Intelligent inquiry method and device and electronic equipment

Abstract

The invention discloses an intelligent inquiry method and device and electronic equipment, and relates to the technical field of intelligent inquiry and the technical field of data processing.The intelligent inquiry method comprises the steps that symptom information provided by a patient in the current round of inquiry is obtained, and the symptom information serves as latest symptom information; generating a current candidate disease set based on the latest symptom information; for each candidate disease in the current candidate disease set, calculating the posterior probability of the candidate disease after the current round of inquiry based on the latest symptom information by adopting a Bayesian algorithm; based on the posterior probability of each candidate disease in the current candidate disease set, the uncertainty entropy of the current candidate disease set is calculated, and the uncertainty entropy represents the uncertainty of the disease diagnosis result of the current candidate disease set; and determining whether to end the intelligent inquiry or not based on a size relationship between the uncertainty entropy and a preset entropy threshold value. By adopting the scheme, the inquiry efficiency and the inquiry integrity in intelligent inquiry are effectively balanced.

Description

Technical Field

[0001] This application relates to the fields of intelligent consultation technology and data processing technology, and in particular to an intelligent consultation method, device and electronic device. Background Technology

[0002] Currently, in the field of AI-assisted medical diagnosis, existing technical solutions are mainly divided into two categories: rule-based diagnosis systems and single-agent large language models generated end-to-end.

[0003] The first type is the rule-based template-based consultation system. These systems typically rely on pre-defined decision trees, question banks, or fixed consultation templates, controlling the questioning order through hard-coded program logic. During the consultation, the system asks the patient questions item by item according to a predetermined tree-like or tabular process (e.g., "Do you have pain?", "If so, where is it?"), and populates the template with the results after information collection. The main drawback is that while the output format is uniform, its core flaw lies in the static and rigid questioning path. The system cannot dynamically adjust its focus based on the patient's actual complaints, lacking the dynamic reasoning of a doctor's "hypothesis-verification" process; essentially, it merely executes a pre-defined sequence of questions, unable to handle complex and varied atypical cases, resulting in low consultation efficiency and a lack of true intelligent reasoning ability.

[0004] The second category is Large Language Models (LLMs), which are single-agent models trained on large-scale medical corpora. These models utilize deep learning techniques to directly interact with patients end-to-end in natural language. Based on the input context, the model uses an internal attention mechanism to predict and generate the next round of questions. The main drawback is that while these models perform excellently in semantic understanding and natural interaction, they suffer from serious uncontrollability and black-box issues when applied to serious medical scenarios.

[0005] For example, the lack of a quantitative convergence mechanism: due to the lack of a clear objective function for consultation, the model often cannot determine "when to end the consultation", which can easily lead to invalid and redundant follow-up questions or premature termination when key information is missing, and the integrity of information collection cannot be guaranteed.

[0006] In summary, the lack of reasonable objective termination conditions makes it difficult to balance the efficiency and completeness of the consultation process. Summary of the Invention

[0007] This application provides an intelligent consultation method, device, and electronic device to solve the problem of the inability to effectively balance consultation efficiency and consultation completeness in the existing intelligent consultation technology.

[0008] This application provides an intelligent consultation method, including:

[0009] Obtain the symptom information provided by the patient in this round of consultation, and use the symptom information as the latest symptom information;

[0010] A current set of candidate diseases is generated based on the latest symptom information;

[0011] For each candidate disease in the current candidate disease set, a Bayesian algorithm is used to calculate the posterior probability of the candidate disease after this round of consultation based on the latest symptom information;

[0012] Based on the posterior probability of each candidate disease in the current candidate disease set, the uncertainty entropy of the current candidate disease set is calculated, whereby the uncertainty entropy represents the magnitude of the uncertainty in the disease diagnosis result for the current candidate disease set.

[0013] Based on the relationship between the uncertainty entropy and the preset entropy threshold, it is determined whether to end the intelligent consultation.

[0014] Furthermore, generating the current candidate disease set based on the latest symptom information includes:

[0015] When this round of consultation is the first round of consultation, an initial set of candidate diseases is generated based on the chief complaint provided by the patient in the first round of consultation, which serves as the current set of candidate diseases. The chief complaint is the latest symptom information in the first round of consultation.

[0016] When this round of consultation is not the first round of consultation, based on the latest symptom information, the current candidate disease set remains unchanged, or the current candidate disease set is updated by removing specific candidate diseases from the current candidate disease set to obtain the latest current candidate disease set.

[0017] Furthermore, for each candidate disease in the current candidate disease set, a Bayesian algorithm is used to calculate the posterior probability of the candidate disease after this round of consultation based on the latest symptom information, including:

[0018] For each candidate disease in the current candidate disease set, based on the latest symptom information, the posterior probability of that candidate disease after this round of consultation is calculated using the following formula:

[0019] ;

[0020] in, Indicate candidate diseases The posterior probability following this round of consultation. Indicate candidate diseases The posterior probability after the previous round of consultation, The initial value is determined based on the prior distribution of epidemiological data. This indicates that if the patient has a candidate disease symptom information Probability, symptom information The symptom information provided to the patient during this round of consultation. The number of candidate diseases included in the current candidate disease set.

[0021] Furthermore, calculating the uncertainty entropy of the current candidate disease set based on the posterior probability of each candidate disease in the current candidate disease set includes:

[0022] Based on the posterior probability of each candidate disease in the current candidate disease set, the uncertainty entropy of the current candidate disease set is calculated using the following formula:

[0023] ;

[0024] in, Represents the current set of candidate diseases Uncertainty entropy, Indicate candidate diseases The posterior probability following this round of consultation. For the current candidate disease set The number of candidate diseases included.

[0025] Furthermore, it also includes:

[0026] When it is determined that the next round of consultation needs to be initiated based on the relationship between the uncertainty entropy and the preset entropy threshold, a current set of candidate questions is generated;

[0027] For each candidate question in the current candidate question set, an information gain score is calculated based on the uncertainty entropy and the conditional entropy of the current candidate disease set for that candidate question. The conditional entropy of the current candidate disease set for that candidate question represents the uncertainty of the disease diagnosis results for the current candidate disease set after the candidate question is raised. The larger the information gain score of the candidate question, the greater the degree to which the uncertainty of the disease diagnosis results for the current candidate disease set can be reduced after the candidate question is raised.

