unstructured clinical cases, nursing records, etc., into semantic vectors, providing support for efficient semantic retrieval.

[0027] Scenario-based templates: The built-in FAQ template library is an abstraction of standardized nursing procedures. Each template defines the standard dialogue path, key question nodes, and standardized answers in a specific clinical scenario (such as "diagnosis of diarrhea in infants and young children"), providing a "script" for controllable and assessable training.

[0028] A secure, controllable, and scenario-adaptive intelligent question-answering engine: This module adopts a "retrieval-enhanced generation" technology approach, and its innovative workflow is as follows: Precise retrieval: First, vectorization technology is used to accurately locate the most relevant knowledge fragments to the current conversation from a massive professional knowledge base, ensuring that the "raw materials" for the answer are correct and authoritative.

[0029] Controlled Generation: The retrieved knowledge fragments, the current dialogue context, and pre-designed system prompts such as "nursing staff identity instructions" and "communication style requirements" are then input into the large language model. This method strictly constrains the generation boundaries of the large model, ensuring that its responses are based on retrieved professional knowledge and conform to the context and norms of nursing communication. This fundamentally eliminates "illusions" and arbitrariness, guaranteeing the safety of teaching.

[0030] State Management: Integrating a dialogue state machine based on a standardized patient consultation process, the system can "understand" the stage of the dialogue (such as information collection, differential diagnosis, and health education), thereby proactively guiding or responding and simulating a complete nursing assessment process with logical progression, rather than an isolated question and answer.

[0031] This invention's assessment module breaks through the limitations of traditional methods that only consider the correctness of the final answer, achieving a process-oriented evaluation of core nursing competencies. Its core lies in establishing a quantifiable and configurable assessment model. This model automatically analyzes whether the student's consultation logic is complete (whether any key steps are missing), whether professional terminology is used accurately, and whether the communication style reflects humanistic care, and then provides a comprehensive score. The assessment result is no longer a simple score, but generates a structured assessment report that clearly identifies the student's strengths and areas for improvement, and provides targeted learning suggestions (such as "suggest strengthening practice in asking questions about 'dehydration assessment in children with diarrhea'"), thus realizing a shift from "scoring" to "empowerment."

[0032] A data-driven backend management platform supporting teaching and research decisions: The core value of the backend module lies in transforming the "process-related big data" generated during the teaching process into "evidence" for teaching optimization. It not only manages resources and tracks individual learning trajectories, but also uses algorithms such as cluster analysis and error rate calculation to identify weaknesses in teaching at the group level (for example, finding that most students scored low on questions related to "pediatric medication dosage calculation"). This data-driven insight enables teachers to make precise teaching interventions, optimize instructional design and knowledge base content, and also provides rich data support for teaching research, forming a complete closed loop of "teaching-assessment-data-optimization".

[0033] Modular Expansion and Integrated Training Capabilities: The system adopts a loosely coupled modular design, offering excellent scalability. For example, through the "virtual simulation operation linkage module," the system can seamlessly initiate virtual simulation exercises when the consultation logic naturally extends to nursing operations (such as "physical cooling" after consultation), achieving an organic link between "communication" and "skills" training. Furthermore, the system supports multimodal input (such as images), providing more realistic training scenarios for identifying physical symptoms such as rashes, further enhancing the system's comprehensive training capabilities.

[0034] The method of this invention embodies the collaborative operation of the above-mentioned system, and by supporting the configuration of the integrated training mode of "job, course, competition and certification" and the adaptive learning path recommendation based on data profile, it realizes personalized training under large-scale teaching, effectively improving the efficiency and quality of pediatric nursing talent training.

[0035] Compared with the prior art, the present invention has the following beneficial effects: This greatly enhances the professionalism, safety, and authenticity of teaching: by constructing a professional knowledge base that deeply integrates the requirements of "job training, course preparation, competition, and certification," and by using retrieval-enhanced generation technology to strictly constrain AI responses, it ensures that all teaching content and interactive feedback comply with medical norms and nursing standards, eliminates misleading information, and provides students with a safe, reliable, and highly realistic digital clinical training environment.

