Method, device, electronic device and readable storage medium for identifying medical consultation intention

Abstract

The present invention relates to artificial intelligence technology, and discloses a method for identifying the intention of a medical consultation, including: obtaining the initial medical consultation information of the medical consultation user, identifying the intention of the medical consultation user based on the initial medical consultation information, obtaining the intention identification result, identifying the symptoms of the initial medical consultation information based on the intention identification result, obtaining the symptom identification result and performing horizontal expansion and vertical expansion, obtaining the horizontal expansion result and the vertical expansion result, conducting multiple rounds of dialogues with the medical consultation user based on the horizontal expansion result and the vertical expansion result, obtaining standard medical consultation information, using a preset joint interactive network to jointly interact with the standard medical consultation information and the intention identification result, and outputting the identified medical consultation intention. In addition, the present invention also relates to blockchain technology, and the initial medical consultation information can be obtained from the nodes of the blockchain. The present invention also proposes a medical consultation intention identification device, an electronic device, and a readable storage medium. The present invention can solve the problem of inaccurate user intention identification.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method, device, electronic device and readable storage medium for identifying medical consultation intention. Background Art

[0002] With the development of science and technology, AI technology has been widely used in the field of intelligent health consultation. Online consultation platforms and some apps for TCM consultation have emerged. When it is inconvenient to see a doctor, you can complete the consultation online. However, the first but extremely important step in the existing online TCM consultation: collecting the main complaint, the accuracy of the main complaint recognition cannot achieve the ideal effect.

[0003] In the scenario of online TCM consultation, the collection of chief complaints refers to the collection of some information about the user's consultation intention, including symptoms, symptom severity, and other information. The commonly used method is to collect chief complaints based on specific rule words, using the voice or text information input by the patient to determine whether there are rule symptom words given by TCM experts in the input information. However, since the information input by the patient is more colloquial, and the given rule words are more formal TCM common symptom words, this results in the input information not being able to match the rule words. At the same time, the number of given rule words is limited and the coverage is not comprehensive, which leads to a low recognition rate of the chief complaint intention. In addition, unlike offline consultations, the amount of information input by online patients in a single round is limited, and effective chief complaint information cannot be collected, which also makes the collected patient intentions and symptoms inaccurate. Summary of the invention

[0004] The present invention provides a method, device, electronic device and readable storage medium for identifying medical consultation intention, the main purpose of which is to solve the problem of inaccurate user intention identification.

[0005] To achieve the above object, the present invention provides a method for identifying medical inquiry intention, comprising:

[0006] Acquire initial consultation information of the consultation user, and perform intention recognition on the consultation user based on the initial consultation information to obtain an intention recognition result;

[0007] Performing symptom recognition on the initial medical inquiry information based on the intention recognition result to obtain a symptom recognition result;

[0008] Expanding the symptom recognition result horizontally and vertically to obtain a horizontal expansion result and a vertical expansion result;

[0009] Conduct multiple rounds of dialogue with the consultation user based on the horizontal expansion results and the vertical expansion results to obtain standard consultation information;

[0010] The preset joint interaction network is used to perform joint interaction on the standard medical inquiry information and the intention recognition result, and the recognized medical inquiry intention is output.

[0011] Optionally, the performing intention recognition on the user asking the question based on the initial consultation information to obtain an intention recognition result includes:

[0012] Using a preset intention classification model to perform text encoding on the initial medical consultation information to obtain a medical consultation encoding vector;

[0013] The intention classification model is used to perform intention classification on the medical inquiry coding vector, and the result of the intention classification is used as the intention recognition result.

[0014] Optionally, performing symptom recognition on the initial medical inquiry information based on the intention recognition result to obtain a symptom recognition result includes:

[0015] Using a preset word segmentation algorithm, the initial medical inquiry information corresponding to the medical inquiry user whose intention recognition result is medical inquiry is processed by word segmentation to obtain the medical inquiry word segmentation text;

[0016] Using a preset medical inquiry vocabulary list to preliminarily screen the medical inquiry word segmentation text to obtain a word segmentation screening text;

[0017] The segmented texts in the segmented screening text are screened for importance, and the segmented texts after importance screening are used as the symptom recognition results.

[0018] Optionally, the importance screening of the segmented texts in the segmented screening text, and using the segmented texts after importance screening as the symptom identification result, includes:

[0019] Finding the frequency of occurrence of the segmented text in the initial medical consultation information;

[0020] Calculate the inverse document frequency of the segmented text in the initial medical inquiry information;

[0021] The importance of all the segmented texts is calculated based on the occurrence frequency and the inverse document frequency, and a preset number of segmented texts are selected as the symptom recognition result based on the importance.

[0022] Optionally, the symptom recognition result is expanded horizontally and vertically to obtain a horizontal expansion result and a vertical expansion result, including:

[0023] Obtaining a preset horizontal question template and a vertical question template, and filling the symptom recognition result into the horizontal question template and the vertical question template;

[0024] The filled horizontal question template is used as the horizontal expansion result, and the filled vertical question template is used as the vertical expansion result.