[0028] Based on the information gain score of each candidate question in the current candidate question set, the questions to be asked in the next round of consultation are determined.

[0029] Furthermore, for each candidate question in the current candidate question set, the information gain score for that candidate question is calculated based on the uncertainty entropy and the conditional entropy of the current candidate disease set for that candidate question, including:

[0030] For each candidate question in the current candidate question set, based on the uncertainty entropy and the conditional entropy of the current candidate disease set for that candidate question, the information gain score of that candidate question is calculated using the following formula:

[0031] ;

[0032] in, For candidate problems Information gain score Represents the current set of candidate diseases Uncertainty entropy, Indicating a question about candidates The current candidate disease set Conditional entropy;

[0033] ;

[0034] in, Indicating a question about candidates The answer set For the answer set The answer in the middle, Indicating a question about candidates Get the answer The predicted probability, To address the candidate question Get the answer The current candidate disease set described later The posterior entropy represents the entropy for the candidate problem. Get the answer The current candidate disease set described later The degree of uncertainty in the diagnosis of the disease;

[0035] ;

[0036] ;

[0037] ;

[0038] in, Indicate candidate diseases The posterior probability after the previous round of consultation, This indicates that the patient does indeed have a candidate disease. Ask the patient candidate questions The answer was obtained later. The probability of.

[0039] Furthermore, before determining the questions to be asked in the next round of consultation based on the information gain scores of each candidate question in the current candidate question set, the process also includes:

[0040] Based on whether the information gain score of each candidate question in the current candidate question set is less than a preset score threshold, it is determined whether to end the intelligent consultation.

[0041] When it is determined that the next round of consultation needs to be initiated, the step of determining the questions to be asked in the next round of consultation is performed based on the information gain score of each candidate question in the current candidate question set.

[0042] Furthermore, it also includes:

[0043] After the intelligent consultation ends, the chief complaint, present medical history and past medical history are identified from all the symptom information provided by the patient during the intelligent consultation. The chief complaint, present medical history and past medical history contain multiple colloquial expressions from the patient.

[0044] For each of the multiple colloquial expressions, based on the similarity between the colloquial expression and each medical terminology expression, the colloquial expression is converted into the corresponding medical terminology expression.

[0045] Based on the converted medical terminology, the medical record for this intelligent consultation is generated. The medical record includes the converted chief complaint, present illness, and past medical history.

[0046] This application also provides an intelligent medical consultation device, including:

[0047] The symptom information acquisition module is used to acquire the symptom information provided by the patient in this round of consultation, and the symptom information is used as the latest symptom information;

[0048] The disease set generation module is used to generate a current candidate disease set based on the latest symptom information;

[0049] The probability calculation module is used to calculate the posterior probability of each candidate disease in the current candidate disease set after this round of consultation based on the latest symptom information using a Bayesian algorithm.

[0050] The entropy calculation module is used to calculate the uncertainty entropy of the current candidate disease set based on the posterior probability of each candidate disease in the current candidate disease set, wherein the uncertainty entropy represents the magnitude of the uncertainty of the disease diagnosis result for the current candidate disease set;

[0051] The consultation and judgment module is used to determine whether to end the intelligent consultation based on the relationship between the uncertainty entropy and the preset entropy threshold.

[0052] Furthermore, the disease set generation module is specifically used to generate an initial candidate disease set based on the chief complaint provided by the patient in the first round of consultation when the current consultation is the first round of consultation, and to serve as the current candidate disease set. The chief complaint is the latest symptom information in the first round of consultation.

[0053] When this round of consultation is not the first round of consultation, based on the latest symptom information, the current candidate disease set remains unchanged, or the current candidate disease set is updated by removing specific candidate diseases from the current candidate disease set to obtain the latest current candidate disease set.

[0054] Furthermore, the probability calculation module is specifically used to calculate the posterior probability of each candidate disease in the current candidate disease set after this round of consultation, based on the latest symptom information, using the following formula:

[0055] ;

[0056] in, Indicate candidate diseases The posterior probability following this round of consultation. Indicate candidate diseases The posterior probability after the previous round of consultation, The initial value is determined based on the prior distribution of epidemiological data. This indicates that if the patient has a candidate disease symptom information Probability, symptom information The symptom information provided to the patient during this round of consultation. The number of candidate diseases included in the current candidate disease set.

[0057] Furthermore, the entropy calculation module is specifically used to calculate the uncertainty entropy of the current candidate disease set based on the posterior probability of each candidate disease in the current candidate disease set, using the following formula:

[0058] ;

[0059] in, Represents the current set of candidate diseases uncertainty entropy Indicate candidate diseases The posterior probability following this round of consultation. For the current candidate disease set The number of candidate diseases included.

[0060] Furthermore, it also includes:

[0061] The question set generation module is used to generate the current candidate question set when it is determined that the next round of consultation needs to be initiated based on the relationship between the uncertainty entropy and the preset entropy threshold.

[0062] The scoring calculation module is used to calculate the information gain score of each candidate question in the current candidate question set based on the uncertainty entropy and the conditional entropy of the current candidate disease set for that candidate question. The conditional entropy of the current candidate disease set for that candidate question represents the uncertainty of the disease diagnosis results for the current candidate disease set after the candidate question is raised. The larger the information gain score of the candidate question, the greater the degree to which the uncertainty of the disease diagnosis results for the current candidate disease set can be reduced after the candidate question is raised.

[0063] The consultation and judgment module is also used to determine the questions to be asked in the next round of consultation based on the information gain score of each candidate question in the current candidate question set.