[0036] It enables refined and process-oriented evaluation of advanced nursing competencies: breaking through the limitations of traditional tests that can only assess knowledge memorization, it uses a quantitative assessment model to automatically analyze, score, and generate personalized feedback reports on students' core professional qualities such as clinical thinking logic, communication skills, and humanistic care, making teaching evaluation more scientific, comprehensive, instructive, and strongly supporting the implementation of formative assessment.

[0037] It has established a data loop connecting "teaching, learning, assessment, and research," enabling continuous optimization of teaching models: the system deeply mines and analyzes process data in teaching, which can not only serve personalized learning recommendations for students, but also provide teachers with empirical evidence for group learning analysis and weak links in teaching, promoting the shift of teaching decisions from experience-driven to data-driven, and facilitating the precise iteration of blended teaching models and the in-depth development of nursing education research.

[0038] It promotes the organic integration of theoretical teaching and skills training, as well as different training modes: through virtual simulation operation linkage and configurable training modes of "job, course, competition and certificate", the system breaks down the barriers between communication training and practical training, and between daily teaching and competition preparation in traditional teaching, and provides an integrated and coherent comprehensive ability training platform that is more in line with the actual clinical work scenario.

[0039] It enhances the immersive, interactive, and personalized learning experience: Based on the multimodal interaction of the emotional digital human, it significantly improves students' learning interest and participation; at the same time, the system intelligently recommends learning paths based on individual ability profiles and learning data, realizing personalized learning for each individual and effectively improving learning efficiency and autonomy.

[0040] It has high scalability and application flexibility: The system adopts a modular architecture, which makes it easy to integrate new teaching cases, knowledge content (such as nursing guidelines for emerging diseases), or expand new training modalities (such as VR devices). It can quickly adapt to the needs of nursing discipline development and teaching reform, and has good promotion value and long life cycle. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 A simplified structural diagram of an AI-based intelligent pediatric nursing teaching system with a digital background, provided in an embodiment of the present invention.

[0043] Figure 2 This is a schematic diagram of the interface of the AI-based pediatric nursing teaching system and method based on a digital background, provided in an embodiment of the present invention.

[0044] Figure 3 This is a schematic diagram of the user interface of the AI ​​consultation system and method provided in this embodiment of the invention in the Zhihuishu shared course teaching platform. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] The purpose of this invention is to provide an AI-based pediatric nursing teaching system and method with a digital background.

[0047] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0048] Example 1: Intelligent consultation teaching process based on infant diarrhea cases: This embodiment uses a typical pediatric nursing teaching case—"Nursing Assessment and Health Guidance for Infants with Diarrhea"—as an example to detail the implementation process of the system and method of the present invention. The deployment architecture of this system may include: a user browser / mobile application as the front end, a cloud server cluster hosting core business modules (speech processing, intelligent question-answering engine, knowledge base data analysis), and data interaction via API. Figure 1-3 As shown.

[0049] 1. System preparation and knowledge base construction: Teachers can create and configure teaching cases on "infant diarrhea" through the teacher-side interface of the back-end management and data analysis module.

[0050] Knowledge base construction: Teachers first enter structured knowledge into the "Teaching Knowledge Base Module".

[0051] Job Standard Integration: Enter the chapter on "Pediatric Diarrhea Care" from the "Pediatric Nursing" curriculum standard, and clarify the knowledge points that need to be mastered (such as dehydration assessment and fluid replacement principles).

[0052] Competition and certification standards are integrated: the scoring standards for infant and toddler health assessment in the "National Nursing Skills Competition" are linked to the assessment points for family care of diarrhea in the "Infant and Toddler Development Instructor" professional skill level certificate.

[0053] Integration of Traditional Chinese Medicine Features: Includes TCM pediatric nursing knowledge, such as the key points of diagnosis for "diarrhea due to spleen deficiency in children" and the operation standards and indications for pediatric massage (such as tonifying the spleen meridian and abdominal massage).

