CN111326251A - Method and device for outputting inquiry questions and electronic equipment - Google Patents
Method and device for outputting inquiry questions and electronic equipment Download PDFInfo
- Publication number
- CN111326251A CN111326251A CN202010091503.XA CN202010091503A CN111326251A CN 111326251 A CN111326251 A CN 111326251A CN 202010091503 A CN202010091503 A CN 202010091503A CN 111326251 A CN111326251 A CN 111326251A
- Authority
- CN
- China
- Prior art keywords
- symptom
- inquiry
- sample
- prior probability
- disease
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Public Health (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Physics & Mathematics (AREA)
- Molecular Biology (AREA)
- Medical Informatics (AREA)
- General Engineering & Computer Science (AREA)
- Epidemiology (AREA)
- Pathology (AREA)
- Primary Health Care (AREA)
- Databases & Information Systems (AREA)
- Medical Treatment And Welfare Office Work (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The application discloses an inquiry question output method, an inquiry question output device and electronic equipment, and relates to the technical field of machine learning. The specific implementation scheme is as follows: predicting disease prior probability according to a first symptom of the first object to obtain the disease prior probability of the first object aiming at the first symptom; and inputting the symptom state vector and the disease prior probability into the trained confrontation generation network, and outputting an inquiry question of a second symptom, wherein a state value corresponding to a first symptom in the symptom state vector indicates that the first symptom exists in the first object. The disease prior probability is considered, and the symptom state vector is also considered, namely the inquiry question of the second symptom is predicted by combining the disease prior probability and the symptom state vector, so that the accuracy of the output inquiry question of the second symptom can be improved.
Description
Technical Field
The present application relates to the field of machine learning technology in computer technology, and in particular, to an inquiry question output method, an apparatus and an electronic device.
Background
With the continuous development of computer technology, more and more intelligent products are available in the medical field, for example, the generation of an intelligent inquiry system brings great convenience to the inquiry of a user, the intelligent inquiry system continuously provides inquiry questions, and the user can complete the intelligent inquiry process by answering the inquiry questions provided by the intelligent inquiry system.
However, currently, the next inquiry question to be presented is generally predicted by the presented inquiry question and the response result, the consideration factors are few, and the accuracy of the easily outputted inquiry question is low.
Disclosure of Invention
The application provides an inquiry question output method, an inquiry question output device and electronic equipment, and aims to solve the problem that the output inquiry question is low in accuracy.
In a first aspect, an embodiment of the present application provides an inquiry question output method, including:
predicting disease prior probability according to a first symptom of a first object to obtain the disease prior probability of the first object aiming at the first symptom;
inputting a symptom state vector and the disease prior probability into a trained countermeasure generation network, and outputting an inquiry question of a second symptom, wherein a state value corresponding to the first symptom in the symptom state vector indicates that the first symptom exists in the first object.
In the inquiry question output method provided by the embodiment of the application, firstly, disease prior probability prediction is carried out according to a first symptom of a first object to obtain the disease prior probability of the first object for the first symptom, then, a symptom state vector and the disease prior probability are input into a trained confrontation generation network, and an inquiry question of a second symptom is predicted and output through the trained confrontation generation network. The accuracy of the output inquiry question of the second symptom can be improved because the disease prior probability is considered in the process of outputting the inquiry question of the second symptom, and the symptom state vector is also considered, namely the inquiry question of the second symptom is predicted by combining the disease prior probability and the symptom state vector.
Optionally, after outputting the inquiry question of the second symptom, the method further includes:
receiving response information input by the first subject for the second symptom;
updating the symptom state vector according to the answer information, adding the second symptom into the first symptom under the condition that the answer information indicates that the first object has the second symptom, and returning to the step of predicting the prior disease probability according to the first symptom of the first object to obtain the prior disease probability of the first object aiming at the first symptom until the prior disease probability meets an inquiry stopping condition.
Therefore, by continuously predicting the disease prior probability and continuously predicting the inquiry question of the second symptom according to the disease prior probability and the first symptom until the disease prior probability meets the inquiry stopping condition, the continuously updated first symptom and the predicted disease prior probability are considered in the inquiry question output process, and the accuracy of the inquiry question is improved.
Optionally, before the predicting the disease prior probability according to the first symptom of the first object and obtaining the disease prior probability of the first object for the first symptom, the method further includes:
receiving first content input by the first object;
performing word segmentation processing on the first content to obtain medical related words;
and obtaining the first symptom according to the medical related word segmentation.
Namely, the first symptom is related to the medical word segmentation, so that the accuracy of the first symptom can be improved, and the accuracy of the output of the inquiry question can be improved.