[0025] Optionally, the step of conducting multiple rounds of dialogue with the consultation user based on the horizontal expansion result and the vertical expansion result to obtain standard consultation information includes:

[0026] Using the horizontal expansion results as the first round of supplementary questioning, and using the vertical expansion results as the second round of supplementary questioning;

[0027] Receive the consultation information provided by the user based on the first round of supplementary questions and the second round of supplementary questions, and return to the step of obtaining the initial consultation information of the user until the number of conversations meets a preset threshold, and obtain the symptom information corresponding to all supplementary questions as the standard consultation information.

[0028] Optionally, the using a preset joint interaction network to jointly interact with the standard medical inquiry information and the intention recognition result, and outputting the recognized medical inquiry intention, includes:

[0029] Using the text recognition model in the joint interactive network to perform text encoding on the standard medical question information and the intention recognition result, obtaining a horizontal symptom encoding vector corresponding to the first round of supplementary questions, a vertical symptom encoding vector corresponding to the second round of supplementary questions, and an intention encoding vector corresponding to the intention recognition result;

[0030] Using the first interaction layer in the joint interaction network to interact with the horizontal symptom coding vector and the vertical symptom coding vector to obtain a first interaction vector;

[0031] Using a second interaction layer in the joint interaction network to interact with the first interaction vector and the intention encoding vector to obtain a second interaction vector;

[0032] The fully connected layer in the joint interaction network is used to output a classification result of the second interaction vector, and the classification result is used as the recognized medical consultation intention.

[0033] In order to solve the above problems, the present invention further provides a device for identifying medical consultation intention, the device comprising:

[0034] A symptom recognition module is used to obtain initial consultation information of the consultation user, perform intention recognition on the consultation user based on the initial consultation information to obtain an intention recognition result, and perform symptom recognition on the initial consultation information based on the intention recognition result to obtain a symptom recognition result;

[0035] An identification expansion module, used to expand the symptom identification result horizontally and vertically to obtain a horizontal expansion result and a vertical expansion result;

[0036] A multi-round consultation module, used for conducting a multi-round dialogue with the consultation user based on the horizontal expansion result and the vertical expansion result to obtain standard consultation information;

[0037] The joint interaction module is used to use a preset joint interaction network to perform joint interaction on the standard medical inquiry information and the intention recognition result, and output the recognized medical inquiry intention.

[0038] In order to solve the above problem, the present invention further provides an electronic device, the electronic device comprising:

[0039] a memory storing at least one computer program; and

[0040] The processor executes the computer program stored in the memory to implement the above-mentioned method for identifying medical consultation intention.

[0041] In order to solve the above problem, the present invention also provides a computer-readable storage medium, in which at least one computer program is stored. The at least one computer program is executed by a processor in an electronic device to implement the above-mentioned method for identifying medical consultation intention.

[0042] The present invention performs intent recognition on the initial medical inquiry information, and then performs symptom recognition based on the intention recognition result to obtain a symptom recognition result, and expands the symptom recognition result horizontally and vertically, thereby further enriching the user's medical inquiry information and improving the accuracy of medical inquiry intention recognition. At the same time, multiple rounds of dialogues are conducted with the medical inquiry user based on the horizontal expansion results and the vertical expansion results, which can improve the accuracy of user symptom recognition, and the standard medical inquiry information and the intention recognition result are jointly interacted, so that the intention vector finally used for recognition can contain information between different data, thereby improving the accuracy of medical inquiry intention recognition. Therefore, the medical inquiry intention recognition method, device, electronic device and computer-readable storage medium proposed in the present invention can solve the problem of inaccurate user intention recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of a flow chart of a method for identifying medical inquiry intentions provided in one embodiment of the present invention;

[0044] Figure 2 for Figure 1 A detailed implementation flow chart of one of the steps in the process;

[0045] Figure 3 for Figure 1 A detailed implementation flow diagram of another step in FIG.

[0046] Figure 4 for Figure 1A detailed implementation flow diagram of another step in FIG.

[0047] Figure 5 for Figure 1 A detailed implementation flow diagram of another step in FIG.

[0048] Figure 6 for Figure 1 Detailed implementation flow diagram of another step in

[0049] Figure 7 A functional module diagram of a device for identifying medical inquiry intentions provided by an embodiment of the present invention;

[0050] Figure 8 A schematic diagram of the structure of an electronic device for implementing the method for identifying medical consultation intention provided by an embodiment of the present invention.

[0051] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0052] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0053] The embodiment of the present invention provides a method for identifying the intention of medical consultation. The execution subject of the method for identifying the intention of medical consultation includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided by the embodiment of the present invention. In other words, the method for identifying the intention of medical consultation can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0054] Reference Figure 1 FIG. 1 is a flow chart of a method for identifying a medical consultation intention provided by an embodiment of the present invention. In this embodiment, the method for identifying a medical consultation intention includes the following steps S1-S5:

[0055] S1. Obtaining initial consultation information of the consultation user, and performing intention recognition on the consultation user based on the initial consultation information to obtain an intention recognition result.

[0056] In an embodiment of the present invention, the consultation user may be a user in an online TCM consultation scenario. The initial consultation information is the user's initial input text, for example, "I have a headache recently", "Hello, I want to see a doctor", etc. Since there are many online consultation users visiting, and some users have no intention of consulting, it is necessary to perform intent recognition to streamline the data and ensure the accuracy of the consultation user identification. The intent recognition is to classify the sentence (initial consultation information) or what we often call query into the corresponding intent category by classification. In the online TCM consultation scenario, it is to determine whether the user has the intention to consult.