[0064] Furthermore, the scoring calculation module is specifically used to calculate the information gain score of each candidate question in the current candidate question set, based on the uncertainty entropy and the conditional entropy of the current candidate disease set for that candidate question, using the following formula:

[0065] ;

[0066] in, For candidate problems Information gain score Represents the current set of candidate diseases uncertainty entropy Indicating a question about candidates The current candidate disease set Conditional entropy;

[0067] ;

[0068] in, Indicating a question about candidates The answer set For the answer set The answer in the middle, Indicating a question about candidates Get the answer The predicted probability, To address the candidate question Get the answer The current candidate disease set described later The posterior entropy represents the entropy for the candidate problem. Get the answer The current candidate disease set described later The degree of uncertainty in the diagnosis of the disease;

[0069] ;

[0070] ;

[0071] ;

[0072] in, Indicate candidate diseases The posterior probability after the previous round of consultation, This indicates that the patient does indeed have a candidate disease. Ask the patient candidate questions The answer was obtained later. The probability of.

[0073] Furthermore, the consultation judgment module is also used to determine whether to end the intelligent consultation before determining the questions to be asked in the next round of consultation based on the information gain scores of each candidate question in the current candidate question set. This is done by checking whether the information gain scores of each candidate question in the current candidate question set are all less than a preset score threshold. When it is determined that the next round of consultation needs to be started, the step of determining the questions to be asked in the next round of consultation based on the information gain scores of each candidate question in the current candidate question set is executed.

[0074] Furthermore, it also includes:

[0075] The content recognition module is used to identify the chief complaint, present medical history and past medical history from all the symptom information provided by the patient during the intelligent consultation process after the consultation ends. The chief complaint, present medical history and past medical history contain multiple colloquial expressions of the patient.

[0076] The expression conversion module is used to convert each of the multiple colloquial expressions into a corresponding medical terminology expression based on the similarity between the colloquial expression and each medical terminology expression.

[0077] The medical record generation module is used to generate the medical record for this intelligent consultation based on multiple medical terminology expressions obtained after conversion. The medical record includes the converted chief complaint, present illness, and past medical history.

[0078] This application also provides an electronic device, including a processor and a machine-readable storage medium, wherein the machine-readable storage medium stores machine-executable instructions that can be executed by the processor, and the processor is prompted by the machine-executable instructions to implement any of the above-described intelligent consultation methods.

[0079] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the intelligent consultation methods described above.

[0080] This application also provides a computer program product containing instructions that, when run on a computer, cause the computer to execute any of the intelligent consultation methods described above.

[0081] The beneficial effects of this application include:

[0082] In the method provided in this application embodiment, after each round of consultation, the symptom information provided by the patient in this round of consultation is obtained as the latest symptom information, and a current candidate disease set is generated based on the latest symptom information. For each candidate disease, a Bayesian algorithm is used to calculate the posterior probability of the candidate disease after this round of consultation, and the uncertainty entropy of the current candidate disease set is further calculated. The uncertainty entropy can represent the uncertainty of the disease diagnosis result for the current candidate disease set. Based on the relationship between the uncertainty entropy and the preset entropy threshold, it is determined whether to end the intelligent consultation. That is, it realizes the determination of whether to end the intelligent consultation based on the uncertainty of information and result as a relatively objective termination condition, thereby ensuring the integrity of the consultation while improving the consultation efficiency, and effectively balancing the consultation efficiency and consultation integrity in intelligent consultation.

[0083] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0084] The accompanying drawings are provided to further illustrate the present application and form part of the specification. They are used together with the embodiments of the present application to explain the application and do not constitute a limitation thereof. In the drawings:

[0085] Figure 1 A flowchart of an intelligent consultation method is provided for embodiments of this application;

[0086] Figure 2 This application provides a flowchart for determining the next round of consultation questions in an intelligent consultation method;

[0087] Figure 3 A flowchart of an intelligent consultation method is provided for another embodiment of this application;

[0088] Figure 4 This application provides a flowchart for generating medical records in an intelligent consultation method.

[0089] Figure 5 A schematic diagram of the structure of the intelligent consultation device is provided for the embodiments of this application;

[0090] Figure 6 A schematic diagram of the structure of an intelligent consultation device is provided for another embodiment of this application;

[0091] Figure 7 A schematic diagram of the structure of an intelligent consultation device is provided for another embodiment of this application;

[0092] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0093] To provide an effective solution for balancing consultation efficiency and completeness in intelligent consultation, this application provides an intelligent consultation method, apparatus, and electronic device. The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustrative and explanatory purposes only and are not intended to limit this application. Furthermore, the embodiments and features described in this application can be combined with each other unless otherwise specified.

[0094] This application provides an intelligent consultation method, such as... Figure 1 As shown, it includes:

[0095] Step 11: Obtain the symptom information provided by the patient in this round of consultation; this symptom information will be used as the latest symptom information.

[0096] Step 12: Generate a current set of candidate diseases based on the latest symptom information;

[0097] Step 13: For each candidate disease in the current candidate disease set, use the Bayesian algorithm to calculate the posterior probability of the candidate disease after this round of consultation based on the latest symptom information;

[0098] Step 14: Based on the posterior probability of each candidate disease in the current candidate disease set, calculate the uncertainty entropy of the current candidate disease set. The uncertainty entropy represents the magnitude of the uncertainty in the diagnosis results for the current candidate disease set.

[0099] Step 15: Based on the relationship between the uncertainty entropy and the preset entropy threshold, determine whether to end the intelligent consultation.

[0100] The intelligent consultation method provided in this application calculates the posterior probability of candidate diseases in the current candidate disease set after this round of consultation based on the latest symptom information provided by the patient. It further calculates the uncertainty entropy of the current candidate disease set and determines whether to end the intelligent consultation based on the relationship between the uncertainty entropy and the preset entropy threshold. This realizes the determination of whether to end the intelligent consultation based on the uncertainty of information and results as a relatively objective termination condition. Thus, it ensures the integrity of the consultation while improving the consultation efficiency, effectively balancing the consultation efficiency and consultation integrity in intelligent consultation.

