Medical digital human simulation test management system and method based on situational dialogue

Abstract

The invention relates to the technical field of artificial intelligence, in particular to a medical digital human simulation test management system and method based on situational dialogues, and the system comprises a situational design module, a digital human dialogue generation module, a simulation test module, an evaluation analysis module, a feedback optimization module and a test management module. Wherein the scene design module is used for designing different scene dialogue tasks according to different medical scenes; the digital human dialogue generation module is used for generating dialogue content matched with the scene; the simulation test module is used for simulating the interaction process between the digital person and the patient or the medical staff; the evaluation analysis module is used for analyzing the data recorded by the simulation test module; and the feedback optimization module is used for adjusting parameters of the medical digital human model based on the evaluation result of the evaluation analysis module. According to the method and the system, the scene dialogue task matched with the medical scene is designed, and a dynamic feedback optimization mechanism is combined, so that the precise generation and continuous optimization of the dialogue content of the medical digital person are realized.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a medical digital human simulation test management system and method based on situational dialogue. Background Art

[0002] With the development of artificial intelligence technology, the medical industry has gradually introduced digital human technology to assist medical services and improve diagnosis and treatment efficiency. However, the application of digital human models in medical scenarios faces many challenges. First, medical scenarios are complex and diverse, and the conversation content, medical problems and patient conditions involved are highly professional and scenario-dependent. Traditional digital human dialogue systems are often unable to effectively adapt to changing medical scenarios, resulting in the lack of pertinence in the dialogue content they generate and the inability to provide accurate medical advice, which in turn affects the effect of actual applications. Secondly, existing simulation testing methods cannot conduct a comprehensive and systematic evaluation of medical digital humans, especially in multi-scenario and multi-task simulation tests, which lack effective task allocation, progress tracking and result analysis mechanisms.

[0003] In view of the above problems, the existing technology often lacks a comprehensive management framework when designing and testing medical digital human dialogue, which leads to the complexity and uncontrollability of the testing process. The existing simulation test often only focuses on the test of a single scenario or task, which is difficult to cover the diverse needs of different medical scenarios. At the same time, due to the lack of accurate evaluation and feedback mechanism, the dialogue quality of digital human and the accuracy of medical advice are difficult to continuously optimize in different scenarios. Therefore, it is necessary to propose a medical digital human simulation test management method based on scenario dialogue to solve the above problems. Summary of the invention

[0004] Based on the above purpose, the present invention provides a medical digital human simulation test management system and method based on situational dialogue.

[0005] The medical digital human simulation test management system based on scenario dialogue includes scenario design module, digital human dialogue generation module, simulation test module, evaluation and analysis module, feedback optimization module and test management module; among which:

[0006] Scenario design module: used to design different scenario dialogue tasks according to different medical scenarios and generate dialogue scripts;

[0007] Digital human dialogue generation module: used to receive the dialogue script generated by the scenario design module, and generate dialogue content matching the scenario in combination with the medical digital human model, and transmit the generated dialogue content to the simulation test module;

[0008] Simulation test module: used to simulate the interaction process between the digital human and the patient or medical staff according to the dialogue content output by the digital human dialogue generation module, and record the data in the dialogue, including dialogue time, response accuracy and patient reaction;

[0009] Evaluation and analysis module: used to analyze the data recorded by the simulation test module, evaluate the quality of the digital human dialogue and the accuracy of the medical advice, and transmit the evaluation results to the feedback optimization module;

[0010] Feedback optimization module: Based on the evaluation results of the evaluation and analysis module, adjust the parameters of the medical digital human model and optimize its dialogue generation strategy;

[0011] Test management module: used to manage the task allocation, progress tracking and test result storage of simulation tests, and supports the query and export of test results.

[0012] Optionally, the scenario design module includes a scenario classification unit, a patient information generation unit, a medical question generation unit, a dialogue flow design unit and a script generation unit; wherein:

[0013] Scenario classification unit: used to classify different medical scenarios according to medical fields and determine the tasks and medical needs of each scenario;

[0014] Patient information generation unit: used to generate patient information related to the task according to the task and medical needs determined by the scenario classification unit, wherein the patient information includes the patient's basic information, disease status, medication history, and allergy history;

[0015] A medical problem generating unit: used to generate corresponding medical problems according to the medical needs and patient information output by the scenario classification unit, wherein the medical problems include diagnosis problems, treatment plan problems and nursing needs problems;

[0016] Dialogue process design unit: used to design dialogue processes corresponding to scenarios based on patient information and medical needs, including the order of questions, the expected pattern of patient responses, and the handling of subsequent scenario feedback;

[0017] Script generation unit: used to integrate patient information, medical questions and dialogue flow to generate a complete scenario dialogue script; the script includes patient information, medical questions and dialogue flow.

[0018] Optionally, the digital human dialogue generation module includes a dialogue content parsing unit, a scene adaptation unit and a dialogue generation unit; wherein:

[0019] Dialogue content parsing unit: used to receive the dialogue script generated by the scenario design module, extract the patient information, medical questions and dialogue process, and convert them into a structured data format that can be processed by the medical digital human model, including JSON format, XML format or CSV format;

[0020] Scenario Adaptation Unit: Adjusts the language mode, tone, intonation and wording according to the parsed dialogue script and the characteristics of the medical digital human model to ensure that the generated dialogue content matches the needs of the medical scenario;

[0021] Dialogue generation unit: Based on the output of the scene adaptation unit and combined with the language generation algorithm of the digital human model, it generates dialogue content that is consistent with the patient information, medical problems and dialogue process, including responses to patient questions, explanations of medical advice and situational feedback during the interaction process.