[0054] Knowledge graph construction: Establish entity relationships, such as: "Infant diarrhea" -- leads to → "dehydration" -- divided into → "mild, moderate, severe"; "Oral rehydration salts" -- used for treatment → "mild dehydration" -- usage and dosage are → "XXX".

[0055] Vectorization: All the above textual knowledge (including treatment guidelines, nursing routines, and typical case descriptions) is converted into high-dimensional vectors through the model and stored in a vector database.

[0056] FAQ Template Creation: Create an "Infant Diarrhea Consultation" template in the "Typical Case Library".

[0057] Standard question set: [“How old is your baby?”, “How many days has your baby had diarrhea?”, “How many times a day does your baby have a bowel movement? What is the consistency of the stool (watery, like egg drop soup)?”, “Does the stool have a sour smell or contain pus or blood?”, “How is the amount of urine compared to usual?”, “How is your baby’s spirit and appetite?”, “Does your baby have a fever or vomiting?”]

[0058] Similar question variations: Set variations for "How many days have you had diarrhea?" ["How long have you had diarrhea?", "How long have you had diarrhea symptoms?"] to enhance the system's ability to understand natural spoken language.

[0059] Standard answer / logic: The system is given a pre-set standard answer to play the role of a "family member". For example, for the question "How many days has the diarrhea lasted?", the answer logic is: randomly select from "1-2 days", "3 days", "more than 3 days" and corresponding to different subsequent disease simulation paths.

[0060] 2. Student-side interaction and intelligent consultation process: Students (users) access the system through a browser. The front-end interactive module loads and displays a digital image of a female doctor in a nurse's uniform with a friendly expression, accompanied by a voice greeting: "Hello, please select the case for this exercise." Steps S1 & S2: Students select "Infant Diarrhea" from the case list. After system confirmation, the digital avatar switches to a clinic background and says, "Now, as the on-duty nurse, please conduct a nursing assessment for this baby with diarrhea. I am the baby's mother, and I will cooperate with you." Step S3: The system initializes the "Diarrhea Consultation" dialogue state machine, with the current state being "Collecting Basic Information". The intelligent question-answering engine generates a guiding statement based on the FAQ template: "My baby's name is Beibei, he is 4 years old, and he has had diarrhea since the day before yesterday." The digital human avatar simultaneously makes a concerned expression and mouths.

[0061] Steps S4 & S5 (First Round of Interaction): Student input: The student asks aloud: "Hello, Beibei's mom, how many times a day does Beibei poop?" (voice).

[0062] Voice processing: The voice processing module converts speech into text using the ASR service: "Hello, Beibei's mom, how many times a day does Beibei poop?" Intelligent question answering: The intelligent question answering engine receives text.

[0063] Search enhancement: The query text was vectorized and searched in a vector database, revealing a high correlation with the FAQ template "How many times a day do you have a bowel movement?" and its variations. Additionally, fragments related to "correlation between bowel movement frequency and dehydration level" were retrieved from the knowledge base.

[0064] State Recognition and Response Generation: The dialogue state machine identifies this question as belonging to the "Assess the Severity of Diarrhea" node. The intelligent question-answering engine calls the FAQ template and combines it with retrieved knowledge to generate a response text that matches the "family member's" identity: "It happens six or seven times a day, all watery and a little yellow. I think I can even see undigested rice grains." This response is one of the preset standard answers and foreshadows the subsequent "nutritional questions."

[0065] Emotionalized output: The generated text is converted into a speech stream with anxiety and worry through a TTS service, and a corresponding phoneme sequence is generated to drive the digital human's mouth animation. The digital human delivers the above answer in an anxious tone and expression.

[0066] Teaching assessment record: The teaching assessment and feedback module records in real time: the student asked the key question "number of bowel movements", the logical node "assess the severity of diarrhea" was covered, the keywords "number of bowel movements" and "watery stool" were correctly extracted, and the communication language was polite (using "hello").