Optionally, the condition for stopping interrogation includes that the disease prior probability includes a target probability greater than a preset probability threshold.
Once the disease prior probability comprises the target probability which is greater than the preset probability threshold value, and the disease prior probability meets the inquiry stopping condition, the output of the inquiry questions is stopped, and the output of the inquiry questions is finished, so that the accuracy of the second needle-shaped inquiry questions in the inquiry question output process can be improved.
Optionally, until the disease prior probability satisfies an inquiry stop condition, the method further includes: outputting an interrogation sequence for the first subject, wherein the interrogation sequence comprises interrogation questions of the second symptoms ordered in output time order.
In this manner, the first subject may be facilitated to view the interrogation sequence.
Optionally, the trained confrontation generation network is obtained by:
obtaining M sample inquiry sequences and a disease diagnosis label of each sample inquiry sequence, wherein M is a positive integer;
obtaining a plurality of sample symptom state vectors corresponding to each sample inquiry sequence respectively, and training disease prior probability aiming at sample symptoms existing in each sample symptom state vector;
training the confrontation generation network according to a plurality of sample symptom state vectors respectively corresponding to each sample inquiry sequence and the training disease prior probability aiming at the sample symptom indicated to exist in each sample symptom state vector to obtain the trained confrontation generation network.
For each sample symptom state vector, the sample symptom present in the sample object corresponding to the sample symptom state vector can be determined, which belongs to the corresponding sample interrogation sequence, and thus, the prior probability of the training disease indicating the present sample symptom in each sample symptom state vector can be obtained. And training the confrontation generation network according to a plurality of sample symptom state vectors respectively corresponding to each sample inquiry sequence and the training disease prior probability indicating the existing sample symptoms in each sample symptom state vector to obtain the trained confrontation generation network, so that the training accuracy is improved.
In a second aspect, an embodiment of the present application provides an inquiry question output apparatus, including:
the prior probability prediction module is used for predicting disease prior probability according to a first symptom of a first object to obtain the disease prior probability of the first object aiming at the first symptom;
and the output module is used for inputting the symptom state vector and the disease prior probability into the trained confrontation generation network and outputting the inquiry question of the second symptom, wherein the state value corresponding to the first symptom in the symptom state vector indicates that the first symptom exists in the first object.
Optionally, the apparatus further includes:
an answer receiving module for receiving answer information input by the first object for the question of the second symptom;
and an updating module, configured to update the symptom state vector according to the answer information, add the second symptom to the first symptom when the answer information indicates that the first subject has the second symptom, and return to the step of predicting the disease prior probability according to the first symptom of the first subject to obtain the disease prior probability of the first subject for the first symptom until the disease prior probability satisfies an inquiry stop condition.
Optionally, the apparatus further includes:
the content receiving module is used for receiving first content input by the first object;
the word segmentation module is used for carrying out word segmentation processing on the first content to obtain medical related words;
a first needle-shaped acquisition module for obtaining the first symptom according to the medical-related segmentation.
Optionally, the condition for stopping interrogation includes that the disease prior probability includes a target probability greater than a preset probability threshold.
Optionally, the apparatus further includes:
and the sequence output module is used for outputting an inquiry sequence aiming at the first object until the disease prior probability meets an inquiry stopping condition, wherein the inquiry sequence comprises inquiry questions of a second symptom which are sequenced in an output time sequence.
Optionally, the apparatus further includes:
the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring M sample inquiry sequences and a disease diagnosis label of each sample inquiry sequence, and M is a positive integer;
a second obtaining module, configured to obtain a plurality of sample symptom state vectors corresponding to each sample inquiry sequence, and a priori probability of a training disease indicating existing sample symptoms in each sample symptom state vector;
and the training module is used for training the confrontation generation network according to a plurality of sample symptom state vectors respectively corresponding to each sample inquiry sequence and the training disease prior probability aiming at the sample symptom indicated to exist in each sample symptom state vector to obtain the trained confrontation generation network.
In a third aspect, an embodiment of the present application further provides an electronic device, including:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein the content of the first and second substances,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the methods provided by the embodiments of the present application.
In a fourth aspect, an embodiment of the present application further provides a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method provided by the embodiments of the present application.
Drawings
The drawings are included to provide a better understanding of the present solution and are not intended to limit the present application. Wherein:
FIG. 1 is a schematic flow chart diagram of an inquiry question output method according to an embodiment provided herein;
FIG. 2 is a schematic diagram of an interrogation question output of one embodiment provided herein;
FIG. 3 is one of the block diagrams of an interrogation question output apparatus of an embodiment provided by the present application;
FIG. 4 is a second block diagram of an inquiry question output device according to an embodiment of the present application;
fig. 5 is a block diagram of an electronic device for implementing an inquiry question output method according to an embodiment of the present application.