[0022] Optionally, the scene adaptation unit includes:

[0023] Parsing scenario information: Extracting the types and characteristics of medical scenarios based on the parsed conversation scripts, including scenario information of diagnosis, treatment, and nursing;

[0024] Determine language style adjustment rules: Set different language style adjustment rules based on the extracted scenario information; Rule 1: In diagnosis and treatment scenarios, the language should be concise, clear and accurate; Rule 2: In nursing scenarios, the language should be caring, warm and gentle in expression;

[0025] Adjusting the tone and intonation: Adjusting the tone and intonation in the digital human model according to the determined language style adjustment rules;

[0026] Adjust the wording: Based on the requirements of the medical scenario, adjust the wording in the digital human model to ensure that the generated dialogue content conforms to medical professionalism and patient comprehensibility; the specific words used in diagnosis and treatment scenarios must comply with the standards of medical terminology; and in diagnosis scenarios, avoid excessive use of medical terms to keep it easy to understand.

[0027] Optionally, the dialogue generation unit includes:

[0028] receiving the output of the scene adaptation unit: obtaining the adjusted language pattern, tone, intonation, wording and other adaptation parameters;

[0029] Extract patient information and medical problems: Extract the patient's basic information and current medical problems from the conversation transcript;

[0030] Constructing a dialogue content generation model: Combining patient information, medical questions, dialogue flow, and the output of the scene adaptation unit, the Transformer model is used to generate dialogue content. The Transformer model uses an encoder-decoder architecture to first encode the input patient information and medical questions, and then generates a language response that meets the scene requirements through the decoder. The generation process includes calculating the conditional probability of each word: P(w t |w1,w2,...,w t-1 )=softmax(W·h t +b), where P(w t |w1,w2,...,w t-1 ) represents the conditional probability of generating the tth word at time t; w t represents the tth word generated; h t represents the hidden state vector of the decoder at time t; W is the weight matrix; b is the bias term; softmax(·) is the normalization function;

[0031] Language optimization: According to the specific needs of the medical scenario, the generated dialogue content is optimized to ensure the clarity of medical terms and expressions in the dialogue and the ease of understanding for patients. This process is optimized through the following formula: Among them, O(w t ) represents the probability of the tth word after optimization; α is the adjustment factor; is the indicator function, when the vocabulary w t When it belongs to the medical term set M, Otherwise, it is 0; M represents the set of medical terms;

[0032] Generate final conversation content: Input the optimized probability value into the decoder to generate a complete conversation text as the digital human's response to the patient or medical staff.

[0033] Optionally, the simulation test module includes a dialogue execution unit and a data recording unit; wherein:

[0034] Dialogue execution unit: used to simulate the interaction process between the digital human and the patient or medical staff according to the dialogue content output by the digital human dialogue generation module;

[0035] Data recording unit: used to record the data in the conversation during the simulation test, including conversation time, response accuracy and patient reaction;

[0036] The data recording unit comprises:

[0037] Dialogue time recording subunit: used to record the start time and end time of each round of dialogue, and calculate the duration of each dialogue;

[0038] Response accuracy recording subunit: used to record the degree of match between the conversation content generated by the digital human and the actual medical needs; the response accuracy is evaluated by comparing the degree of match between the conversation content and the expected medical questions, using a matching scoring algorithm, the formula is: Among them, the number of matching parts represents the part of the conversation content that matches the expected medical question; the total represents all elements of the conversation;

[0039] Patient response recording subunit: used to record the patient or medical staff's response to the digital human dialogue, including the correctness of the answer, the delay of the answer and the emotional expression.

[0040] Optionally, the evaluation and analysis module includes a data analysis unit and a quality evaluation unit; wherein:

[0041] Data analysis unit: used to comprehensively analyze the data recorded by the simulation test module, specifically analyzing various indicators in the conversation through data mining technology to quantify the quality of the digital human conversation;

[0042] The data analysis unit comprises:

[0043] Conversation time analysis subunit: used to calculate the duration of each round of conversation based on the conversation time record and evaluate the fluency of the conversation; if the conversation duration exceeds the predetermined threshold, it is considered that there is a delay or length problem in the conversation and needs to be optimized;

[0044] Response accuracy analysis subunit: used to compare the matching degree between the conversation content generated by the digital human and the expected medical questions based on the response accuracy records;

[0045] Patient response analysis subunit: used to evaluate the patient's acceptance and response quality to the digital human dialogue based on the recorded patient response data;

[0046] Quality assessment unit: assesses the quality of digital human dialogue based on the analysis results of the data analysis unit;

[0047] The quality assessment unit comprises:

[0048] Dialogue quality assessment subunit: Generates a dialogue quality score based on the comprehensive indicators analyzed by the dialogue time analysis subunit, response accuracy analysis subunit, and patient response analysis subunit. The score range is set from 1 to 5, where 1 represents very poor and 5 represents excellent.

[0049] Medical advice accuracy assessment subunit: Based on the response accuracy analysis, it evaluates the accuracy of the medical advice given in the digital human dialogue and generates a medical advice accuracy score.