[0101] In one embodiment of this application, when it is determined that the next round of consultation needs to be initiated based on the relationship between the calculated uncertainty entropy of the current candidate disease set and a preset entropy threshold, such as... Figure 2 As shown, the following steps can be used to determine the questions to be asked in the next round of consultation:

[0102] Step 21: Generate the current candidate question set, which contains all candidate questions for the next round of consultation;

[0103] Step 22: For each candidate question in the current candidate question set, calculate the information gain score of the candidate question based on the uncertainty entropy and the conditional entropy of the current candidate disease set for the candidate question. The conditional entropy of the current candidate disease set for the candidate question represents the uncertainty of the disease diagnosis results for the current candidate disease set after the candidate question is proposed. The larger the information gain score of the candidate question, the greater the degree to which the uncertainty of the disease diagnosis results for the current candidate disease set can be reduced after the candidate question is proposed.

[0104] Step 23: Based on the information gain score of each candidate question in the current candidate question set, determine the questions to be asked in the next round of consultation.

[0105] In the above embodiments of this application, for each candidate question in the current candidate set, the larger the information gain score of the candidate question, the greater the degree to which the uncertainty of the disease diagnosis results for the current candidate disease set can be reduced after the candidate question is raised. Therefore, based on the information gain scores of each candidate question in the current candidate question set, the questions to be raised in the next round of consultation can be determined, which can gradually reduce the uncertainty of the disease diagnosis results for the current candidate disease set. As multiple rounds of consultation are carried out, the disease diagnosis results for the patient tend to be clearer and more certain, which helps to improve the relevance of the questions in each round of consultation, reduce the number of invalid consultations, and thus improve consultation efficiency.

[0106] The method and apparatus provided in this application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0107] In one embodiment of this application, an intelligent consultation method is provided, such as... Figure 3 As shown, it includes:

[0108] Step 31: Obtain the symptom information provided by the patient in this round of consultation. This symptom information will be used as the latest symptom information.

[0109] After the intelligent consultation is initiated for a patient, the symptom information provided by the patient in the first round of consultation is the chief complaint. The chief complaint is a general description of the patient's condition, and the disease category to which the patient's condition belongs can often be initially determined through the chief complaint.

[0110] Step 32: Generate a current set of candidate diseases based on the latest symptom information provided by the patient in this round of consultation.

[0111] In this step, when this round of consultation is the first round of consultation, an initial set of candidate diseases can be generated based on the chief complaint provided by the patient in the first round of consultation, which will serve as the current set of candidate diseases. The chief complaint is the latest symptom information from the first round of consultation.

[0112] In practical applications, the chief complaint can be directly input into a pre-trained big oracle model. The big oracle model can then generate an initial set of candidate diseases based on the chief complaint, which will serve as the current set of candidate diseases.

[0113] Alternatively, the main complaint can be segmented first to obtain multiple segments, and medical-related words can be identified from them. These medical-related words can then be input into a pre-trained big oracle model to obtain the corresponding disease category. Further search can be conducted for candidate diseases included in that disease category to form an initial candidate disease set.

[0114] When this round of consultation is not the first round of consultation, the current candidate disease set can be kept unchanged based on the latest symptom information, or the current candidate disease set can be updated by removing specific candidate diseases from the current candidate disease set to obtain the latest current candidate disease set;

[0115] In practical applications, this can also be achieved through a big oracle model. For example, the latest symptom information can be directly input into the big oracle model, or the latest symptom information can be segmented into words, and the resulting medical-related words can be input into the big oracle model to achieve dynamic updates of the current candidate disease set and obtain the latest current candidate disease set.

[0116] Dynamic updates to the candidate disease set can reduce the number of candidate diseases, thereby gradually improving the certainty of disease diagnosis results through multiple rounds of consultation.

[0117] In this step, the symptom information on which the Big Prophecy model infers candidate diseases must be strictly derived from information provided by patients; model speculation or rewriting is not allowed.

[0118] In this embodiment of the application, after each acquisition of the latest symptom information provided by the patient, it can be appended to the symptom record and accumulated and saved in a structured format.

[0119] Step 33: For each candidate disease in the current candidate disease set, use the Bayesian algorithm to calculate the posterior probability of the candidate disease after this round of consultation based on the latest symptom information.

[0120] In this step, specifically, the posterior probability of the candidate disease after this round of consultation can be calculated using the following formula:

[0121] ;

[0122] in, Indicate candidate diseases The posterior probability following this round of consultation. Indicate candidate diseases The posterior probability after the previous round of consultation, The initial value is determined based on the prior distribution of epidemiological data. This indicates that if the patient has a candidate disease symptom information Probability, symptom information The symptom information provided to the patient during this round of consultation. This represents the number of candidate diseases included in the current candidate disease set.

[0123] In practical applications, The likelihood probability can be pre-set based on a medical knowledge base.

[0124] denominator in the formula As a normalization factor, it ensures that the sum of the posterior probabilities of all candidate diseases is 1.

[0125] In the above formula for calculating the posterior probability of a candidate disease, for a candidate disease, the posterior probability after the previous round of consultation is used as the prior probability of the current round of consultation t. Combined with the latest symptom information obtained in the current round of consultation t, the posterior probability of the next round of consultation t+1 is calculated, which is the posterior probability after the current round of consultation, and is used as the updated post-confidence that the patient has the candidate disease.

[0126] Step 34: Calculate the uncertainty entropy of the current candidate disease set based on the posterior probability of each candidate disease in the current candidate disease set.

[0127] In this step, specifically, the uncertainty entropy of the current candidate disease set can be calculated using the following formula:

[0128] ;

[0129] in, Represents the current set of candidate diseases Uncertainty entropy, Indicate candidate diseases The posterior probability following this round of consultation. For the current set of candidate diseases The number of candidate diseases included.

[0130] In this embodiment of the application, the uncertainty entropy calculated for the current candidate disease set represents the magnitude of uncertainty in the disease diagnosis results for the current candidate disease set. The higher the value of the uncertainty entropy, the greater the uncertainty in the disease diagnosis; the lower the value of the uncertainty entropy, the smaller the uncertainty in the disease diagnosis.