Detailed Description
The following description of the exemplary embodiments of the present application, taken in conjunction with the accompanying drawings, includes various details of the embodiments of the application for the understanding of the same, which are to be considered exemplary only. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present application. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
As shown in fig. 1, according to an embodiment of the present application, there is provided an inquiry question output method including:
step S101: and predicting the disease prior probability according to the first symptom of the first object to obtain the disease prior probability of the first object aiming at the first symptom.
The first object may be an object capable of content input, for example, the first object may be a user or a patient, etc. who needs an inquiry. In one example, an inquiry input interface is provided for a first object, the first object can enter content at the inquiry input interface, and the first stitch can be a symptom entered by the first object.
The method can be applied to an electronic device, after obtaining the first symptom of the first object, the disease prior probability prediction can be performed according to the first symptom of the first object, so as to obtain the disease prior probability of the first object for the first symptom, and the prior probability for a certain disease can be understood as the probability that the disease exists in the first object, and the larger the value, the more likely the disease exists in the first object.
S102: and inputting the symptom state vector and the disease prior probability into the trained confrontation generation network, and outputting an inquiry question of a second symptom, wherein a state value corresponding to a first symptom in the symptom state vector indicates that the first symptom exists in the first object.
In this embodiment, the symptom state vector and the disease prior probability may be input into a trained confrontation generating network, the next inquiry question of the symptom to be proposed is predicted by the confrontation generating network, and the inquiry question of the next predicted symptom is output, where the second symptom may be understood as the next predicted symptom, and the second symptom obtained in each execution of step 102 is different, so that the inquiry questions of a plurality of second symptoms may be output, and the inquiry sequence includes the inquiry questions of a plurality of second symptoms. In one example, the second symptom of the interview question can be output by a trained evaluator in the trained challenge generation network,
the symptom state vector includes state values of N symptoms, where N is an integer greater than 1, and the N symptoms may be understood as all symptoms counted according to historical diagnostic data, that is, all symptoms in a symptom space. The state value is used to indicate whether the first object has a corresponding symptom, for example, a state value of 1 for symptom a indicates that the first object has symptom a, a state value of-1 for symptom B indicates that the first object does not have symptom B, and a state value of 0 for symptom C indicates that the first object has symptom C unknown. The symptom state vector is a vector composed of different state values. In this embodiment, the first needle is a symptom of the first object, and the state value corresponding to the first symptom in the symptom state vector indicates that the first object has the first symptom, for example, the state value corresponding to the first symptom may be 1.
In the inquiry question output method provided by the embodiment of the application, firstly, disease prior probability prediction is carried out according to a first symptom of a first object to obtain the disease prior probability of the first object for the first symptom, then, a symptom state vector and the disease prior probability are input into a trained confrontation generation network, and an inquiry question of a second symptom is predicted and output through the trained confrontation generation network. The accuracy of the output inquiry question of the second symptom can be improved because the disease prior probability is considered in the process of outputting the inquiry question of the second symptom, and the symptom state vector is also considered, namely the inquiry question of the second symptom is predicted by combining the disease prior probability and the symptom state vector.
In one embodiment, after outputting the inquiry question of the second symptom, the method further comprises: receiving response information input by the first object for the inquiry questions of the second symptom; and updating the symptom state vector according to the answer information, adding the second symptom into the first symptom under the condition that the answer information indicates that the first object has the second symptom, and returning to the step of predicting the disease prior probability according to the first symptom of the first object to obtain the disease prior probability of the first object aiming at the first symptom until the disease prior probability meets the inquiry stopping condition.
And stopping outputting the inquiry question of the next predicted symptom until the disease prior probability meets the inquiry stopping condition, namely stopping outputting the inquiry question of the second symptom output by the trained confrontation generation network, and ending the output of the inquiry question. The second symptom is added to the first symptom, the first symptom is updated (for example, before updating, the first symptom comprises a symptom A, the output inquiry question of the second symptom is an inquiry question of the symptom B, the answer information of the first object is yes, namely, the symptom B exists, the symptom B can be added to the first symptom, and the updated first symptom comprises the symptom A and the symptom B), the step of predicting the disease prior probability according to the first symptom of the first object is returned to obtain the disease prior probability of the first object aiming at the first symptom, and the prediction of the disease prior probability is carried out again, at this time, the first symptom is updated and is the latest first symptom. And after the disease prior probability is predicted again, continuously inputting the symptom state vector and the disease prior probability which is predicted again into the trained confrontation generation network, and outputting the inquiry question of the second symptom again, namely the inquiry question of the second symptom which is output again at this time is different from the second symptom which is output before. And (5) sequentially circulating until the disease prior probability meets the inquiry stopping condition, stopping outputting the inquiry questions, and finishing outputting the inquiry questions. Therefore, by continuously predicting the disease prior probability and continuously predicting the inquiry question of the second symptom according to the disease prior probability and the first symptom until the disease prior probability meets the inquiry stopping condition, the continuously updated first symptom and the predicted disease prior probability are considered in the inquiry question output process, and the accuracy of the inquiry question is improved.