[0050] Optionally, the feedback optimization module includes an evaluation result receiving unit, a model parameter adjustment unit, and a dialogue generation strategy optimization unit; wherein:

[0051] Evaluation result receiving unit: used to receive the evaluation results output by the evaluation and analysis module, including the dialogue quality score and the medical advice accuracy score;

[0052] Model parameter adjustment unit: used to adjust the parameters of the medical digital human model according to the evaluation results;

[0053] The model parameter adjustment unit comprises:

[0054] Dialogue quality optimization subunit: used to adjust the parameters related to dialogue generation in the digital human model, including tone and intonation, according to the dialogue quality score; the formula for the amplitude of parameter adjustment is:

[0055] Δθ=α·(dialogue quality score-preset target score), where Δθ represents the amplitude of parameter adjustment and α is the adjustment factor used to control the adjustment amplitude;

[0056] Medical advice accuracy optimization subunit: used to adjust the recommendation algorithm in the medical digital human model according to the medical advice accuracy score and optimize the accuracy of medical advice; the formula is:

[0057] in, is the adjustment range of the parameters related to medical advice, and β is the adjustment factor used to control the adjustment range;

[0058] Dialogue generation strategy optimization unit: used to optimize the digital human dialogue generation strategy based on the adjustment of model parameters. Specifically, according to the adjusted parameters, the generation algorithm is updated and the language generation strategy of the digital human model is adjusted, including the voice output method, the flexibility of the dialogue process and the contextual consideration when responding.

[0059] Optionally, the test management module includes

[0060] Task allocation unit: used to allocate simulation test tasks to testers or test groups

[0061] Progress tracking unit: used to track the progress of test tasks in real time, and record the execution status, completion time and remaining time of each test task;

[0062] Test result storage unit: used to store the result data of each test task, including dialogue content, evaluation data and test records;

[0063] Result query unit: used to query the stored test results by conditions, supporting retrieval based on multiple conditions such as scene, time or task;

[0064] Result export unit: used to export test results, supports exporting test data and analysis results to Excel, CSV or PDF format.

[0065] The medical digital human simulation test management method based on scenario dialogue is implemented by the above-mentioned medical digital human simulation test management system based on scenario dialogue, and includes the following steps:

[0066] S1: Design different medical scenarios, including diagnosis, treatment, and care, based on the needs of the medical field, and determine the tasks and medical needs of each scenario;

[0067] S2: Based on the medical scenario designed in S1, generate a scenario dialogue task matching the scenario and form a dialogue script, which includes patient information, medical questions and dialogue flow;

[0068] S3: Input the dialogue script generated by S2 into the digital human dialogue generation module, combine it with the medical digital human model, and generate dialogue content matching the scenario according to the dialogue script;

[0069] S4: The conversation content generated by S3 is transmitted to the simulation test module to simulate the interaction process between the digital human and the patient or medical staff, and the data during the conversation is recorded, including the conversation time, response accuracy and patient reaction;

[0070] S5: Transmit the conversation data recorded in S4 to the evaluation and analysis module, analyze the data, evaluate the quality of the digital human conversation and the accuracy of the medical advice, and generate an evaluation result;

[0071] S6: Based on the evaluation results of S5, the optimization suggestions are transmitted to the feedback optimization module to adjust the parameters of the medical digital human model and optimize its dialogue generation strategy;

[0072] S7: Based on the medical digital human model optimized in S6, the conversation content is regenerated, and task assignment, progress tracking and test result storage are performed through the test management module.

[0073] Beneficial effects of the present invention:

[0074] The present invention effectively simulates the interaction process between digital humans and patients or medical staff by designing dialogue tasks that match the scenarios and generating corresponding dialogue content in combination with the medical digital human model, thereby ensuring the authenticity and efficiency of the dialogue content. This technical solution can cover a variety of medical scenarios such as diagnosis, treatment, and nursing, meet different testing needs, and ensure the comprehensiveness and efficiency of the testing process.

[0075] The present invention uses real-time recording and analysis of conversation data, and adjusts the generation strategy of the medical digital human model in combination with the evaluation results, thereby continuously optimizing the conversation quality and the accuracy of medical advice. This technology can ensure that the performance of digital humans in different medical scenarios is gradually improved through a dynamic feedback mechanism, thereby improving their reliability and practicality in actual applications, and further promoting the application and development of medical digital humans in the medical industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0077] Figure 1 This is a schematic diagram of a medical digital human simulation test management system according to an embodiment of the present invention;

[0078] Figure 2 Schematic diagram of a medical digital human simulation test management method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0079] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0080] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0081] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0082] like Figure 1 As shown, the medical digital human simulation test management system based on scenario dialogue includes a scenario design module, a digital human dialogue generation module, a simulation test module, an evaluation and analysis module, a feedback optimization module and a test management module; wherein:

[0083] Scenario design module: used to design different scenario dialogue tasks according to different medical scenarios (such as diagnosis, treatment, nursing, etc.) and generate dialogue scripts;

[0084] Digital human dialogue generation module: used to receive the dialogue script generated by the scenario design module, and generate dialogue content matching the scenario in combination with the medical digital human model, and transmit the generated dialogue content to the simulation test module;

[0085] Simulation test module: used to simulate the interaction process between the digital human and the patient or medical staff according to the dialogue content output by the digital human dialogue generation module, and record the data in the dialogue, including the dialogue time, response accuracy and patient reaction; the simulation test module transmits the test data to the evaluation and analysis module;

[0086] Evaluation and analysis module: used to analyze the data recorded by the simulation test module, evaluate the quality of the digital human dialogue and the accuracy of the medical advice, and transmit the evaluation results to the feedback optimization module;

[0087] Feedback optimization module: Based on the evaluation results of the evaluation and analysis module, adjust the parameters of the medical digital human model, optimize its dialogue generation strategy, and improve the dialogue effect;

[0088] Test management module: used to manage the task allocation, progress tracking and test result storage of simulation tests, and supports the query and export of test results.