[0131] Step 35: Based on the relationship between the uncertainty entropy and the preset entropy threshold, determine whether to end the intelligent consultation.

[0132] Specifically, when the uncertainty quotient is less than the preset entropy threshold, it means that the uncertainty of the disease diagnosis result for the current candidate disease set is small enough that the disease can be basically diagnosed, and then the intelligent consultation ends.

[0133] When the uncertainty entropy is not less than the preset entropy threshold, it means that the uncertainty of the disease diagnosis result for the current candidate disease set is still relatively large, and the disease cannot be diagnosed. It is necessary to continue to obtain the patient's latest symptom information through the next round of consultation. Then it is determined that the next round of consultation needs to be started, and step 36 is executed.

[0134] Step 36: When it is determined that the next round of consultation needs to be initiated, generate the current set of candidate questions.

[0135] In this embodiment of the application, the initial set of candidate questions may include pre-set candidate questions, or it may include candidate questions generated by a pre-trained large oracle model based on the patient's chief complaint.

[0136] When the current consultation is not the first consultation, the questions from the previous consultation can be removed from the candidate question set. Alternatively, new candidate questions can be generated based on the latest symptom information using the big oracle model and added to the candidate question set to obtain the latest current candidate question set.

[0137] Step 37: For each candidate question in the current candidate question set, calculate the information gain score of the candidate question based on the uncertainty entropy and the conditional entropy of the current candidate disease set for that candidate question.

[0138] In this embodiment of the application, the conditional entropy of the current candidate disease set for the candidate question represents the degree of uncertainty in the disease diagnosis results for the current candidate disease set after the candidate question is raised.

[0139] The higher the information gain score of a candidate question, the greater the degree to which the uncertainty of the diagnosis results for the current candidate disease set can be reduced after proposing the candidate question. Conversely, the lower the information gain score, the less the degree to which the uncertainty of the diagnosis results for the current candidate disease set can be reduced after proposing the candidate question.

[0140] In this step, the information gain score of the candidate problem can be calculated using the following formula:

[0141] ;

[0142] in, For candidate problems Information gain score Represents the current set of candidate diseases Uncertainty entropy, Indicating a question about candidates Current candidate disease set The conditional entropy, also known as the expected conditional entropy, represents the conditional entropy at which the candidate problem is posed. Then, for the current set of candidate diseases, how much uncertainty is expected in the diagnosis of the remaining diseases?

[0143] ;

[0144] in, Indicating a question about candidates The set of answers, for example, the set of answers in general. The answers that can be included include: yes, no, not sure. For the answer set One of the answers, for example, That is true;

[0145] Indicating a question about candidates Get the answer The predicted probability, that is, given the current known information, the patient's probability of answering the candidate question. answer How likely is it?

[0146] To address the candidate question Get the answer Post-current candidate disease set The posterior entropy represents the entropy for the candidate problem. Get the answer Post-current candidate disease set The magnitude of uncertainty in a disease diagnosis, i.e., the assumption that the patient is facing a candidate problem. answer The current disease set is then recalculated. Uncertainty entropy;

[0147] ;

[0148] ;

[0149] ;

[0150] in, Indicate candidate diseases The posterior probability after the previous round of consultation (equivalent to the probability in step 33 above). ), This indicates that the patient does indeed have a candidate disease. Ask the patient candidate questions The answer was obtained later. The probability is the medical prior likelihood probability, which can be pre-set based on the medical knowledge base.

[0151] Since the information gain score of a candidate question can represent the degree to which the uncertainty of the diagnosis result for the current candidate disease set can be reduced after the candidate question is proposed, the questions to be proposed in the next round of consultation can be determined based on the information gain score of each candidate question in the current candidate question set.

[0152] Furthermore, before determining the questions to be asked in the next round of consultations, the following step 38 can be performed first.

[0153] Step 38: Based on the information gain scores of each candidate question in the current candidate question set, determine whether the intelligent consultation should end if all scores are less than the preset score threshold.

[0154] Specifically, if all scores are below the preset scoring threshold, the intelligent frame is terminated; if none scores are below the preset scoring threshold, the next round of consultation needs to be initiated, and step 39 is executed.

[0155] When all scores are below the preset scoring threshold, it means that the degree to which all candidate questions can reduce the uncertainty of the diagnosis results for the current candidate disease set is very small. Therefore, there is no need to select questions for the next round of consultation from the current candidate question set, and it is possible to end the intelligent consultation.

[0156] In this step, based on the information gain score of each candidate question in the current candidate question set, it is determined whether to end the intelligent consultation. Furthermore, if continuing the consultation may not yield more effective symptom information, unnecessary multiple rounds of consultation are avoided, the number of ineffective consultations is reduced, and the consultation efficiency is improved. Moreover, if it is necessary to continue the consultation to improve the certainty of the disease diagnosis result, the consultation is not terminated prematurely, ensuring the integrity of the consultation. In other words, the consultation efficiency and consultation integrity in intelligent consultation are further effectively balanced.

[0157] Step 39: Based on the information gain score of each candidate question in the current candidate question set, determine the questions to be asked in the next round of consultation.

[0158] In this step, the candidate question with the highest information gain score can be selected from the current set of candidate questions as the question to be asked in the next round of consultation.

[0159] After initiating the next round of consultation based on the identified issues, you can return to step 31 above to proceed with the processing flow for the next round of consultation.

[0160] In this embodiment of the application, based on the above-described intelligent consultation method, a further step is to generate a medical record for this intelligent consultation after it ends, such as... Figure 4 As shown, it may include the following steps:

[0161] Step 41: After the intelligent consultation ends, identify the chief complaint, present medical history and past medical history from all the symptom information provided by the patient during the intelligent consultation.

[0162] In this step, the symptom information provided by the patient in the first round of consultation is the main complaint. In subsequent rounds of consultation, the symptom information provided by the patient in each round of consultation can be input into a pre-trained big oracle model, which will then identify medical history and past medical history.

[0163] Since the symptom information is provided by the patient, the identified chief complaint, present medical history, and past medical history contain multiple colloquial expressions from the patient.