[0067] Steps S3-S6 (subsequent rounds of interaction): The student continues to ask: "Does the stool smell bad or contain blood?" (covering the "Differential Diagnosis" node), "Is the baby's urination normal? How is the baby's spirit?" (covering the core "Dehydration Assessment" node). The system dynamically generates corresponding answers based on the progression of the dialogue state machine and the FAQ template. If the student misses a key question (such as "urination status"), the dialogue state machine can proactively prompt through a digital human after a certain number of rounds: "Nurse, I noticed that the baby hasn't had their diaper changed enough today, is that a problem?"

[0068] Step S6 (End of Consultation): After the student inquires about key information such as signs of dehydration, diet, and mental state, the dialogue state machine determines that the core information has been collected and can proceed to the "Health Guidance" stage. The student might say, "Based on what you've said, Beibei may have mild diarrhea. The most important thing right now is to prevent dehydration..." The system can respond simply to this, and finally, the digital human summarizes: "Thank you for the nurse's detailed explanation and guidance. I understand." Step S7 (Assessment and Feedback): The teaching assessment and feedback module immediately starts analysis the moment the consultation ends.

[0069] Quantitative Scoring: The evaluation model was invoked. c_logic: Analysis of the student's question sequence covered key nodes such as "basic information," "diarrhea characteristics," "accompanying symptoms," and "dehydration assessment," achieving a high completeness score. c_keyword: Keywords such as "frequency of bowel movements," "characteristics," "urination," and "mental state" were extracted, showing high matching accuracy. c_communication: NLP model analysis showed that the student's statements were mostly interrogative, polite, and professional, with moderate empathy (e.g., no proactive expression of comfort). c_accuracy: When the system asked, "Given your baby's condition, what precautions should be taken for home care?", the student's answer was compared with the standard answer in the knowledge base.

[0070] Report Generation: The total score S=86 is calculated based on weights (e.g., w1=0.4, w2=0.3, w3=0.2, w4=0.1). The system generates an evaluation report: "This consultation was logically sound (Excellent), the terminology was accurate (Excellent), and the communication was professional, but empathy could be improved (Good). Suggestions: 1. After inquiring about the symptoms, add a sentence like 'Please don't worry too much, we'll work together to find a solution' to show care. 2. Further inquire, 'Does your baby have any history of allergies to any medications or foods?' Click here to enhance your learning on the topic of 'Health Education for Children with Diarrhea'." Virtual simulation linkage (trigger): The assessment report also suggests: "If you need to practice preparing oral rehydration salts, you can enter the simulation operation." After the student clicks, the virtual simulation operation linkage module is activated, displaying a 3D virtual operating table, measuring cup, and oral rehydration salts, guiding the student to practice virtual preparation according to the standard procedure. The accuracy and time of operation are recorded and fed back.

[0071] Backend data analysis and personalized recommendations: Step S8 (Group Data Analysis): The backend management and data analysis module summarizes the data of all students who have completed this case.

[0072] Calculate the population error rate E K Error rate E found in the knowledge point of "dehydration degree assessment (mild / moderate)" K Up to 35%.

[0073] Cluster analysis: Analysis of the low-scoring student group revealed that a common error was the failure to strongly associate "significantly reduced urine output" with "moderate dehydration".

[0074] Visualization and Hints: On the teacher's dashboard, the "Dehydration Assessment" knowledge point is highlighted in red. Based on this, the teacher decides to focus on explaining this knowledge point in the next lesson and considers optimizing the explanation and FAQ template for this section in the knowledge base.

[0075] Adaptive learning path recommendation: Student A's historical data shows their score (S) in the "Medication Guidance" module. hist The error rate in "drug dosage calculation" was low, while population data showed an error rate of low. k The importance weight of this knowledge point in the course is high (I). k High. The system calculates and recommends priority P. rec The score was relatively high, so after student A logged in, the front end prominently displayed: "We intelligently recommend the next exercises for you: 'Simulated Medical Consultation for Commonly Used Pediatric Drug Dosage Calculation' and 'Simulated Nursing Care Exercise for Childhood Diarrhea'." Details are as follows: (II) Simulated consultation for calculating dosages of commonly used pediatric drugs: 1. Case Overview: This case study is designed around "medication guidance for children with fever and cough". It focuses on training students to accurately calculate the dosage of commonly used pediatric drugs (such as ibuprofen suspension and ambroxol oral solution) based on the child's age, weight, symptoms and other information, and to complete the complete communication process for medication guidance.