[0089] The scenario design module includes a scenario classification unit, a patient information generation unit, a medical question generation unit, a dialogue flow design unit, and a script generation unit; among which:

[0090] Scenario classification unit: used to classify different medical scenarios according to medical fields (such as diagnosis, treatment, nursing, etc.) and determine the tasks and medical needs of each scenario. This unit clarifies the theme of the scenario dialogue task and specific medical problems based on the characteristics of various medical scenarios;

[0091] Patient information generation unit: used to generate task-related patient information based on the tasks and medical needs determined by the scenario classification unit. The patient information includes the patient's basic information (such as age, gender, medical history, etc.), disease status, medication history, and allergy history. This unit ensures that the generated patient information is consistent with the selected medical scenario and task;

[0092] Medical problem generation unit: used to generate corresponding medical problems based on the medical needs and patient information output by the scenario classification unit. Medical problems include diagnosis problems, treatment plan problems and nursing needs problems. Questions are raised based on the patient's specific condition and treatment stage to ensure that the medical problems match the task requirements.

[0093] Dialogue process design unit: used to design dialogue processes corresponding to scenarios based on patient information and medical needs, including the order of questions, the expected pattern of patient responses, and the handling of subsequent scenario feedback;

[0094] Script generation unit: used to integrate patient information, medical problems and dialogue processes to generate a complete scenario dialogue script; the script includes patient information, medical problems and dialogue processes; the above units can ensure that the generated scenario dialogue script is authentic and effective through accurate classification of different medical scenarios, generation of patient information, reasonable design of medical problems, and optimal arrangement of dialogue processes, and can adapt to the interaction needs in different medical scenarios, thereby improving the interaction quality of medical digital humans and the accuracy of simulation tests.

[0095] The digital human dialogue generation module includes a dialogue content analysis unit, a scene adaptation unit and a dialogue generation unit; wherein:

[0096] Dialogue content parsing unit: used to receive the dialogue script generated by the scenario design module, extract the patient information, medical questions and dialogue process, and convert them into a structured data format that can be processed by the medical digital human model, including JSON format, XML format or CSV format, so as to provide input data for the subsequent generation of dialogue content;

[0097] Scenario Adaptation Unit: Adjusts the language mode, tone, intonation and wording according to the parsed dialogue script and the characteristics of the medical digital human model to ensure that the generated dialogue content matches the needs of the medical scenario;

[0098] Dialogue generation unit: Based on the output of the scenario adaptation unit and combined with the language generation algorithm of the digital human model, it generates dialogue content that is consistent with the patient information, medical problems and dialogue process, including responses to patient questions, explanations of medical advice and situational feedback during the interaction process; through the synergy of various units, the digital human dialogue generation module can effectively convert the dialogue script generated by the scenario design module into dialogue content that meets the needs of actual medical scenarios, thereby improving the application effect and interaction quality of the medical digital human model in simulation testing.

[0099] The scene adaptation unit includes:

[0100] Parsing scenario information: Extracting the types and characteristics of medical scenarios based on the parsed conversation scripts, including scenario information of diagnosis, treatment, and nursing;

[0101] Determine language style adjustment rules: Set different language style adjustment rules based on the extracted scenario information; Rule 1: In diagnosis and treatment scenarios, the language should be concise, clear and accurate; Rule 2: In nursing scenarios, the language should be caring, warm and gentle in expression;

[0102] Adjust the tone and intonation: According to the determined language style adjustment rules, adjust the tone and intonation in the digital human model; specifically, model the emotional intensity of each speech in the conversation (such as "affirmation", "question" or "suggestion"), and use the following formula to calculate the tone weight:

[0103] Among them, the intensity of sentiment words i Represents the strength value of the i-th sentiment word and the weight of sentiment expression i It indicates the weight of the word in the specified scenario (e.g., the weight of the word "confirmation" is higher in diagnosis), and n is the number of sentiment words;

[0104] Adjust the wording: Based on the requirements of the medical scenario, adjust the wording in the digital human model to ensure that the generated dialogue content conforms to medical professionalism and patient comprehensibility; the specific wording in the diagnosis and treatment scenarios must comply with the norms of medical terminology; and in the diagnosis scenario, avoid excessive use of medical terms to keep it easy to understand; through the above steps, the scenario adaptation unit can ensure that the generated dialogue content is highly matched with the needs of the medical scenario, improve the applicability of the digital human model in different medical tasks, and enhance the interaction effect with patients or medical staff.

[0105] The dialogue generation unit includes:

[0106] Receive the output of the scene adaptation unit: obtain the adjusted language pattern, tone, intonation, wording and other adaptation parameters, which will serve as the basic input for generating dialogue content;

[0107] Extract patient information and medical problems: Extract the patient's basic information (such as age, gender, medical history, etc.) and current medical problems (such as diagnosis, treatment plan, care needs, etc.) from the conversation script in order to generate corresponding conversation content for specific situations;

[0108] Constructing a dialogue content generation model: Combining patient information, medical questions, dialogue flow, and the output of the scene adaptation unit, the Transformer model is used to generate dialogue content. The Transformer model uses an encoder-decoder architecture to first encode the input patient information and medical questions, and then generates a language response that meets the scene requirements through the decoder. The generation process includes calculating the conditional probability of each word: P(w t |w1,w2,...,w t-1)=softmax(W·h t +b), where P(w t |w1,w2,...,w t-1 ) represents the conditional probability of generating the tth word at time t; w t represents the tth word generated; h t represents the hidden state vector of the decoder at time t; W is the weight matrix, which represents the processing weight of the decoder on the current hidden state; b is the bias term, which represents the offset of the decoder to the generated vocabulary; softmax(·) is a normalization function that converts the output of the decoder into a probability distribution;