[0057] For details, refer to Figure 2 As shown, the intention recognition of the user who is asking the question based on the initial question information to obtain the intention recognition result includes the following steps S10-S11:

[0058] S10, using a preset intention classification model to perform text encoding on the initial medical inquiry information to obtain a medical inquiry encoding vector;

[0059] S11. Use the intention classification model to perform intent classification on the medical inquiry coding vector, and use the intention classification result as the intention recognition result.

[0060] In an embodiment of the present invention, the preset intent classification model may be a convolutional neural network, for example, the intent classification model may be CNN (Convolutional Neural Networks) + softmax, LSTM (Long Short Term) + softmax, etc., wherein CNN or LSTM is used to perform text encoding on the initial consultation information to obtain a vector representation as a query classification feature, and softmax classifies the features to identify the intent of the user's search query. For example, the intent recognition results include: consultation, non-consultation, and unrecognized results.

[0061] S2. Perform symptom recognition on the initial medical inquiry information based on the intention recognition result to obtain a symptom recognition result.

[0062] In the embodiment of the present invention, for the user whose intention recognition result is "consultation", the main symptom or disease information contained in the text of the initial consultation information is extracted as the symptom recognition result. For example, "headache" is extracted from "I have a headache" as the symptom recognition result.

[0063] Specifically, refer to Figure 3 As shown, the symptom recognition is performed on the initial medical inquiry information based on the intention recognition result to obtain the symptom recognition result, including the following steps S20-S22:

[0064] S20, using a preset word segmentation algorithm to perform word segmentation processing on the initial medical inquiry information corresponding to the medical inquiry user whose intention recognition result is medical inquiry, to obtain a medical inquiry word segmentation text;

[0065] S21, using a preset medical inquiry vocabulary to preliminarily screen the medical inquiry segmentation text to obtain a segmentation screening text;

[0066] S22, performing importance screening on the segmented texts in the segmented screening text, and using the segmented texts after importance screening as the symptom recognition result.

[0067] In an optional embodiment of the present invention, the preset word segmentation algorithm can be any existing word segmentation method, such as NLPIR of the Institute of Computing Technology of the Chinese Academy of Sciences, LTP of Harbin Institute of Technology, THULAC of Tsinghua University, Jieba word segmentation, Stanford word segmenter, etc.

[0068] In another optional embodiment of the present invention, for the consultation user whose intention recognition result is non-consultation and unidentified consultation user, return to S1.

[0069] Optionally, the preset consultation vocabulary list can be a common disease table compiled by medical experts. In addition, the importance of each word in the word segmentation screening text is calculated using the TF-IDF algorithm, and the word segmentation texts are sorted according to the importance, and a preset proportion of word segmentation texts are selected as the symptom recognition result.

[0070] In the embodiment of the present invention, the importance screening of the segmented text in the segmented screening text, and using the segmented text after importance screening as the symptom recognition result, includes:

[0071] Finding the frequency of occurrence of the segmented text in the initial medical consultation information;

[0072] Calculate the inverse document frequency of the segmented text in the initial medical inquiry information;

[0073] The importance of all the segmented texts is calculated based on the occurrence frequency and the inverse document frequency, and a preset number of segmented texts are selected as the symptom recognition result based on the importance.

[0074] In an optional embodiment of the present invention, the importance of all segmented texts is calculated based on the occurrence frequency and the inverse document frequency, including:

[0075] The importance Z of the segmented text is calculated using the following formula:

[0076]

[0077] Among them, TF i is the frequency of occurrence of word segmentation text i, N is the number of sentences in the initial consultation information, and wi is the number of sentences containing word segmentation text i.

[0078] S3. Expand the symptom recognition result horizontally and vertically to obtain a horizontal expansion result and a vertical expansion result.

[0079] In the embodiment of the present invention, the horizontal expansion and vertical expansion refer to extending the inquiry of information such as the location, degree, and time of the symptom based on the symptom recognition result.

[0080] For details, refer to Figure 4 As shown, the symptom recognition result is expanded horizontally and vertically to obtain the horizontal expansion result and the vertical expansion result, including the following steps S30-S31:

[0081] S30, obtaining a preset horizontal question template and a vertical question template, and filling the symptom recognition result into the horizontal question template and the vertical question template;

[0082] S31. Using the filled horizontal question template as a horizontal expansion result, and using the filled vertical question template as a vertical expansion result.

[0083] In an optional embodiment of the present invention, the horizontal question template is constructed based on the location of the symptom, etc., and the vertical question template is constructed based on the degree and occurrence time of the symptom. For example, if the symptom recognition result is "headache", the horizontal expansion result is "Where is the location of the headache?", and the vertical expansion result is: "How long does the headache last?", "Are there other symptoms besides headache?", etc.

[0084] In the embodiment of the present invention, more medical consultation information can be obtained through horizontal expansion and vertical expansion, thereby improving the accuracy of identifying the user's medical consultation.

[0085] S4. Conduct multiple rounds of dialogue with the consultation user based on the horizontal expansion results and the vertical expansion results to obtain standard consultation information.