[0076] 2. Knowledge base construction and configuration: Job integration: Based on the "Principles of Calculating Medication for Children" in "Pediatric Nursing" and the "Prescription Management Measures", "Calculating Dosage by Weight" is identified as the core competency.

[0077] Integration of competition and certification: The scoring item of "Medication Safety and Communication" in the Nursing Skills Competition and the assessment point of "Guidance on Home Medication Use for Common Diseases" in the "Infant and Toddler Development Guide" certificate are linked.

[0078] Knowledge graph construction: Entities include "drug name", "dosage form", "weight range", "dosage formula", "dosage interval", "contraindications", etc., and establish "drug-dosage-age / weight" association rules.

[0079] FAQ Template Design: Standard set of questions: "How old is your child? How much does he / she weigh?" "What is his / her fever?" "Does he / she have a cough or runny nose?" "Does he / she have any history of drug allergies?" "Has your child used similar medications before?" The logic of the standard answer: The system acts as a family member, providing information such as the child's age (e.g., 3 years old), weight (14kg), body temperature (38.5℃), and no history of allergies.

[0080] Dosage calculation trigger point: After the student has asked for basic information, the system can proactively ask: "Nurse, how should I take this ibuprofen suspension? How much should I take at once?" 3. Example of interaction flow: Student: "What's your baby's temperature now? Does your baby have any other discomfort?" System (digital human playing the role of parent): "Current temperature is 38.6℃, has a slight cough, but is otherwise in good spirits." Student: "How much does your child weigh? Does he / she have any drug allergies?" System: "Weight 14 kg, never heard of any drug allergies." The system proactively guided the user: "So, how should this ibuprofen suspension be taken?" Student (who needs to calculate and answer): "Based on weight, the recommended dose is 5-10mg / kg. Your child can use 70-140mg, which is equivalent to 3.5-7ml. It is recommended to give 5ml first for observation." System evaluation points: Do you ask about weight and allergy history? Is the dosage calculation correct? Does it specify the dosing interval, maximum daily dose, and precautions (such as taking it after meals)?

[0081] 4. Evaluation and Feedback: The teaching assessment module automatically calculates the dosage accuracy score and generates a report based on communication standardization and logical integrity.

[0082] Virtual simulation linkage: If a student makes a calculation error, the system can prompt: "Do you want to enter the 'Drug Dosage Calculation Simulator' for specialized training?" Backend data analysis: Teachers can view the "dosage calculation error rate" and common error types (such as unit conversion errors, weight not updated, etc.).

[0083] (III) Simulated nursing practice for pediatric diarrhea (including integration of traditional Chinese medicine features and massage techniques): 1. Case Overview: This case study, building upon routine diarrhea consultations, strengthens the diagnostic thinking of traditional Chinese medicine pediatrics, guiding students to inquire about information from the four diagnostic methods of traditional Chinese medicine, such as tongue coating and accompanying symptoms, and recommends and simulates pediatric massage techniques based on the diagnostic results.

[0084] 2. Knowledge base construction and configuration: Integration of Traditional Chinese Medicine Knowledge: Enter knowledge of traditional Chinese medicine pediatric diagnosis, such as: "Spleen deficiency diarrhea": loose stools, pale tongue with white coating, and poor appetite.

[0085] "Damp-heat diarrhea": foul-smelling stool, red tongue with yellow and greasy coating, and scanty, dark urine.

[0086] Structured operating procedures: The operation steps, duration, and indications of pediatric massage techniques (such as tonifying the spleen meridian, abdominal massage, and pushing up the seven vertebrae) are all transformed into structured knowledge segments and operation scoring standards.