[0109] Language optimization: According to the specific needs of the medical scenario, the generated dialogue content is optimized to ensure the clarity of medical terms and expressions in the dialogue and the ease of understanding for patients. This process is optimized through the following formula: Among them, O(w t ) represents the probability of the tth word after optimization; α is the adjustment factor used to control the influence of medical terms; is the indicator function, when the vocabulary w t When it belongs to the medical term set M, Otherwise, it is 0; M represents the set of medical terms;

[0110] Generate final dialogue content: Input the optimized probability value into the decoder to generate a complete dialogue text as the digital human's response to the patient or medical staff, ensuring that the generated dialogue content meets the requirements of the medical scenario and follows the logical order of the dialogue process; through the above steps, the dialogue generation unit can generate dialogue content that meets the requirements of the medical scenario based on the output of the scenario adaptation unit, patient information and medical problems, combined with the language generation algorithm of the digital human model, thereby achieving high-quality medical dialogue simulation.

[0111] The simulation test module includes a dialogue execution unit and a data recording unit; wherein:

[0112] Dialogue execution unit: used to simulate the interaction process between the digital human and the patient or medical staff according to the dialogue content output by the digital human dialogue generation module;

[0113] The interactive process follows these steps:

[0114] Step 1, receiving the dialogue content output by the digital human dialogue generation module;

[0115] Step 2: Input the conversation content into the digital human model to generate voice or text, simulating the digital human responding to the patient or medical staff;

[0116] Step 3: Based on the patient or medical staff's feedback (such as answers, expressions, actions, etc.), continue the dialogue process and generate new dialogue content.

[0117] Data recording unit: used to record the data in the conversation during the simulation test, including conversation time, response accuracy and patient reaction;

[0118] The data logging unit includes:

[0119] Conversation time recording subunit: used to record the start time and end time of each round of conversation, and calculate the duration of each conversation. The specific calculation formula is: conversation duration = end time - start time;

[0120] Response accuracy recording subunit: used to record the degree of match between the conversation content generated by the digital human and the actual medical needs; the response accuracy is evaluated by comparing the degree of match between the conversation content and the expected medical questions, using a matching scoring algorithm, the formula is: Among them, the number of matching parts indicates the part of the conversation content that matches the expected medical question; the total represents all elements in the conversation, including patient information, medical questions, and conversation process. The higher the matching degree, the higher the response accuracy.

[0121] Patient response recording subunit: used to record the patient or medical staff's response to the digital human dialogue, including the correctness of the answer, the delay of the answer and the emotional expression; through the above unit, the simulation test module can truly simulate the interaction process between the digital human and the patient or medical staff according to the dialogue content output by the digital human dialogue generation module, and accurately record the key data in the dialogue, including the dialogue time, response accuracy and patient reaction, so as to provide accurate support for data analysis in the simulation test process.

[0122] The evaluation and analysis module includes a data analysis unit and a quality evaluation unit; wherein:

[0123] Data analysis unit: used to comprehensively analyze the data recorded by the simulation test module. Specifically, it uses data mining technology to analyze various indicators in the dialogue, quantify the quality of the digital human dialogue, and provide support for subsequent optimization;

[0124] The data analysis unit includes:

[0125] Conversation time analysis subunit: used to calculate the duration of each round of conversation based on the conversation time record and evaluate the fluency of the conversation; if the conversation duration exceeds the preset threshold, it is considered that the conversation has delay or length problems and needs to be optimized; the specific calculation formula is: conversation duration deviation = actual conversation duration - preset conversation duration; if the deviation is greater than the threshold, it is marked as an abnormal conversation;

[0126] Response accuracy analysis subunit: used to compare the match between the conversation content generated by the digital human and the expected medical questions based on the response accuracy records; calculate the response accuracy score, the formula is: Among them, the number of matching parts indicates the part of the conversation content that matches the medical question, and the total number of conversation elements includes all question and answer elements;

[0127] Patient response analysis subunit: used to evaluate the patient's acceptance and response quality of the digital human dialogue based on the recorded patient response data; the response quality is calculated by calculating the relationship between response delay and response accuracy using the following formula: Among them, γ is the adjustment factor, which is used to control the effect of response delay on response quality score. Response delay refers to the time from when the patient asks a question to when he gives an answer;

[0128] Quality assessment unit: assesses the quality of digital human dialogue based on the analysis results of the data analysis unit;

[0129] The quality assessment units include:

[0130] Conversation quality assessment subunit: Based on the comprehensive indicators analyzed by the conversation time analysis subunit, response accuracy analysis subunit and patient response analysis subunit, a conversation quality score is generated, and the score range is set to 1 to 5, where 1 represents very poor and 5 represents excellent; the comprehensive score is calculated according to the following weighted formula: Conversation quality score = w1·conversation duration score + w2·response accuracy score + w3·response quality score, where w1, w2, w3 are weight coefficients, indicating the contribution ratio of each score to the final quality score, and satisfying: w1+w2+w3=1;

[0131] Medical advice accuracy assessment subunit: Based on the response accuracy analysis, the accuracy of the medical advice given by the digital human in the dialogue is assessed to generate a medical advice accuracy score; specifically, the medical advice given by the digital human is compared with the actual medical plan in terms of consistency, using the following calculation formula:

[0132] Among them, the number of matching suggestions indicates the number of times the suggestions given by the digital human are consistent with the actual medical standards or diagnostic guidelines, and the total number of suggestions is the total number of all suggestions made by the digital human. Through the above units, the evaluation and analysis module can conduct a comprehensive analysis of the data recorded by the simulation test module, evaluate the quality of the digital human dialogue and the accuracy of the medical suggestions, and thus provide effective feedback for the continuous optimization of the medical digital human.