In one embodiment, before the disease prior probability prediction is performed according to the first symptom of the first subject and the disease prior probability of the first subject for the first symptom is obtained, the method further includes: receiving first content input by a first object; performing word segmentation processing on the first content to obtain medical related words; the first symptom is obtained according to the medical related word segmentation.
That is, the first object actively inputs the first content, and the first content may be understood as the subject matter of the first object, for example, the first object may input the symptom of itself in the form of a sentence, for example, "i am with headache today", or may input the symptom of itself in the form of a word, for example, "headache belly pain". After the first content is obtained, word segmentation processing can be performed on the first content, and the word segmentation processing mode has various modes, and the application is not limited. After the word segmentation processing is performed on the first content, the medical-related word segmentation can be obtained from the obtained multiple word segmentations (there are various ways, for example, the medical-related word segmentation can be identified by medical entity, and the like, and the present application is not limited thereto). After the medical related segmentation is obtained, the first symptom can be obtained according to the medical related segmentation, namely the first symptom is related to the medical related segmentation, so that the accuracy of the first symptom can be improved, and the accuracy of output of the inquiry question can be improved. There are various ways to obtain the first symptom according to the medical-related segmentation, and the present application is not limited thereto, for example, the medical-related segmentation can be used as the determined first symptom, and for example, the medical-related segmentation and other attribute features (for example, duration, signs, etc.) can be used as the first symptom.
In one embodiment, performing a disease prior probability prediction based on a first symptom of a first subject to obtain a disease prior probability of the first subject for the first symptom comprises: text feature extraction is carried out on the first symptom of the first object based on a trained combined prediction model based on text feature extraction and disease prior probability prediction, and disease prior probability prediction is carried out according to the extracted text feature, so that the disease prior probability of the first object aiming at the first symptom is obtained.
Then, the disease prior probability prediction is carried out according to the extracted text features through a disease prior probability prediction module in the combined prediction model, so that the disease prior probability of the first object for the first symptom is obtained, the disease prediction in a certain range can be realized, the possible diseases of the first object are locked in the certain range, and the network output narrowing range is generated for the subsequent confrontation. That is, in the embodiment, text feature extraction and disease prior prediction can be performed on the first symptom of the first object by using a combination mode of two deep neural networks, so that disease prediction in a certain range can be realized, possible diseases of the first object can be locked in a certain range, and a network output reduction range is generated for subsequent countermeasures. It should be noted that there may be many combinations of two deep neural networks, and an example of the present application provides a specific possible combination of a combined prediction model based on text feature extraction and disease prior probability prediction, that is, a combination of a Bi-directional long-and-short time memory unit (Bi-LSTM) and a Convolutional Neural Network (CNN) is used. As an example, the bidirectional long-and-short time memory unit (Bi-LSTM) and the convolutional neural network can be in a series combination or a parallel combination.
As an example, before performing text feature extraction on the first symptom of the first object based on the trained combined prediction model based on text feature extraction and disease prior probability prediction, the method further includes: and performing word vectorization on the first symptom of the first object to obtain a word vector of the first symptom. It will be appreciated that if a plurality of symptoms are included in the first symptom, a word vector for each of the first symptoms is obtained. In this way, the text feature extraction of the first symptom of the first object based on the trained combined prediction model based on text feature extraction and disease prior probability prediction includes: and performing text feature extraction on the word vector of the first symptom based on a trained combined prediction model based on text feature extraction and disease prior probability prediction. That is, what is input to the combined predictive model is the word vector for the first symptom. In one example, there are various techniques for performing Word vectorization, and the present application is not limited thereto, for example, the techniques for performing Word vectorization include Word2Vec, GloVe, and the like, and the Word vector is obtained by mapping the participles to the expression of the vector through the techniques for performing Word vectorization.