[0087] FAQ Template Extension: Add a Traditional Chinese Medicine (TCM) consultation section to the standard question set: "What color is your baby's tongue coating? Is it thick?" "Are their palms and soles hot?" "Is their tummy bloated?" "Does their stool have any unusual odor?" 3. Example of interaction flow: The system guides (a digital human plays the role of the parent): "My baby has had diarrhea for three days. The stool looks like egg drop soup and has a slightly sour smell." Student: "Do you have a fever? How much urine is coming in?" System: "No fever, but less urination than usual." Student (who needs to proactively inquire about TCM signs): "Could you please check the baby's tongue coating? Is the baby's stomach bloated?" The system (uploaded simulated tongue coating image or description): "The tongue coating is a bit yellow, thick and greasy, and the stomach feels a bit bloated." The student (whose diagnosis should be "damp-heat diarrhea") said: "From the perspective of traditional Chinese medicine, this is damp-heat diarrhea. In addition to fluid replacement, pediatric massage can be used as an adjunct treatment." The system triggers virtual simulation linkage: "Do you wish to proceed with the pediatric massage simulation? Please select the correct technique based on your diagnosis." 4. Linkage between virtual simulation operation and scoring: The virtual simulation operation linkage module is activated, displaying a three-dimensional pediatric model and massage technique icons.

[0088] Student operation process: Choose "Clear the Large Intestine Meridian" and "Retreat the Six Fu Organs" (suitable for damp-heat diarrhea).

[0089] Follow the animation prompts to simulate the operation techniques, directions, and number of repetitions on the virtual model.

[0090] Operation rating: Whether the chosen method is correct (whether it conforms to dialectical thinking).

[0091] Is the operation sequence standardized?

[0092] Whether the force and speed are appropriate (assessed through gesture recognition or click rhythm).

[0093] Data feedback: The operational scores are incorporated into the overall evaluation of this consultation to generate a comprehensive capability report.

[0094] 5. Evaluation and Feedback: In addition to the four regular scoring items, the teaching evaluation module adds a scoring item for "the rationality of TCM syndrome differentiation".

[0095] Report example: "The TCM consultation process was complete (Excellent), the diagnosis was accurate (damp-heat diarrhea), and the massage techniques were correctly selected (Excellent). Suggestion: Further explanation of the principles of massage could be provided to enhance communication with parents." Backend data analysis: Teachers can view the "missing rate of TCM consultation" and "correctness rate of massage operation" to provide a basis for course optimization.

[0096] Examples of multimodal and advanced features: Multimodal Input: During the consultation, the system's digital human can say, "Some parents take pictures of their baby's stool. If you have a photo, you can upload it to me to help with the diagnosis." The student uploads a photo. The intelligent question-answering engine's multimodal big data analysis of the image identifies it as "egg drop soup-like stool" and, combined with the text dialogue context, generates the response: "Judging from the picture, the stool resembles egg drop soup, which further supports the diagnosis of viral gastroenteritis. More attention needs to be paid to fluid replacement." Switching between on-the-job training and competition modes: When teachers are preparing for the competition, they can switch the case study mode to "competition" in the background. At this time, the dialogue process is more compact, the system plays the role of a "standardized patient" and answers strictly in accordance with the competition scoring sheet, the timer is started, and the scoring weights w2 (keywords) and w4 (answer accuracy) of the teaching evaluation module are increased to simulate the competition scoring environment.

[0097] Through the detailed description of the above embodiments, those skilled in the art can clearly understand the composition, connection relationship, data flow, and specific working method of each module of the system of the present invention, and can realize the technical solution protected by the present invention based on existing technologies such as cloud computing, artificial intelligence, WebGL, and databases. The present invention deeply integrates professional nursing teaching knowledge, advanced artificial intelligence interactive technology, and scientific teaching evaluation theory to construct an implementable, assessable, and optimizable intelligent teaching closed loop, effectively improving the digitalization and intelligentization level of pediatric nursing talent training.