[0164] Step 42: For each of the multiple colloquial expressions, based on the similarity between the colloquial expression and each medical terminology expression, convert the colloquial expression into the corresponding medical terminology expression.

[0165] In this embodiment of the application, the medical terminology expressions can be pre-generated based on a medical information database.

[0166] In this step, the similarity between a colloquial expression and a medical terminology expression can be calculated using the following formula:

[0167] ;

[0168] in, The text embedding vector for the patient's colloquial expressions. For medical terminology embedding vectors, Text embedding vector Embedded vectors of medical terms The similarity between them also indicates the similarity between the colloquial expression and the medical terminology expression.

[0169] Text embedding vector It can be obtained by using a pre-built word vector model to map colloquial text into high-dimensional vectors.

[0170] In this step, for a colloquial expression, the medical terminology expression with the highest similarity can be selected, and the colloquial expression can be converted into the corresponding medical terminology expression.

[0171] In practical applications, it is possible that for a given colloquial expression, all medical terminological expressions have low similarity, for example, all below the similarity threshold. In this case, the patient's original colloquial expression can be retained unchanged.

[0172] Step 43: Based on the multiple medical terminology expressions obtained after conversion, generate the medical record for this intelligent consultation. The medical record includes the converted chief complaint, present illness, and past medical history.

[0173] In this step, the medical records are output according to a fixed three-part structure, as follows:

[0174] Chief complaint: Extract the patient's core symptoms and their duration;

[0175] Present illness history: Based on the timeline, summarize the evolution of symptoms and key negative physical signs;

[0176] Past medical history: List important medical history, medication history and allergy history.

[0177] In existing technologies, medical record generation methods based on large models for intelligent consultation often have very obvious colloquial characteristics, making it difficult to achieve structured alignment with hospital electronic medical record (EMR) systems based on specific standards (such as ICD-10), thus limiting their practical implementation in clinical processes.

[0178] In this embodiment of the application, the above-mentioned Figure 4 The medical record generation scheme shown achieves a precise mapping from patient-language language to standard medical terminology, ensuring that the generated medical records comply with clinical document writing standards.

[0179] The intelligent consultation method and medical record generation scheme provided in the embodiments of this application can be implemented based on an intelligent consultation system containing two intelligent agents, one of which is a consultation intelligent agent used to implement the above. Figure 3 The intelligent consultation method flow shown includes another intelligent agent, the medical record generation intelligent agent, used to implement the above. Figure 4 The medical record generation scheme is shown.

[0180] In addition, the intelligent consultation system may also include a collaborative control module, which is used to determine whether to end the intelligent consultation based on the uncertainty entropy of the current candidate disease set and the information gain score of each candidate question in the current candidate question set. After determining to end the intelligent consultation, the collaborative control consultation agent sends the stored symptom records to the medical record generation agent, which then generates the medical record.

[0181] By adopting the above-described solution provided in the embodiments of this application, the following technical effects can be achieved:

[0182] 1. Decoupled Collaborative Architecture Based on Dual Agents: An innovative collaborative architecture is proposed, separating the "diagnosis agent" and the "medical record generation agent." The diagnosis agent focuses on probability-based reasoning and decision-making, while the medical record generation agent focuses on document construction based on semantic norms. This decoupled design resolves the contradiction between "medical reasoning logic" and "text generation norms" that is difficult to balance in traditional single-model architectures, significantly improving the system's robustness.

[0183] 2. Information Gain-Driven Proactive Consultation Strategy: Unlike traditional rule-tree-based consultation, this application introduces information gain as the core indicator for questioning decisions. The intelligent consultation system can dynamically calculate and select the question with the highest discriminative power based on the current diagnostic uncertainty (entropy). This mechanism achieves "goal-oriented" consultation, effectively avoiding ineffective follow-up questions and reducing the average number of consultation rounds by approximately 25%.

[0184] 3. Explainable Chain of Thought (CoT) Externalization Mechanism: This application's embodiments visualize and deconstruct the thinking process of the black box model through an explicit three-stage structure of "fact recording - confidence distribution - questioning strategy". Doctors can clearly trace the system's "why it suspects this disease" and "why it asks this question".

[0185] 4. Automatic Termination Mechanism with Dual Convergence: This application's embodiments construct a dual termination judgment logic based on "diagnostic certainty (entropy threshold)" and "information acquisition value (scoring threshold)". The system can automatically and accurately truncate the consultation process when the probability of diagnosis is high enough or when effective information cannot be obtained, solving the problems of "endless dialogue" or "premature termination" in existing large-scale model consultations.

[0186] 5. Standardized terminology generation based on semantic vectors: To address the issue of patients' colloquial expressions being difficult to directly input into the database, the medical record generation agent introduces a semantic similarity matching algorithm to achieve automatic mapping from "patients' colloquial language" to "standard medical terminology," ensuring that the generated medical records conform to clinical document writing standards.

[0187] Based on the same inventive concept, and according to the intelligent consultation method provided in the above embodiments of this application, another embodiment of this application also provides an intelligent consultation device, the structural schematic diagram of which is shown below. Figure 5 As shown, it specifically includes:

[0188] Symptom information acquisition module 51 is used to acquire symptom information provided by the patient in this round of consultation, and the symptom information is used as the latest symptom information;

[0189] Disease set generation module 52 is used to generate a current candidate disease set based on the latest symptom information;

[0190] The probability calculation module 53 is used to calculate the posterior probability of each candidate disease in the current candidate disease set after this round of consultation based on the latest symptom information using a Bayesian algorithm.

[0191] Entropy calculation module 54 is used to calculate the uncertainty entropy of the current candidate disease set based on the posterior probability of each candidate disease in the current candidate disease set, wherein the uncertainty entropy represents the magnitude of the uncertainty of the disease diagnosis result for the current candidate disease set;

[0192] The consultation judgment module 55 is used to determine whether to end the intelligent consultation based on the relationship between the uncertainty entropy and the preset entropy threshold.