[0086] In the embodiment of the present invention, since it is difficult to obtain accurate consultation information through a single round of dialogue, more abundant symptom information can be obtained by conducting multiple rounds of dialogue with the consultation user through horizontal expansion results and vertical expansion results.

[0087] For details, refer to Figure 5 As shown, the method of conducting multiple rounds of dialogue with the consulting user based on the horizontal expansion result and the vertical expansion result to obtain standard consultation information includes the following steps S40-S41:

[0088] S40, using the horizontal expansion result as the first round of supplementary questioning, and using the vertical expansion result as the second round of supplementary questioning;

[0089] S41, receiving the consultation information provided by the user based on the first round of supplementary questions and the second round of supplementary questions, and returning to the step of obtaining the initial consultation information of the user, until the number of conversations meets the preset threshold, and obtaining the symptom information corresponding to all supplementary questions as the standard consultation information.

[0090] In the embodiment of the present invention, the standard medical inquiry information includes the symptoms answered by the user after horizontal expansion and vertical expansion, which enriches the medical inquiry data and improves the accuracy of the medical inquiry.

[0091] S5. Use a preset joint interaction network to jointly interact with the standard medical inquiry information and the intention recognition result, and output the recognized medical inquiry intention.

[0092] In the embodiment of the present invention, since the standard medical consultation information includes the symptom information answered by the user based on horizontal expansion and vertical expansion, it is closely related to the intention recognition result. For example, the information between the horizontal and vertical symptoms is closely related, and both are performed through intention recognition and then symptom recognition, the intention information is also related to the symptom information. Therefore, richer feature information can be extracted through joint interaction.

[0093] In another optional embodiment of the present invention, before using a preset joint interaction network to jointly interact with the standard medical inquiry information and the intention recognition result, the method also includes: obtaining a pre-built text recognition model, and connecting a first interaction layer, a second interaction layer, and a fully connected layer after the text recognition model to obtain the preset joint interaction network.

[0094] In an optional embodiment of the present invention, the joint interactive network is a composite network, including a text recognition model, a first interactive layer, a second interactive layer and a fully connected layer, etc., wherein the text recognition model can be a Bert_base model, which is used to perform text encoding on standard medical consultation information and intent recognition results. Compared with the use of RNN and CNN models for text encoding, the advantage of Bert lies in the Mask LM in its pre-training task, which allows the model to make full use of the context information of the sentence for encoding, and in the selection of the encoder layer, a deeper transformer encoder with better parallelism is used, so that each word can ignore the direction and distance, so that each word in each sentence has the opportunity to be encoded, and a more semantically rich text representation is obtained; the first interactive layer and the second interactive layer can be two Attention layers, which are used to fuse text information; the fully connected layer is used to classify user intent and output the recognized chief complaint medical consultation intent.

[0095] For details, refer to Figure 6As shown, the method of using a preset joint interaction network to jointly interact with the standard medical inquiry information and the intention recognition result and outputting the recognized medical inquiry intention includes the following steps S50-S53:

[0096] S50, using the text recognition model in the joint interactive network to perform text encoding on the standard medical question information and the intention recognition result, to obtain the horizontal symptom encoding vector corresponding to the first round of supplementary questions, the vertical symptom encoding vector corresponding to the second round of supplementary questions, and the intention encoding vector corresponding to the intention recognition result;

[0097] S51, using the first interaction layer in the joint interaction network to interact with the horizontal symptom coding vector and the vertical symptom coding vector to obtain a first interaction vector;

[0098] S52, using the second interaction layer in the joint interaction network to interact with the first interaction vector and the intention encoding vector to obtain a second interaction vector;

[0099] S53: Use the fully connected layer in the joint interaction network to output the classification result of the second interaction vector, and use the classification result as the identified medical consultation intention.

[0100] In the embodiment of the present invention, by performing two joint interactions on the medical inquiry information through two interaction layers, more chief complaint information can be collected from the medical inquiry information, thereby improving the accuracy of identifying the user's chief complaint intention. For example, the medical inquiry intention of user A is "I have a headache on my forehead for three days, I want to see a doctor", etc.

[0101] The present invention performs intent recognition on the initial medical inquiry information, and then performs symptom recognition based on the intention recognition result to obtain a symptom recognition result, and expands the symptom recognition result horizontally and vertically, thereby further enriching the user's medical inquiry information and improving the accuracy of medical inquiry intention recognition. At the same time, multiple rounds of dialogues are conducted with the medical inquiry user based on the horizontal expansion results and the vertical expansion results, which can improve the accuracy of user symptom recognition, and the standard medical inquiry information and the intention recognition results are jointly interacted, so that the intention vector finally used for recognition can contain information between different data, thereby improving the accuracy of medical inquiry intention recognition. Therefore, the medical inquiry intention recognition method proposed in the present invention can solve the problem of inaccurate user intention recognition.

[0102] like Figure 7 , which is a functional module diagram of a device for identifying medical inquiry intentions provided in one embodiment of the present invention.