In one example, the trained combined prediction model based on text feature extraction and disease prior probability prediction can be trained by the following training modes:
acquiring a symptom sample of a training object and the category of the symptom sample;
acquiring a medical text data sample, and taking medical dialogue data as a label of the medical text data sample;
and training the combined prediction model based on the text feature extraction and the disease prior probability prediction according to the symptom sample, the category of the symptom sample, the medical text data sample and the label of the medical text data sample of the training object to obtain the trained combined prediction model based on the text feature extraction and the disease prior probability prediction.
In one example, obtaining a symptom sample and a category of the symptom sample for the training subject may include: acquiring input content of a training object, performing word segmentation processing on the input content to obtain training words related to medical treatment, and taking the training words related to medical treatment as symptom samples; and marking the training word segmentation class to obtain the class of the symptom sample.
In one example, the present application embodiments may build a specialized medical entity database for text in the medical field. Based on sentence segmentation of the medical entity library, word segmentation processing can be carried out on input contents of the training object to obtain corresponding training words. In one example, the input content of the training subject may be segmented into corresponding segments based on sentence segmentation of the medical entity library, and the segments related to medical treatment may be labeled with corresponding categories, such as "vomiting" labeled as "symptom", "pneumonia" labeled as "disease", and so on.
In one example, before a combined prediction model based on text feature extraction and disease prior probability prediction is trained according to a symptom sample, a category of the symptom sample, a medical text data sample and a label of medical text data of a training object, training participles related to medical treatment and the medical text data sample can be mapped to vector expressions by using a word vectorization technology, namely, word vectorization is carried out, and a training word vector is obtained. And then, training the combined prediction model based on the text feature extraction and the disease prior probability prediction through the training word vector to obtain the trained combined prediction model based on the text feature extraction and the disease prior probability prediction.
As an example, compared with a traditional neural network (e.g., DNN, RNN) framework, a combined prediction model based on a bidirectional long-and-short-term memory unit and a convolutional neural network considers the sequential relationship between words in a sentence and better conforms to the basic assumption of natural language processing (i.e., expression of word order influence semantics), and on the other hand, effectively solves the problems of gradient explosion (gradient expansion) and gradient diffusion (gradient variation) existing in the traditional Recurrent Neural Network (RNN), so that model training is more stable. Because the semantic features of words and sentences are considered, the combined prediction model is obtained based on the training method, the possible diseases of the patient can be predicted in a certain range, the possible diseases of the patient can be locked in the certain range, and the network output narrowing range is generated for subsequent countermeasures.
In one embodiment, the interrogation cessation condition includes the inclusion of a probability of interest in the prior probability of disease that is greater than a preset probability threshold.
The method comprises the steps of utilizing a first symptom of a first object to predict disease prior probability to obtain the disease prior probability of a plurality of diseases, determining that the disease prior probability meets an inquiry stopping condition under the condition that the disease prior probability has a target larger than a preset probability threshold, namely, determining that the disease prior probability meets the inquiry stopping condition every time the disease prior probability is predicted, comparing the obtained disease prior probability with the preset probability threshold, if the target probability is not larger than the preset threshold, continuing outputting an inquiry problem of next predicted inquiry symptom, and stopping outputting the inquiry problem and finishing outputting the inquiry problem once the disease prior probability comprises the target probability larger than the preset probability threshold and the disease prior probability meets the inquiry stopping condition. In one example, until the disease prior probability satisfies the inquiry stop condition, the method may further include: and outputting the target disease corresponding to the target probability, so that the first object can see the predicted target disease, and the first object can perform subsequent further verification, treatment and the like according to the target disease.
In one embodiment, until the disease prior probability satisfies the inquiry stop condition, the method further comprises: and outputting an inquiry sequence aiming at the first object, wherein the inquiry sequence comprises inquiry questions of the second symptoms which are ordered in the output time sequence.
That is, in this embodiment, the method further includes: until the disease prior probability satisfies an interrogation stop condition, an interrogation sequence for the first subject may be output. And in the process of continuously outputting the inquiry questions of the second symptoms, the first object replies aiming at different output second spicules to realize inquiry interaction, under the condition that the disease prior probability meets the inquiry stopping condition, the output of the inquiry question of the next predicted symptom is stopped, and an inquiry sequence aiming at the first object can be output, namely the inquiry questions of a plurality of second symptoms output in the process of outputting the inquiry questions are output as the inquiry sequence, so that the user can view and know the symptom condition of the user in the inquiry interaction process. It should be noted that the questions of the second symptom included in the inquiry sequence are ordered by the output time sequence (e.g., may be chronological) for better viewing of the first object.