[0193] Furthermore, the disease set generation module 52 is specifically used to generate an initial candidate disease set based on the chief complaint provided by the patient in the first round of consultation when the current consultation is the first round of consultation, and to serve as the current candidate disease set. The chief complaint is the latest symptom information in the first round of consultation.

[0194] When this round of consultation is not the first round of consultation, based on the latest symptom information, the current candidate disease set remains unchanged, or the current candidate disease set is updated by removing specific candidate diseases from the current candidate disease set to obtain the latest current candidate disease set.

[0195] Furthermore, the probability calculation module 53 is specifically used to calculate the posterior probability of each candidate disease in the current candidate disease set after this round of consultation, based on the latest symptom information, using the following formula:

[0196] ;

[0197] in, Indicate candidate diseases The posterior probability following this round of consultation. Indicate candidate diseases The posterior probability after the previous round of consultation, The initial value is determined based on the prior distribution of epidemiological data. This indicates that if the patient has a candidate disease symptom information Probability, symptom information The symptom information provided to the patient during this round of consultation. The number of candidate diseases included in the current candidate disease set.

[0198] Furthermore, the entropy calculation module 54 is specifically used to calculate the uncertainty entropy of the current candidate disease set based on the posterior probability of each candidate disease in the current candidate disease set, using the following formula:

[0199] ;

[0200] in, Represents the current set of candidate diseases Uncertainty entropy, Indicate candidate diseases The posterior probability following this round of consultation. For the current candidate disease set The number of candidate diseases included.

[0201] Furthermore, such as Figure 6 As shown, it also includes:

[0202] The question set generation module 56 is used to generate the current candidate question set when it is determined that the next round of consultation needs to be started based on the relationship between the uncertainty entropy and the preset entropy threshold.

[0203] The scoring calculation module 57 is used to calculate the information gain score of each candidate question in the current candidate question set based on the uncertainty entropy and the conditional entropy of the current candidate disease set for that candidate question. The conditional entropy of the current candidate disease set for that candidate question represents the uncertainty of the disease diagnosis results for the current candidate disease set after the candidate question is raised. The larger the information gain score of the candidate question, the greater the degree to which the uncertainty of the disease diagnosis results for the current candidate disease set can be reduced after the candidate question is raised.

[0204] The consultation and judgment module 55 is also used to determine the questions to be asked in the next round of consultation based on the information gain score of each candidate question in the current candidate question set.

[0205] Furthermore, the scoring calculation module 57 is specifically used to calculate the information gain score of each candidate question in the current candidate question set, based on the uncertainty entropy and the conditional entropy of the current candidate disease set for that candidate question, using the following formula:

[0206] ;

[0207] in, For candidate problems Information gain score Represents the current set of candidate diseases Uncertainty entropy, Indicating a question about candidates The current candidate disease set Conditional entropy;

[0208] ;

[0209] in, Indicating a question about candidates The answer set For the answer set The answer in the middle, Indicating a question about candidates Get the answer The predicted probability, To address the candidate question Get the answer The current candidate disease set described later The posterior entropy represents the entropy for the candidate problem. Get the answer The current candidate disease set described later The degree of uncertainty in the diagnosis of the disease;

[0210] ;

[0211] ;

[0212] ;

[0213] in, Indicate candidate diseases The posterior probability after the previous round of consultation, This indicates that the patient does indeed have a candidate disease. Ask the patient candidate questions The answer was obtained later. The probability of.

[0214] Furthermore, the consultation judgment module 55 is also used to determine whether to end the intelligent consultation before determining the questions to be asked in the next round of consultation based on the information gain scores of each candidate question in the current candidate question set. This is done by checking whether the information gain scores of each candidate question in the current candidate question set are all less than a preset score threshold. When it is determined that the next round of consultation needs to be started, the step of determining the questions to be asked in the next round of consultation based on the information gain scores of each candidate question in the current candidate question set is executed.

[0215] Furthermore, such as Figure 7 As shown, it also includes:

[0216] The content recognition module 58 is used to identify the chief complaint, present medical history and past medical history from all the symptom information provided by the patient during the intelligent consultation after the consultation ends. The chief complaint, present medical history and past medical history contain multiple colloquial expressions of the patient.

[0217] The expression conversion module 59 is used to convert each of the multiple colloquial expressions into a corresponding medical terminology expression based on the similarity between the colloquial expression and each medical terminology expression.

[0218] The medical record generation module 510 is used to generate the medical record for this intelligent consultation based on the multiple medical terminology expressions obtained after conversion. The medical record includes the converted chief complaint, present illness, and past medical history.

[0219] The functions of the above modules can be corresponding to Figures 1 to 4 The corresponding processing steps in the process shown will not be repeated here.

[0220] The intelligent consultation device provided in the embodiments of this application can be implemented through a computer program. Those skilled in the art should understand that the above-described module division method is only one of many module division methods. Whether it is divided into other modules or not divided into modules, as long as the intelligent consultation device has the above-described functions, it should be within the protection scope of this application.

[0221] This application also provides an electronic device, such as... Figure 8 As shown, it includes a processor 81 and a machine-readable storage medium 82, the machine-readable storage medium 82 storing machine-executable instructions that can be executed by the processor 81, the processor 81 being prompted by the machine-executable instructions to implement any of the intelligent consultation methods described above.

[0222] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the intelligent consultation methods described above.

[0223] This application also provides a computer program product containing instructions that, when run on a computer, cause the computer to execute any of the intelligent consultation methods described above.

[0224] The machine-readable storage medium in the aforementioned electronic device may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0225] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0226] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of devices, electronic devices, computer-readable storage media, and computer program products are basically similar to the method embodiments, and therefore the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0227] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0228] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0229] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0230] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0231] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. An intelligent consultation method, characterized in that, include: Obtain the symptom information provided by the patient in this round of consultation, and use the symptom information as the latest symptom information; A current set of candidate diseases is generated based on the latest symptom information; For each candidate disease in the current candidate disease set, a Bayesian algorithm is used to calculate the posterior probability of the candidate disease after this round of consultation based on the latest symptom information; Based on the posterior probability of each candidate disease in the current candidate disease set, the uncertainty entropy of the current candidate disease set is calculated, whereby the uncertainty entropy represents the magnitude of the uncertainty in the disease diagnosis result for the current candidate disease set. Based on the relationship between the uncertainty entropy and the preset entropy threshold, it is determined whether to end the intelligent consultation.