[0133] The feedback optimization module includes an evaluation result receiving unit, a model parameter adjustment unit, and a dialogue generation strategy optimization unit; wherein:

[0134] Evaluation result receiving unit: used to receive the evaluation results output by the evaluation and analysis module, including the dialogue quality score and the medical advice accuracy score;

[0135] Model parameter adjustment unit: used to adjust the parameters of the medical digital human model according to the evaluation results;

[0136] The model parameter adjustment unit includes:

[0137] Dialogue quality optimization subunit: used to adjust the parameters related to dialogue generation in the digital human model, including tone and intonation, according to the dialogue quality score; the formula for the amplitude of parameter adjustment is:

[0138] Δθ = α·(conversation quality score - preset target score), where Δθ represents the amplitude of parameter adjustment, α is the adjustment factor used to control the adjustment amplitude, and the preset target score is the ideal conversation quality score defined by the system to ensure that the conversation content is more in line with the patient's needs and the standards of the medical scenario;

[0139] Medical advice accuracy optimization subunit: used to adjust the recommendation algorithm in the medical digital human model according to the medical advice accuracy score and optimize the accuracy of medical advice; the formula is:

[0140] in, is the adjustment range of the parameters related to medical advice, β is the adjustment factor used to control the adjustment range, and the preset target accuracy is the standard medical advice accuracy score; by optimizing the recommendation algorithm in the model, the accuracy of the digital human in medical advice is improved;

[0141] Dialogue generation strategy optimization unit: used to optimize the digital human dialogue generation strategy based on the adjustment of model parameters. Specifically, according to the adjusted parameters, the generation algorithm is updated and the language generation strategy of the digital human model is adjusted, including the voice output method, the flexibility of the dialogue process and the contextual consideration when answering. Through the above units, the feedback optimization module can adjust the parameters of the medical digital human model based on the evaluation results output by the evaluation and analysis module, thereby optimizing its dialogue generation strategy and improving the performance of the digital human in medical dialogues and the accuracy of medical advice.

[0142] The test management module includes

[0143] Task allocation unit: used to allocate simulation test tasks to testers or test groups. This unit automatically or manually selects appropriate testers based on test scenarios and personnel resources to ensure the rationality of task allocation;

[0144] Progress tracking unit: used to track the progress of test tasks in real time, record the execution status, completion time and remaining time of each test task, and ensure that the test process is carried out according to the predetermined plan;

[0145] Test result storage unit: used to store the result data of each test task, including dialogue content, evaluation data and test records; and store the results in the database in a standard format to ensure data integrity and accessibility;

[0146] Result query unit: used to query the stored test results by conditions, support retrieval based on multiple conditions such as scene, time or task, and provide fast access to results;

[0147] Result export unit: used to export test results, supports exporting test data and analysis results to Excel, CSV or PDF formats; through the above units, the test management module realizes efficient allocation of test tasks, real-time tracking of progress, storage and management of results, and provides flexible query and export functions to ensure the smooth progress of the test process.

[0148] like Figure 2 As shown, the medical digital human simulation test management method based on scenario dialogue is implemented by the above-mentioned medical digital human simulation test management system based on scenario dialogue, and includes the following steps:

[0149] S1: Design different medical scenarios, including diagnosis, treatment, and care, based on the needs of the medical field, and determine the tasks and medical needs of each scenario;

[0150] S2: Based on the medical scenario designed in S1, generate scenario dialogue tasks that match the scenario and form a dialogue script, which includes patient information, medical questions and dialogue flow;

[0151] S3: Input the dialogue script generated by S2 into the digital human dialogue generation module, combine it with the medical digital human model, and generate dialogue content matching the scenario according to the dialogue script;

[0152] S4: The conversation content generated by S3 is transmitted to the simulation test module to simulate the interaction process between the digital human and the patient or medical staff, and the data during the conversation is recorded, including the conversation time, response accuracy and patient reaction;

[0153] S5: Transmit the conversation data recorded in S4 to the evaluation and analysis module, analyze the data, evaluate the quality of the digital human conversation and the accuracy of the medical advice, and generate an evaluation result;

[0154] S6: Based on the evaluation results of S5, the optimization suggestions are transmitted to the feedback optimization module to adjust the parameters of the medical digital human model and optimize its dialogue generation strategy;

[0155] S7: Based on the medical digital human model optimized in S6, the conversation content is regenerated, and task assignment, progress tracking and test result storage are performed through the test management module.