[0103] The medical consultation intention recognition device 100 of the present invention can be installed in an electronic device. According to the functions to be implemented, the medical consultation intention recognition device 100 can include a symptom recognition module 101, a recognition expansion module 102, a multi-round medical consultation module 103 and a joint interaction module 104. The module of the present invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0104] In this embodiment, the functions of each module / unit are as follows:

[0105] The symptom identification module 101 is used to obtain initial consultation information of the consultation user, perform intention identification on the consultation user based on the initial consultation information to obtain an intention identification result, and perform symptom identification on the initial consultation information based on the intention identification result to obtain a symptom identification result;

[0106] The identification expansion module 102 is used to expand the symptom identification result horizontally and vertically to obtain a horizontal expansion result and a vertical expansion result;

[0107] The multi-round consultation module 103 is used to conduct multi-round dialogues with the consultation user based on the horizontal expansion results and the vertical expansion results to obtain standard consultation information;

[0108] The joint interaction module 104 is used to use a preset joint interaction network to perform joint interaction on the standard medical inquiry information and the intention recognition result, and output the recognized medical inquiry intention.

[0109] In detail, the specific implementation of each module of the medical inquiry intention recognition device 100 is as follows:

[0110] Step 1: Obtain the initial consultation information of the consultation user, and perform intention recognition on the consultation user based on the initial consultation information to obtain an intention recognition result.

[0111] In an embodiment of the present invention, the consultation user may be a user in an online TCM consultation scenario. The initial consultation information is the user's initial input text, for example, "I have a headache recently", "Hello, I want to see a doctor", etc. Since there are many online consultation users visiting, and some users have no intention of consulting, it is necessary to perform intent recognition to streamline the data and ensure the accuracy of the consultation user identification. The intent recognition is to classify the sentence (initial consultation information) or what we often call query into the corresponding intent category by classification. In the online TCM consultation scenario, it is to determine whether the user has the intention to consult.

[0112] In detail, the performing of intention recognition on the user asking the question based on the initial consultation information to obtain the intention recognition result includes:

[0113] Using a preset intention classification model to perform text encoding on the initial medical consultation information to obtain a medical consultation encoding vector;

[0114] The intention classification model is used to perform intention classification on the medical inquiry coding vector, and the result of the intention classification is used as the intention recognition result.

[0115] In an embodiment of the present invention, the preset intent classification model may be a convolutional neural network, for example, the intent classification model may be CNN (Convolutional Neural Networks) + softmax, LSTM (Long Short Term) + softmax, etc., wherein CNN or LSTM is used to perform text encoding on the initial consultation information to obtain a vector representation as a query classification feature, and softmax classifies the features to identify the intent of the user's search query. For example, the intent recognition results include: consultation, non-consultation, and unrecognized results.

[0116] Step 2: Perform symptom recognition on the initial medical inquiry information based on the intention recognition result to obtain a symptom recognition result.

[0117] In the embodiment of the present invention, for the user whose intention recognition result is "consultation", the main symptom or disease information contained in the text of the initial consultation information is extracted as the symptom recognition result. For example, "headache" is extracted from "I have a headache" as the symptom recognition result.

[0118] Specifically, performing symptom recognition on the initial medical inquiry information based on the intention recognition result to obtain a symptom recognition result includes:

[0119] Using a preset word segmentation algorithm, the initial medical inquiry information corresponding to the medical inquiry user whose intention recognition result is medical inquiry is processed by word segmentation to obtain the medical inquiry word segmentation text;

[0120] Using a preset medical inquiry vocabulary list to preliminarily screen the medical inquiry word segmentation text to obtain a word segmentation screening text;

[0121] The segmented texts in the segmented screening text are screened for importance, and the segmented texts after importance screening are used as the symptom recognition results.

[0122] In an optional embodiment of the present invention, the preset word segmentation algorithm can be any existing word segmentation method, such as NLPIR of the Institute of Computing Technology of the Chinese Academy of Sciences, LTP of Harbin Institute of Technology, THULAC of Tsinghua University, Jieba word segmentation, Stanford word segmenter, etc.

[0123] In another optional embodiment of the present invention, for the consultation user whose intention recognition result is non-consultation and unidentified consultation user, return to step one.

[0124] Optionally, the preset consultation vocabulary list can be a common disease table compiled by medical experts. In addition, the importance of each word in the word segmentation screening text is calculated using the TF-IDF algorithm, and the word segmentation texts are sorted according to the importance, and a preset proportion of word segmentation texts are selected as the symptom recognition result.

[0125] In the embodiment of the present invention, the importance screening of the segmented text in the segmented screening text, and using the segmented text after importance screening as the symptom recognition result, includes:

[0126] Finding the frequency of occurrence of the segmented text in the initial medical consultation information;

[0127] Calculate the inverse document frequency of the segmented text in the initial medical inquiry information;

[0128] The importance of all the segmented texts is calculated based on the occurrence frequency and the inverse document frequency, and a preset number of segmented texts are selected as the symptom recognition result based on the importance.

[0129] In an optional embodiment of the present invention, the importance of all segmented texts is calculated based on the occurrence frequency and the inverse document frequency, including:

[0130] The importance Z of the segmented text is calculated using the following formula:

[0131]

[0132] Among them, TF i is the frequency of occurrence of word segmentation text i, N is the number of sentences in the initial consultation information, and w i is the number of sentences containing the word segmentation text i.

[0133] Step three: expand the symptom recognition results horizontally and vertically to obtain horizontal expansion results and vertical expansion results.