2. The method as described in claim 1, characterized in that, The process of generating the current candidate disease set based on the latest symptom information includes: When this round of consultation is the first round of consultation, an initial set of candidate diseases is generated based on the chief complaint provided by the patient in the first round of consultation, which serves as the current set of candidate diseases. The chief complaint is the latest symptom information in the first round of consultation. When this round of consultation is not the first round of consultation, based on the latest symptom information, the current candidate disease set remains unchanged, or the current candidate disease set is updated by removing specific candidate diseases from the current candidate disease set to obtain the latest current candidate disease set.

3. The method as described in claim 1, characterized in that, For each candidate disease in the current candidate disease set, a Bayesian algorithm is used to calculate the posterior probability of that candidate disease after this round of consultation based on the latest symptom information, including: For each candidate disease in the current candidate disease set, based on the latest symptom information, the posterior probability of that candidate disease after this round of consultation is calculated using the following formula: ； in, Indicate candidate diseases The posterior probability following this round of consultation. Indicate candidate diseases The posterior probability after the previous round of consultation, The initial value is determined based on the prior distribution of epidemiological data. This indicates that if the patient has a candidate disease symptom information Probability, symptom information The symptom information provided to the patient during this round of consultation. The number of candidate diseases included in the current candidate disease set.

4. The method as described in claim 1, characterized in that, The step of calculating the uncertainty entropy of the current candidate disease set based on the posterior probability of each candidate disease in the current candidate disease set includes: Based on the posterior probability of each candidate disease in the current candidate disease set, the uncertainty entropy of the current candidate disease set is calculated using the following formula: ； in, Represents the current set of candidate diseases uncertainty entropy Indicate candidate diseases The posterior probability following this round of consultation. For the current candidate disease set The number of candidate diseases included.

5. The method as described in claim 1, characterized in that, Also includes: When it is determined that the next round of consultation needs to be initiated based on the relationship between the uncertainty entropy and the preset entropy threshold, a current set of candidate questions is generated; For each candidate question in the current candidate question set, an information gain score is calculated based on the uncertainty entropy and the conditional entropy of the current candidate disease set for that candidate question. The conditional entropy of the current candidate disease set for that candidate question represents the uncertainty of the disease diagnosis results for the current candidate disease set after the candidate question is raised. The larger the information gain score of the candidate question, the greater the degree to which the uncertainty of the disease diagnosis results for the current candidate disease set can be reduced after the candidate question is raised. Based on the information gain score of each candidate question in the current candidate question set, the questions to be asked in the next round of consultation are determined.

6. The method as described in claim 5, characterized in that, For each candidate question in the current candidate question set, the information gain score of the candidate question is calculated based on the uncertainty entropy and the conditional entropy of the current candidate disease set for that candidate question, including: For each candidate question in the current candidate question set, based on the uncertainty entropy and the conditional entropy of the current candidate disease set for that candidate question, the information gain score of that candidate question is calculated using the following formula: ； in, For candidate problems Information gain score Represents the current set of candidate diseases uncertainty entropy Indicating a question about candidates The current candidate disease set Conditional entropy; ； in, Indicating a question about candidates The answer set For the answer set The answer in the middle, Indicating a question about candidates Get the answer The predicted probability, To address the candidate question Get the answer The current candidate disease set described later The posterior entropy represents the entropy for the candidate problem. Get the answer The current candidate disease set described later The degree of uncertainty in the diagnosis of the disease; ； ； ； in, Indicate candidate diseases The posterior probability after the previous round of consultation, This indicates that the patient does indeed have a candidate disease. Ask the patient candidate questions The answer was obtained later. The probability of.

7. The method as described in claim 5, characterized in that, Before determining the questions to be asked in the next round of consultation based on the information gain scores of each candidate question in the current candidate question set, the procedure further includes: Based on whether the information gain score of each candidate question in the current candidate question set is less than a preset score threshold, it is determined whether to end the intelligent consultation. When it is determined that the next round of consultation needs to be initiated, the step of determining the questions to be asked in the next round of consultation is performed based on the information gain score of each candidate question in the current candidate question set.

8. The method as described in claim 1, characterized in that, Also includes: After the intelligent consultation ends, the chief complaint, present medical history and past medical history are identified from all the symptom information provided by the patient during the intelligent consultation. The chief complaint, present medical history and past medical history contain multiple colloquial expressions from the patient. For each of the multiple colloquial expressions, based on the similarity between the colloquial expression and each medical terminology expression, the colloquial expression is converted into the corresponding medical terminology expression. Based on the converted medical terminology, the medical record for this intelligent consultation is generated. The medical record includes the converted chief complaint, present illness, and past medical history.

9. An intelligent consultation device, characterized in that, include: The symptom information acquisition module is used to acquire the symptom information provided by the patient in this round of consultation, and the symptom information is used as the latest symptom information; The disease set generation module is used to generate a current candidate disease set based on the latest symptom information; The probability calculation module is used to calculate the posterior probability of each candidate disease in the current candidate disease set after this round of consultation based on the latest symptom information using a Bayesian algorithm. The entropy calculation module is used to calculate the uncertainty entropy of the current candidate disease set based on the posterior probability of each candidate disease in the current candidate disease set, wherein the uncertainty entropy represents the magnitude of the uncertainty of the disease diagnosis result for the current candidate disease set; The consultation and judgment module is used to determine whether to end the intelligent consultation based on the relationship between the uncertainty entropy and the preset entropy threshold.

10. An electronic device, characterized in that, The method includes a processor and a machine-readable storage medium storing machine-executable instructions that can be executed by the processor, the processor being prompted by the machine-executable instructions to perform the method of any one of claims 1-8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in any one of claims 1-8.

Images