[0156] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0157] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A medical digital human simulation test management system based on situational dialogue, characterized in that: It includes scenario design module, digital human dialogue generation module, simulation test module, evaluation and analysis module, feedback optimization module and test management module; among which: Scenario design module: used to design different scenario dialogue tasks according to different medical scenarios and generate dialogue scripts; Digital human dialogue generation module: used to receive the dialogue script generated by the scenario design module, and generate dialogue content matching the scenario in combination with the medical digital human model, and transmit the generated dialogue content to the simulation test module; Simulation test module: used to simulate the interaction process between the digital human and the patient or medical staff according to the dialogue content output by the digital human dialogue generation module, and record the data in the dialogue, including dialogue time, response accuracy and patient reaction; Evaluation and analysis module: used to analyze the data recorded by the simulation test module, evaluate the quality of the digital human dialogue and the accuracy of the medical advice, and transmit the evaluation results to the feedback optimization module; Feedback optimization module: Based on the evaluation results of the evaluation and analysis module, adjust the parameters of the medical digital human model and optimize its dialogue generation strategy; Test management module: used to manage the task allocation, progress tracking and test result storage of simulation tests, and supports the query and export of test results.

2. The medical digital human simulation test management system based on scenario dialogue according to claim 1 is characterized in that: The scenario design module includes a scenario classification unit, a patient information generation unit, a medical question generation unit, a dialogue flow design unit and a script generation unit; wherein: Scenario classification unit: used to classify different medical scenarios according to medical fields and determine the tasks and medical needs of each scenario; A patient information generation unit: used to generate patient information related to the task according to the task and medical needs determined by the scenario classification unit, wherein the patient information includes the patient's basic information, disease status, medication history, and allergy history; A medical problem generating unit: used to generate corresponding medical problems according to the medical needs and patient information output by the scenario classification unit, wherein the medical problems include diagnosis problems, treatment plan problems and nursing needs problems; Dialogue process design unit: used to design dialogue processes corresponding to scenarios based on patient information and medical needs, including the order of questions, the expected pattern of patient responses, and the handling of subsequent scenario feedback; Script generation unit: used to integrate patient information, medical questions and dialogue flow to generate a complete scenario dialogue script; the script includes patient information, medical questions and dialogue flow.

3. The medical digital human simulation test management system based on scenario dialogue according to claim 1 is characterized in that: The digital human dialogue generation module includes a dialogue content analysis unit, a scene adaptation unit and a dialogue generation unit; wherein: Dialogue content parsing unit: used to receive the dialogue script generated by the scenario design module, extract the patient information, medical questions and dialogue process, and convert them into a structured data format that can be processed by the medical digital human model, including JSON format, XML format or CSV format; Scenario Adaptation Unit: Adjusts the language mode, tone, intonation and wording according to the analyzed dialogue script and the characteristics of the medical digital human model to ensure that the generated dialogue content matches the needs of the medical scenario; Dialogue generation unit: Based on the output of the scene adaptation unit and combined with the language generation algorithm of the digital human model, it generates dialogue content that is consistent with the patient information, medical problems and dialogue process, including responses to patient questions, explanations of medical advice and situational feedback during the interaction process.

4. The medical digital human simulation test management system based on scenario dialogue according to claim 3 is characterized in that: The scene adaptation unit comprises: Parsing scenario information: Extracting the types and characteristics of medical scenarios based on the parsed conversation scripts, including scenario information of diagnosis, treatment, and nursing; Determine language style adjustment rules: Set different language style adjustment rules based on the extracted scenario information; Rule 1: In diagnosis and treatment scenarios, the language should be concise, clear and accurate; Rule 2: In nursing scenarios, the language should be caring, warm and gentle in expression; Adjusting the tone and intonation: Adjusting the tone and intonation in the digital human model according to the determined language style adjustment rules; Adjust the wording: Based on the requirements of the medical scenario, adjust the wording in the digital human model to ensure that the generated dialogue content conforms to medical professionalism and patient comprehensibility; the specific words used in diagnosis and treatment scenarios must comply with the standards of medical terminology; and in diagnosis scenarios, avoid excessive use of medical terms to keep it easy to understand.

5. The medical digital human simulation test management system based on scenario dialogue according to claim 4 is characterized in that: The dialogue generation unit comprises: receiving the output of the scene adaptation unit: obtaining the adjusted language pattern, tone, intonation, wording and other adaptation parameters; Extract patient information and medical problems: Extract the patient's basic information and current medical problems from the conversation transcript; Constructing a dialogue content generation model: Combining patient information, medical questions, dialogue flow, and the output of the scene adaptation unit, the Transformer model is used to generate dialogue content. The Transformer model uses an encoder-decoder architecture to first encode the input patient information and medical questions, and then generates a language response that meets the scene requirements through the decoder. The generation process includes calculating the conditional probability of each word: P(w t |w1,w2,...,w t-1 )=softmax(W·h t +b), where P(w t |w1,w2,...,w t-1 ) represents the conditional probability of generating the tth word at time t; w t represents the tth word generated; h t represents the hidden state vector of the decoder at time t; W is the weight matrix; b is the bias term; softmax(·) is the normalization function; Language optimization: According to the specific needs of the medical scenario, the generated dialogue content is optimized to ensure the clarity of medical terms and expressions in the dialogue and the ease of understanding for patients. This process is optimized through the following formula: Among them, O(w t ) represents the probability of the tth word after optimization; α is the adjustment factor; is the indicator function, when the vocabulary w t When it belongs to the medical term set M, Otherwise, it is 0; M represents the set of medical terms; Generate final conversation content: Input the optimized probability value into the decoder to generate a complete conversation text as the digital human's response to the patient or medical staff.