[0098] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0099] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A digital-based intelligent pediatric nursing teaching AI consultation system, characterized in that, include: The front-end interaction module is used to render and display the doctor's digital human image on the user terminal, receive the user's voice or text input, and synchronously play the voice response generated by the system and drive the mouth animation of the digital human image. The voice processing module is connected in communication with the aforementioned front-end interaction module and is used to convert the user's voice input into text and convert the system-generated text questions and answers into a voice stream. The teaching knowledge base module stores structured pediatric nursing teaching knowledge data, which includes at least: pediatric nursing curriculum standards reconstructed based on job competency requirements, typical cases related to nursing skills competition standards, evaluation standards for the professional skill levels of infant and toddler development guides, and characteristic knowledge and operational norms of traditional Chinese medicine pediatric nursing; the knowledge data is organized into knowledge graphs and / or vectorized representations to support semantic retrieval; The intelligent question-answering engine module is communicatively connected to both the speech processing module and the teaching knowledge base module. It receives the converted user query text, performs enhanced retrieval based on the teaching knowledge base module, and generates text responses that conform to pediatric nursing teaching standards and medical accuracy by combining preset dialogue management strategies. The dialogue management strategy includes a dialogue state machine based on a standardized patient consultation process, which guides users through the complete nursing process from nursing assessment and diagnosis to nursing interventions and health guidance. The teaching evaluation and feedback module is communicatively connected to the front-end interaction module and the intelligent question-answering engine module. It is used to record and analyze the user's interaction process data with the system in real time. The interaction process data includes at least: the completeness of the questioning logic, the accuracy of keyword extraction, the standardization of communication language, and the matching degree with the standard answers in the teaching knowledge base. Based on the interaction process data, a multi-dimensional ability evaluation report and personalized learning suggestions are generated. The backend management and data analysis module is used by teachers to manage teaching cases, update the knowledge base, view student learning trajectories, and conduct macro-analysis of teaching effectiveness based on the interaction data of all students, in order to support the continuous optimization of teaching models.

2. The AI-powered pediatric nursing teaching system based on a digital background as described in claim 1, characterized in that, The teaching knowledge base module also includes a dynamically updated typical case library, which contains multiple standardized pediatric nursing consultation scenarios. Each scenario corresponds to a structured FAQ template, which includes a standard set of questions, allowed variations of similar questions, and standard answers or answer generation logic that conform to nursing standards. During the dialogue, the intelligent question-answering engine module prioritizes matching and calling the FAQ template based on the current dialogue state to provide standardized and assessable interactive training.

3. The AI-based intelligent pediatric nursing teaching system for digital backgrounds as described in claim 2, characterized in that, The assessment model of the teaching evaluation and feedback module introduces a quantitative scoring function to calculate the score of a single consultation exercise. S : in, C logic The score represents the completeness of the consultation logic, which is calculated by comparing the coverage of the user's question sequence with the key nodes of the standard consultation process. C keyword The score represents the accuracy of keyword extraction, calculated by analyzing the matching degree between key medical terms extracted from user questions and a standard keyword list. C communication The score represents the standardization of communication language, and evaluates the politeness, clarity, and empathy of user statements based on a natural language processing model. C accuracy The score represents the answer matching degree. When a user answers questions in the system while playing the role of a nurse, the similarity between their answer and the standard answer in the knowledge base is calculated. w 1, w 2, w 3, w 4 represents the weighting coefficient for each sub-item score, and w 1+ w 2+ w 3+ w 4=1, and the weighting coefficient can be adjusted by the teacher according to the teaching focus.

4. The AI-powered pediatric nursing teaching system based on a digital background as described in claim 1, characterized in that, The retrieval enhancement generation method used by the intelligent question answering and engine module specifically includes: (a) Vectorize the user query text to obtain the query vector. V q ; (b) Calculate in the vector database of the teaching knowledge base module. V q Based on the similarity with all knowledge fragment vectors, retrieve the Top-K most relevant knowledge fragments. D 1, D 2, ..., D K ; (c) Combine the user's query text with the relevant knowledge fragments retrieved. D 1, D 2, ..., D K The prompt words, along with the current dialogue history, together constitute the prompt words, which are then input into the large language model; (d) The large language model generates the final response text based on the prompt words. The response text must reference or follow the content of the retrieved relevant knowledge fragments and conform to the context of nursing communication.