[0134] In the embodiment of the present invention, the horizontal expansion and vertical expansion refer to extending the inquiry of information such as the location, degree, and time of the symptom based on the symptom recognition result.

[0135] In detail, the symptom recognition result is expanded horizontally and vertically to obtain a horizontal expansion result and a vertical expansion result, including:

[0136] Obtaining a preset horizontal question template and a vertical question template, and filling the symptom recognition result into the horizontal question template and the vertical question template;

[0137] The filled horizontal question template is used as the horizontal expansion result, and the filled vertical question template is used as the vertical expansion result.

[0138] In an optional embodiment of the present invention, the horizontal question template is constructed based on the location of the symptom, etc., and the vertical question template is constructed based on the degree and occurrence time of the symptom. For example, if the symptom recognition result is "headache", the horizontal expansion result is "Where is the location of the headache?", and the vertical expansion result is: "How long does the headache last?", "Are there other symptoms besides headache?", etc.

[0139] In the embodiment of the present invention, more medical consultation information can be obtained through horizontal expansion and vertical expansion, thereby improving the accuracy of identifying the user's medical consultation.

[0140] Step 4: Conduct multiple rounds of dialogue with the consultation user based on the horizontal expansion results and the vertical expansion results to obtain standard consultation information.

[0141] In the embodiment of the present invention, since it is difficult to obtain accurate consultation information through a single round of dialogue, more abundant symptom information can be obtained by conducting multiple rounds of dialogue with the consultation user through horizontal expansion results and vertical expansion results.

[0142] Specifically, the user who is inquiring is given multiple rounds of dialogue based on the horizontal expansion result and the vertical expansion result to obtain standard consultation information, including:

[0143] Using the horizontal expansion results as the first round of supplementary questioning, and using the vertical expansion results as the second round of supplementary questioning;

[0144] Receive the consultation information provided by the user based on the first round of supplementary questions and the second round of supplementary questions, and return to the step of obtaining the initial consultation information of the user until the number of conversations meets a preset threshold, and obtain the symptom information corresponding to all supplementary questions as the standard consultation information.

[0145] In the embodiment of the present invention, the standard medical inquiry information includes the symptoms answered by the user after horizontal expansion and vertical expansion, which enriches the medical inquiry data and improves the accuracy of the medical inquiry.

[0146] Step 5: Use a preset joint interaction network to jointly interact with the standard medical inquiry information and the intention recognition result, and output the recognized medical inquiry intention.

[0147] In the embodiment of the present invention, since the standard medical consultation information includes the symptom information answered by the user based on horizontal expansion and vertical expansion, it is closely related to the intention recognition result. For example, the information between the horizontal and vertical symptoms is closely related, and both are performed through intention recognition and then symptom recognition, the intention information is also related to the symptom information. Therefore, richer feature information can be extracted through joint interaction.

[0148] In another optional embodiment of the present invention, before using a preset joint interaction network to jointly interact with the standard medical inquiry information and the intention recognition result, the method also includes: obtaining a pre-built text recognition model, and connecting a first interaction layer, a second interaction layer, and a fully connected layer after the text recognition model to obtain the preset joint interaction network.

[0149] In an optional embodiment of the present invention, the joint interactive network is a composite network, including a text recognition model, a first interactive layer, a second interactive layer and a fully connected layer, etc., wherein the text recognition model can be a Bert_base model, which is used to perform text encoding on standard medical consultation information and intent recognition results. Compared with the use of RNN and CNN models for text encoding, the advantage of Bert lies in the Mask LM in its pre-training task, which allows the model to make full use of the context information of the sentence for encoding, and in the selection of the encoder layer, a deeper transformer encoder with better parallelism is used, so that each word can ignore the direction and distance, so that each word in each sentence has the opportunity to be encoded, and a more semantically rich text representation is obtained; the first interactive layer and the second interactive layer can be two Attention layers, which are used to fuse text information; the fully connected layer is used to classify user intent and output the recognized chief complaint medical consultation intent.

[0150] In detail, the method of using a preset joint interaction network to jointly interact with the standard medical inquiry information and the intention recognition result, and outputting the recognized medical inquiry intention, includes:

[0151] Using the text recognition model in the joint interactive network to perform text encoding on the standard medical question information and the intention recognition result, obtaining a horizontal symptom encoding vector corresponding to the first round of supplementary questions, a vertical symptom encoding vector corresponding to the second round of supplementary questions, and an intention encoding vector corresponding to the intention recognition result;

[0152] Using the first interaction layer in the joint interaction network to interact with the horizontal symptom coding vector and the vertical symptom coding vector to obtain a first interaction vector;

[0153] Using a second interaction layer in the joint interaction network to interact with the first interaction vector and the intention encoding vector to obtain a second interaction vector;

[0154] The fully connected layer in the joint interaction network is used to output a classification result of the second interaction vector, and the classification result is used as the recognized medical consultation intention.

[0155] In the embodiment of the present invention, by performing two joint interactions on the medical inquiry information through two interaction layers, more chief complaint information can be collected from the medical inquiry information, thereby improving the accuracy of identifying the user's chief complaint intention. For example, the medical inquiry intention of user A is "I have a headache on my forehead for three days, I want to see a doctor", etc.