6. The medical digital human simulation test management system based on scenario dialogue according to claim 1 is characterized in that: The simulation test module includes a dialogue execution unit and a data recording unit; wherein: Dialogue execution unit: used to simulate the interaction process between the digital human and the patient or medical staff according to the dialogue content output by the digital human dialogue generation module; Data recording unit: used to record the data in the conversation during the simulation test, including conversation time, response accuracy and patient reaction; The data recording unit comprises: Dialogue time recording subunit: used to record the start time and end time of each round of dialogue, and calculate the duration of each dialogue; Response accuracy recording subunit: used to record the degree of match between the conversation content generated by the digital human and the actual medical needs; the response accuracy is evaluated by comparing the consistency between the conversation content and the expected medical questions, using a matching scoring algorithm, the formula is: Among them, the number of matching parts indicates the part of the conversation content that matches the expected medical question; the total represents all elements of the conversation; Patient response recording subunit: used to record the patient or medical staff's response to the digital human dialogue, including the correctness of the answer, the delay of the answer and the emotional expression.

7. The medical digital human simulation test management system based on scenario dialogue according to claim 6 is characterized in that: The evaluation and analysis module includes a data analysis unit and a quality evaluation unit; wherein: Data analysis unit: used to comprehensively analyze the data recorded by the simulation test module, specifically analyzing various indicators in the conversation through data mining technology to quantify the quality of the digital human conversation; The data analysis unit comprises: Conversation time analysis subunit: used to calculate the duration of each round of conversation based on the conversation time record and evaluate the fluency of the conversation; if the conversation duration exceeds the predetermined threshold, it is considered that there is a delay or length problem in the conversation and needs to be optimized; Response accuracy analysis subunit: used to compare the matching degree between the conversation content generated by the digital human and the expected medical questions based on the response accuracy records; Patient response analysis subunit: used to evaluate the patient's acceptance and response quality to the digital human dialogue based on the recorded patient response data; Quality assessment unit: assesses the quality of digital human dialogue based on the analysis results of the data analysis unit; The quality assessment unit comprises: Dialogue quality assessment subunit: Generates a dialogue quality score based on the comprehensive indicators analyzed by the dialogue time analysis subunit, response accuracy analysis subunit, and patient response analysis subunit. The score range is set from 1 to 5, where 1 represents very poor and 5 represents excellent. Medical advice accuracy assessment subunit: Based on the response accuracy analysis, it evaluates the accuracy of the medical advice given in the digital human dialogue and generates a medical advice accuracy score.

8. The medical digital human simulation test management system based on scenario dialogue according to claim 1 is characterized in that: The feedback optimization module includes an evaluation result receiving unit, a model parameter adjustment unit, and a dialogue generation strategy optimization unit; wherein: Evaluation result receiving unit: used to receive the evaluation results output by the evaluation and analysis module, including the dialogue quality score and the medical advice accuracy score; Model parameter adjustment unit: used to adjust the parameters of the medical digital human model according to the evaluation results; The model parameter adjustment unit comprises: Dialogue quality optimization subunit: used to adjust the parameters related to dialogue generation in the digital human model, including tone and intonation, according to the dialogue quality score; the formula for the amplitude of parameter adjustment is: Δθ=α·(dialogue quality score-preset target score), where Δθ represents the amplitude of parameter adjustment and α is the adjustment factor used to control the adjustment amplitude; Medical advice accuracy optimization subunit: used to adjust the recommendation algorithm in the medical digital human model according to the medical advice accuracy score and optimize the accuracy of medical advice; the formula is: in, is the adjustment range of the parameters related to medical advice, and β is the adjustment factor used to control the adjustment range; Dialogue generation strategy optimization unit: used to optimize the digital human dialogue generation strategy based on the adjustment of model parameters. Specifically, according to the adjusted parameters, the generation algorithm is updated and the language generation strategy of the digital human model is adjusted, including the voice output method, the flexibility of the dialogue process and the contextual consideration when responding.

9. The medical digital human simulation test management system based on scenario dialogue according to claim 1, characterized in that: The test management module includes Task allocation unit: used to allocate simulation test tasks to testers or test groups; Progress tracking unit: used to track the progress of test tasks in real time, and record the execution status, completion time and remaining time of each test task; Test result storage unit: used to store the result data of each test task, including dialogue content, evaluation data and test records Result query unit: used to query the stored test results by conditions, supporting retrieval based on multiple conditions such as scene, time or task; Result export unit: used to export test results, supports exporting test data and analysis results to Excel, CSV or PDF format.

10. A medical digital human simulation test management method based on scenario dialogue, implemented by the medical digital human simulation test management system based on scenario dialogue according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1: Design different medical scenarios according to the needs of the medical field, including diagnosis, treatment and nursing, and determine the tasks and medical needs of each scenario; S2: Based on the medical scenario designed in S1, generate a scenario dialogue task matching the scenario and form a dialogue script, which includes patient information, medical questions and dialogue flow; S3: Input the dialogue script generated by S2 into the digital human dialogue generation module, combine it with the medical digital human model, and generate dialogue content matching the scenario according to the dialogue script; S4: The conversation content generated by S3 is transmitted to the simulation test module to simulate the interaction process between the digital human and the patient or medical staff, and the data during the conversation is recorded, including the conversation time, response accuracy and patient reaction; S5: Transmit the conversation data recorded in S4 to the evaluation and analysis module, analyze the data, evaluate the quality of the digital human conversation and the accuracy of the medical advice, and generate an evaluation result; S6: Based on the evaluation results of S5, the optimization suggestions are transmitted to the feedback optimization module to adjust the parameters of the medical digital human model and optimize its dialogue generation strategy; S7: Based on the medical digital human model optimized in S6, the conversation content is regenerated, and task assignment, progress tracking and test result storage are performed through the test management module.

Images