In one embodiment, the trained counter-generating network is obtained by:
obtaining M sample inquiry sequences and a disease diagnosis label of each sample inquiry sequence, wherein M is a positive integer;
obtaining a plurality of sample symptom state vectors corresponding to each sample inquiry sequence respectively, and training disease prior probability aiming at sample symptoms existing in each sample symptom state vector;
and training the confrontation generation network according to a plurality of sample symptom state vectors respectively corresponding to each sample inquiry sequence and the training disease prior probability aiming at the sample symptom indicated in each sample symptom state vector to obtain the trained confrontation generation network.
The sample inquiry sequence comprises a plurality of sample symptoms which are ordered according to the sequence of the historical inquiry time, and a plurality of sample symptom state vectors corresponding to the sample inquiry sequence can be obtained according to the sequence of the historical inquiry time. For example, the sample inquiry sequence includes a sample symptom E, a sample symptom F, a sample symptom G, and a sample symptom H, where the sample symptom E and the sample symptom F are symptoms actively input by a sample object, and it can be understood that the historical inquiry time is input time and is the same, at this time, a state value corresponding to the sample symptom E and the sample symptom F in an obtained sample symptom state vector indicates that the corresponding object has the sample symptom E and the sample symptom F, and it can be understood that the obtained sample symptom state vector corresponds to the sample symptoms E and F, and the sample symptom indicated in the sample symptom state vector includes the sample symptoms E and F, and the corresponding training disease prior probability is obtained according to the sample symptom E and the sample symptom F indicated in the sample symptom state vector. The question of the inquiry of the sample symptom G is subsequently presented, it is known that the sample symptom G exists in the object by means of inquiry and the like, the historical inquiry time of the sample symptom G is later than the historical inquiry time of the sample symptom E, the state values corresponding to the sample symptom E, the sample symptom F and the sample symptom G in the next available sample symptom state vector indicate that the corresponding object has the sample symptom E, the sample symptom F and the sample symptom G, it can be understood that the sample symptom state vector obtained at this time corresponds to the sample symptoms E, F and G, the sample symptoms indicated to exist in the sample symptom state vector include the sample symptoms E, F and G, and the prior probability of the corresponding training disease is obtained according to the sample symptoms indicated to exist in the sample symptom state vector, i.e., the sample symptom E, the sample symptom F and the sample symptom G. And analogizing in sequence until sample symptom state vectors corresponding to the sample symptom E, the sample symptom F, the sample symptom G and the sample symptom H are obtained, wherein each sample state vector corresponds to a training disease prior probability. Thus, 3 sample symptom state vectors can be obtained for the sample interrogation sequence.
For each sample symptom state vector, the sample symptom present in the sample object corresponding to the sample symptom state vector can be determined, which belongs to the corresponding sample interrogation sequence, and thus, the prior probability of the training disease indicating the present sample symptom in each sample symptom state vector can be obtained. And training the confrontation generation network according to a plurality of sample symptom state vectors respectively corresponding to each sample inquiry sequence and the training disease prior probability indicating the existing sample symptoms in each sample symptom state vector to obtain the trained confrontation generation network, so that the training accuracy is improved.
The confrontation generating network comprises a generator (namely, a generating model) and a discriminator (a discriminating model), wherein the discriminator can be understood as a two-classification model, real probability (probability of the real sequence) can be predicted according to an inquiry question sequence of training symptoms output by the generator (a set of inquiry questions of the training symptoms output each time and ordered according to output time sequence), the real probability can be transmitted to the generator, the generator can continuously change the generator according to the real probability, namely, the generator is adjusted, so that the inquiry questions output by the generator are more accurate, the process is circulated until a training stopping condition is met, training of the confrontation generating network is completed, and the trained confrontation generating network is obtained, and comprises the trained generator and the trained discriminator.
It can be understood that the input of the generator includes a plurality of sample symptom state vectors corresponding to each sample inquiry sequence respectively and training disease prior probabilities indicating existing sample symptoms in each sample symptom state vector, the output of the generator is an inquiry question of a next training symptom, the input of the discriminator includes an inquiry question sequence of the training symptom output by the generator, the output of the discriminator is a true probability for the training symptom question sequence, and the output of the discriminator is input to the generator.
In one example, the second symptom interrogation question is output by a trained generator, i.e., the second symptom interrogation question is output by inputting the symptom state vector and the prior probability of the disease into a trained generator in a trained antagonistic generation network.
The generator of the countermeasure generation network can give the optimal selection symptom a (a ∈ Sunknown, which represents the set of symptoms that have not been queried currently) in the current state S, namely, give the inquiry question that needs to be presented next time of interaction when the current symptom is known to be observed.
The process of the above method is described in detail below with an embodiment.