[0156] The present invention performs intent recognition on the initial medical inquiry information, and then performs symptom recognition based on the intention recognition result to obtain a symptom recognition result, and expands the symptom recognition result horizontally and vertically, thereby further enriching the user's medical inquiry information and improving the accuracy of medical inquiry intention recognition. At the same time, multiple rounds of dialogues are conducted with the medical inquiry user based on the horizontal expansion results and the vertical expansion results, which can improve the accuracy of user symptom recognition, and the standard medical inquiry information and the intention recognition results are jointly interacted, so that the intention vector finally used for recognition can contain information between different data, thereby improving the accuracy of medical inquiry intention recognition. Therefore, the medical inquiry intention recognition method proposed in the present invention can solve the problem of inaccurate user intention recognition.

[0157] like Figure 8 FIG. 1 is a schematic diagram of the structure of an electronic device for implementing a method for identifying medical consultation intentions provided by an embodiment of the present invention.

[0158] The electronic device may include a processor 10, a memory 11, a communication interface 12 and a bus 13, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as a medical consultation intention recognition program.

[0159] Wherein, the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (for example: SD or DX memory, etc.), magnetic memory, disk, optical disk, etc. The memory 11 may be an internal storage unit of an electronic device in some embodiments, such as a mobile hard disk of the electronic device. The memory 11 may also be an external storage device of an electronic device in other embodiments, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device. Further, the memory 11 may also include both an internal storage unit of the electronic device and an external storage device. The memory 11 may not only be used to store application software and various types of data installed in the electronic device, such as the code of the consultation intention recognition program, but also be used to temporarily store data that has been output or is to be output.

[0160] The processor 10 may be composed of an integrated circuit in some embodiments, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and a combination of various control chips, etc. The processor 10 is the control core (Control Unit) of the electronic device, and uses various interfaces and lines to connect the various components of the entire electronic device, and executes or executes the programs or modules (such as the inquiry intention recognition program, etc.) stored in the memory 11, and calls the data stored in the memory 11 to execute various functions of the electronic device and process data.

[0161] The communication interface 12 is used for communication between the above-mentioned electronic device and other devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device and other electronic devices. The user interface may be a display (Display), an input unit (such as a keyboard (Keyboard)), and optionally, the user interface may also be a standard wired interface, a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, etc. Among them, the display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device and to display a visual user interface.

[0162] The bus 13 may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus 13 may be divided into an address bus, a data bus, a control bus, etc. The bus 13 is configured to realize connection and communication between the memory 11 and at least one processor 10, etc.

[0163] Figure 8 Only an electronic device with components is shown, and those skilled in the art will understand that Figure 8 The structure shown does not constitute a limitation on the electronic device, and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0164] For example, although not shown, the electronic device may also include a power source (such as a battery) for supplying power to each component. Preferably, the power source may be logically connected to the at least one processor 10 through a power management device, so that the power management device can realize functions such as charging management, discharging management, and power consumption management. The power source may also include one or more DC or AC power sources, recharging devices, power failure detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device may also include a variety of sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.

[0165] Furthermore, the electronic device may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device and other electronic devices.

[0166] Optionally, the electronic device may further include a user interface, which may be a display, an input unit (such as a keyboard), or a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-Emitting Diode) touch device. The display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device and to display a visual user interface.

[0167] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0168] The medical consultation intention recognition program stored in the memory 11 in the electronic device is a combination of multiple instructions. When running in the processor 10, it can achieve:

[0169] Acquire initial consultation information of the consultation user, and perform intention recognition on the consultation user based on the initial consultation information to obtain an intention recognition result;

[0170] Performing symptom recognition on the initial medical inquiry information based on the intention recognition result to obtain a symptom recognition result;

[0171] Expanding the symptom recognition result horizontally and vertically to obtain a horizontal expansion result and a vertical expansion result;

[0172] Conduct multiple rounds of dialogue with the consultation user based on the horizontal expansion results and the vertical expansion results to obtain standard consultation information;

[0173] The preset joint interaction network is used to perform joint interaction on the standard medical inquiry information and the intention recognition result, and the recognized medical inquiry intention is output.

[0174] Specifically, the specific implementation method of the processor 10 for the above instructions can refer to the description of the relevant steps in the corresponding embodiment of the accompanying drawings, which will not be repeated here.

[0175] Furthermore, if the module / unit integrated in the electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, and a read-only memory (ROM).

[0176] The present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor of an electronic device, the computer program can implement:

[0177] Acquire initial consultation information of the consultation user, and perform intention recognition on the consultation user based on the initial consultation information to obtain an intention recognition result;

[0178] Performing symptom recognition on the initial medical inquiry information based on the intention recognition result to obtain a symptom recognition result;

[0179] Expanding the symptom recognition result horizontally and vertically to obtain a horizontal expansion result and a vertical expansion result;

[0180] Conduct multiple rounds of dialogue with the consultation user based on the horizontal expansion results and the vertical expansion results to obtain standard consultation information;

[0181] The preset joint interaction network is used to perform joint interaction on the standard medical inquiry information and the intention recognition result, and the recognized medical inquiry intention is output.

[0182] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0183] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0184] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0185] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0186] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.