5. The AI-powered pediatric nursing teaching system based on a digital background according to claim 1, characterized in that, The system further includes a virtual simulation operation linkage module. When the answer generated by the intelligent question-and-answer engine module or the suggestion from the teaching assessment and feedback module involves a specific nursing operation, the virtual simulation operation linkage module is triggered, presenting a corresponding three-dimensional virtual simulation operation interface on the user terminal, guiding the user to practice the simulated operation, and feeding back the operation result data to the teaching assessment and feedback module.

6. The AI-powered pediatric nursing teaching system based on a digital background according to claim 1, characterized in that, The front-end interaction module supports multimodal input, allowing users to upload partial photos or simple line drawings of the child in addition to voice and text. The intelligent question-answering engine module integrates a multimodal large model, which can perform comprehensive analysis and answer by combining image information with the text dialogue context.

7. The AI-powered pediatric nursing teaching system based on a digital background according to claim 1, characterized in that, The backend management and data analysis module provides a function for analyzing teaching weaknesses based on group learning data. This function is implemented through the following steps: (a) Collect data on all students' interaction processes and assessment scores on specific teaching cases or knowledge points; (b) For students whose scores are below the threshold, cluster analysis is performed to identify common error types or missing consultation steps; (c) Calculate the overall error rate for a specific knowledge point. E k , ,in N error,k For knowledge points k The number of students who made the error N total,k To access knowledge points k Total number of student visits; (d) Present the analysis results to teachers in the form of visual charts and automatically mark knowledge points with high error rates, indicating that teaching needs to be strengthened or the knowledge base content needs to be optimized.

8. A digital-based intelligent pediatric nursing teaching AI consultation method, applied to the system as described in any one of claims 1-7, characterized in that, Includes the following steps: S1: The system starts up, the front-end interactive module loads and displays the designated doctor's digital avatar and initial greeting; S2: Receive the user's request for a consultation practice and determine the topic of the pediatric nursing case for the practice; S3: Based on the selected case, initialize the dialogue state machine. The intelligent question-answering engine module generates guiding questions or acts as a patient / family member to make statements based on the current state and in conjunction with the teaching knowledge base module. S4: Receives user's voice or text input, converts it through the voice processing module, processes it through the intelligent question-answering engine module, combines knowledge retrieval and dialogue history to generate answers that conform to nursing standards, and provides feedback through the front-end interaction module in the form of digital human voice and animation. S5: The teaching assessment and feedback module monitors and records the interaction process between steps S3 and S4 in real time, and extracts assessment features; S6: Repeat steps S3 to S5 until the dialogue state machine determines that the consultation process has been completed or the user has actively ended it. S7: The teaching assessment and feedback module generates and pushes a multi-dimensional assessment report and learning suggestions based on the complete interaction record. S8: The backend management and data analysis module summarizes the current and historical interaction data and updates the group learning analysis report.

9. The method according to claim 8, characterized in that, In step S3, the system supports a training mode that integrates job training, course preparation, competition, and certification. Teachers can configure the focus of each exercise in the background: when the focus is on job preparation, the dialogue process and assessment standards are closely aligned with the actual clinical job requirements; when the focus is on competition, the system simulates the standardized patient site assessment scenario and scoring rules of the nursing skills competition; when the focus is on certification, the question and answer content and assessment points are based on the assessment points of the infant and toddler development guide professional skills level certificate.

10. The method according to claim 8, characterized in that, The method also includes an adaptive learning path recommendation step: Based on the evaluation results of users' multiple historical practice sessions from the teaching assessment and feedback module, a user ability profile is constructed to identify their strengths and weaknesses. Based on the group's weaknesses information provided by the backend management and data analysis modules, the system intelligently recommends the next most suitable training case or knowledge point topic to the current user, with recommendation priority. P rec The calculation factors include: the user's historical scores for the corresponding knowledge point. S hist The group error rate of this knowledge point E k And the importance and weight of this knowledge point in the overall curriculum system. I k The calculation formula is: ; The recommendation results will be presented to the user through the front-end interaction module.

Images