And predicting the prior probability of the disease aiming at the main complaint content of the patient based on the two-way length memory unit and the convolutional neural network. The Bi-LSTM-based deep learning method is widely applied to various fields of natural language processing, including automatic text classification, emotion analysis and machine translation. In the embodiment of the application, the text feature extraction and classification method based on the deep learning method of Bi-LSTM and CNN can be applied to disease prior probability prediction according to the patient complaint content. A large number of medical text data samples are used as supports, professional medical dialogue data are used as labels of the medical text data samples, a combined prediction model can be trained, accuracy of the combined prediction model is improved, and the premise that network output is generated for follow-up countermeasures is met.
The specific training process is as follows, firstly, the main complaint content of the training object is participled, different from the participle of the traditional text, and aiming at the text in the medical field, a special medical entity database is established. Based on sentence segmentation of the medical entity library, the sentences can be segmented into corresponding segmented words, and meanwhile, the segmented words related to medical treatment can be marked with corresponding categories, such as 'vomiting' can be marked as 'symptom', and 'pneumonia' can be marked as 'disease'. Then, after acquiring the participles and after medical text data samples, we map the participles to the expression of the vector using Word vectorization technology (Word2Vec, GloVe). Then, text feature extraction is carried out on the word vectors by using a combined prediction model based on Bi-LSTM and CNN, and the output of the prior probability is trained. Compared with the traditional neural network (DNN, RNN) framework, the deep network model based on the Bi-LSTM and the CNN considers the sequence relation between words in sentences and better accords with the basic assumption of natural language processing (expression of language order influence semantics), and effectively solves the problems of gradient explosion (gradient expansion) and gradient diffusion (gradient variation) existing in the traditional Recurrent Neural Network (RNN), so that the model training is more stable. Because the semantic features of words and sentences are considered, the disease prior probability prediction model is obtained based on the training method, the possible diseases of the patient can be locked in a certain range in the disease prediction in a certain range, and the range is reduced for the symptom space output of the generator of the subsequent confrontation network.
Then, training against a network-generating Discriminator (Discriminator) is performed. In the conventional intelligent inquiry method, because the authenticity of the inquiry sequence is not judged, the generated inquiry sequence may be unnatural, and meanwhile, enough information may not be collected to finally give a diagnosis of the disease. Therefore, the countermeasure generation network is adopted and comprises a generator and a discriminator, the countermeasure generation network discriminator and the generator are trained, namely the training of the countermeasure generation network is completed, and the discriminator model is trained to distinguish whether the inquiry sequence generated by the generator is from the real inquiry data or from the generator. The real inquiry data is derived from real inquiry dialogue data of doctors and patients, and the judgment result of the final discriminator is fed back to the generator in a real probability mode for further optimizing the generator.
The specific process is as follows, and the existing medical inquiry dialogue database includes, but is not limited to, medical books, medical documents, medical reports, case analysis, etc., and also includes the inquiry data of patients and doctors generated by simulation. Since these medical data are elaborately compiled by the relevant medical experts, there is a standardized annotated format. Therefore, the inquiry information flow of doctors and patients, namely the sample inquiry sequence can be extracted from the databases, and the final disease diagnosis label of the inquiry information flow can be obtained. The above sample interrogation sequences can be used as real interrogation data. Thus, a discriminator D of the antagonistic generating network can be constructed, which determines, based on a deep network model, whether the sequence of inquiry questions generated by the generator is from the real inquiry data or from the generator G of the antagonistic network. And finally, the output result of the discriminator D is fed back to the generator in a true probability mode, and the inquiry problem generated by the generator G is guided to be more natural.
Secondly, based on training against the Generator (Generator) that generates the network. In the traditional intelligent inquiry method, the interaction between a doctor (an intelligent inquiry system) and a patient is not considered, so that the obtained inquiry question sequence may not be consistent with the inquiry logic of a real doctor. The inquiry question of the next symptom needing to be output is generated through a generator according to the existing large amount of inquiry data.
On the one hand, methods based on Bi-directional length memory cells (Bi-LSTM) and Convolutional Neural Networks (CNN) may give a candidate set of coarse-grained disease predictions within a certain range, but eventually require interactive interrogation logic to determine the disease step by step. On the other hand, for the discriminator and the generator against the generation network, under the known symptom of the current user, the optimum inquiry symptom under the current situation is output, and the discriminator considers whether the inquiry sequence can be more real after the addition of the symptom.