[0187] The embodiments of the present invention can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0188] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0189] The blockchain referred to in this invention is a new application mode of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0190] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the system claim can also be implemented by one unit or device through software or hardware. The second and other words are used to indicate names, but not to indicate any particular order.

[0191] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for identifying medical inquiry intention, characterized in that: The method comprises: Acquire initial consultation information of the consultation user, and perform intention recognition on the consultation user based on the initial consultation information to obtain an intention recognition result; Performing symptom recognition on the initial medical inquiry information based on the intention recognition result to obtain a symptom recognition result; Expanding the symptom recognition result horizontally and vertically to obtain a horizontal expansion result and a vertical expansion result; Conduct multiple rounds of dialogue with the consultation user based on the horizontal expansion results and the vertical expansion results to obtain standard consultation information; Using a preset joint interaction network to jointly interact with the standard medical inquiry information and the intention recognition result, and outputting the recognized medical inquiry intention; The horizontal expansion and vertical expansion of the symptom recognition result to obtain the horizontal expansion result and the vertical expansion result include: obtaining a preset horizontal question template and a preset vertical question template, filling the symptom recognition result into the horizontal question template and the preset vertical question template; using the filled horizontal question template as the horizontal expansion result, and using the filled vertical question template as the vertical expansion result; The method of conducting multiple rounds of dialogues with the user inquiring based on the horizontal expansion results and the vertical expansion results to obtain standard medical inquiry information includes: using the horizontal expansion results as the first round of supplementary questions, and using the vertical expansion results as the second round of supplementary questions; receiving the medical inquiry information fed back by the user inquiring based on the first round of supplementary questions and the second round of supplementary questions, and returning to the step of obtaining the initial medical inquiry information of the user inquiring, until the number of dialogues meets a preset number threshold, obtaining the symptom information corresponding to all supplementary questions as the standard medical inquiry information; The method of using a preset joint interactive network to jointly interact with the standard medical inquiry information and the intention recognition result and outputting the recognized medical inquiry intention includes: using a text recognition model in the joint interactive network to text encode the standard medical inquiry information and the intention recognition result to obtain a horizontal symptom coding vector corresponding to the first round of supplementary questions, a vertical symptom coding vector corresponding to the second round of supplementary questions, and an intention coding vector corresponding to the intention recognition result; using the first interactive layer in the joint interactive network to interact with the horizontal symptom coding vector and the vertical symptom coding vector to obtain a first interactive vector; using the second interactive layer in the joint interactive network to interact with the first interactive vector and the intention coding vector to obtain a second interactive vector; using the fully connected layer in the joint interactive network to output a classification result of the second interaction vector, and using the classification result as the recognized medical inquiry intention.

2. The method for identifying medical inquiry intention according to claim 1, characterized in that: The performing intention recognition on the user asking the question based on the initial consultation information to obtain the intention recognition result includes: Using a preset intention classification model to perform text encoding on the initial medical consultation information to obtain a medical consultation encoding vector; The intention classification model is used to perform intention classification on the medical inquiry coding vector, and the result of the intention classification is used as the intention recognition result.

3. The method for identifying medical inquiry intention as claimed in claim 1, characterized in that: The performing symptom recognition on the initial medical inquiry information based on the intention recognition result to obtain a symptom recognition result includes: Using a preset word segmentation algorithm, the initial medical inquiry information corresponding to the medical inquiry user whose intention recognition result is medical inquiry is processed by word segmentation to obtain the medical inquiry word segmentation text; Using a preset medical inquiry vocabulary list to preliminarily screen the medical inquiry word segmentation text to obtain a word segmentation screening text; The segmented texts in the segmented screening text are screened for importance, and the segmented texts after importance screening are used as the symptom recognition results.

4. The method for identifying medical inquiry intention according to claim 3, characterized in that: The step of screening the importance of the segmented text in the segmented screening text and using the segmented text after importance screening as the symptom recognition result includes: Finding the frequency of occurrence of the segmented text in the initial medical consultation information; Calculate the inverse document frequency of the segmented text in the initial medical inquiry information; The importance of all the segmented texts is calculated based on the occurrence frequency and the inverse document frequency, and a preset number of segmented texts are selected as the symptom recognition result based on the importance.

5. A device for identifying medical consultation intention, used to implement the method for identifying medical consultation intention as claimed in any one of claims 1 to 4, characterized in that: The device comprises: A symptom recognition module is used to obtain initial consultation information of the consultation user, perform intention recognition on the consultation user based on the initial consultation information to obtain an intention recognition result, and perform symptom recognition on the initial consultation information based on the intention recognition result to obtain a symptom recognition result; An identification expansion module, used to expand the symptom identification result horizontally and vertically to obtain a horizontal expansion result and a vertical expansion result; A multi-round consultation module, used for conducting a multi-round dialogue with the consultation user based on the horizontal expansion result and the vertical expansion result to obtain standard consultation information; The joint interaction module is used to use a preset joint interaction network to perform joint interaction on the standard medical inquiry information and the intention recognition result, and output the recognized medical inquiry intention.

6. An electronic device, characterized in that: The electronic device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for identifying medical consultation intention as described in any one of claims 1 to 4.

7. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for identifying medical consultation intention according to any one of claims 1 to 4 is implemented.

Images