After the training is generated, the inquiry question of the next symptom (second symptom) can be predicted according to the symptom state vector and the disease prior probability of the training-completed confrontation generation network. As shown in fig. 2, first, a first object inputs a first content (i.e. the main complaint content in fig. 2), performs word segmentation on the first content to obtain a word segment related to medical treatment, i.e. a first symptom, performs word vectorization on the word segment to obtain a word vector of the first symptom, performs disease prior probability prediction according to the word vector of the first symptom by using a combined prediction model to obtain a disease prior probability for the first symptom of the first object, and performs prediction of an inquiry question a (i.e. a next symptom inquiry action) of a next symptom through a trained countermeasure generation network in combination with the disease prior probability and the first symptom, for example, the output inquiry question of the next symptom is a, the first object can answer the inquiry question of the given next symptom a, complete an inquiry interaction, and update a set S (for example, the symptom state vector is updated according to the answer information of the first object, and updating the first symptom according to the answer information, obtaining a new disease prior probability according to the updated first symptom, so that the symptom state vector is updated and the disease prior probability is updated, and then S is updated accordingly), then predicting the inquiry question of the next symptom according to the updated set S by the trained generator, and then sequentially and iteratively completing the interaction between the patient and a doctor (an intelligent inquiry system), until the disease prior probability comprises a target probability which is greater than a preset probability threshold, stopping the output of the inquiry question of the next symptom, finishing the intelligent inquiry, giving the inquiry sequence and the target disease generated by the generator, and completing the whole disease inquiry and prediction process.
According to the method, how to simulate the inquiry logic of the doctor and the response of the patient are considered in the training process, so that the inquiry experience is optimized under the condition of improving the problem accuracy, the inquiry is more consistent with the logic of the doctor, and the reasonability and the user experience of the inquiry logic of the related diseases are optimized.
As shown in fig. 3, in one embodiment, there is also provided an inquiry question output device 300, which includes:
the prior probability prediction module 301 is configured to perform disease prior probability prediction according to a first symptom of the first object, so as to obtain a disease prior probability of the first object for the first symptom.
An output module 302, configured to input a symptom state vector and the disease prior probability into the trained confrontation generating network, and output an inquiry question of a second symptom, where a state value corresponding to the first symptom in the symptom state vector indicates that the first symptom exists in the first object.
As shown in fig. 4, in an embodiment, the apparatus further includes:
an answer receiving module 303, configured to receive answer information input by the first object for the second symptom.
An updating module 304, configured to update the symptom state vector according to the answer information, and in a case that the answer information indicates that the first subject has the second symptom, add the second symptom to the first symptom, and return to the step of performing disease prior probability prediction according to the first symptom of the first subject, so as to obtain a disease prior probability of the first subject for the first symptom until the disease prior probability satisfies an inquiry stop condition.
In one embodiment, the above apparatus further comprises:
and the content receiving module is used for receiving the first content input by the first object.
And the word segmentation module is used for carrying out word segmentation processing on the first content to obtain medical related words.
A first needle-shaped acquisition module for obtaining the first symptom according to the medical-related segmentation.
In one embodiment, the interrogation stop condition includes that the prior probability of the disease includes a target probability greater than a preset probability threshold.
In one embodiment, the above apparatus further comprises:
and the sequence output module is used for outputting an inquiry sequence aiming at the first object until the disease prior probability meets an inquiry stopping condition, wherein the inquiry sequence comprises inquiry questions of a second symptom which are sequenced in an output time sequence.
In one embodiment, the above apparatus further comprises:
the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring M sample inquiry sequences and a disease diagnosis label of each sample inquiry sequence, and M is a positive integer;
a second obtaining module, configured to obtain a plurality of sample symptom state vectors corresponding to each sample inquiry sequence, and a priori probability of a training disease indicating existing sample symptoms in each sample symptom state vector;
and the training module is used for training the confrontation generation network according to a plurality of sample symptom state vectors respectively corresponding to each sample inquiry sequence and the training disease prior probability aiming at the sample symptom indicated to exist in each sample symptom state vector to obtain the trained confrontation generation network.
The inquiry question output device of each embodiment is a device for implementing the inquiry question output method of each embodiment, and has corresponding technical features and technical effects, and details are not repeated herein.
According to an embodiment of the present application, an electronic device and a readable storage medium are also provided.
Fig. 5 is a block diagram of an electronic device according to an inquiry question output method in an embodiment of the present application. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the present application that are described and/or claimed herein.
As shown in fig. 5, the electronic apparatus includes: one or more processors 501, memory 502, and interfaces for connecting the various components, including high-speed interfaces and low-speed interfaces. The various components are interconnected using different buses and may be mounted on a common motherboard or in other manners as desired. The processor may process instructions for execution within the electronic device, including instructions stored in or on the memory to display graphical information of a GUI on an external input/output apparatus (such as a display device coupled to the interface). In other embodiments, multiple processors and/or multiple buses may be used, along with multiple memories and multiple memories, as desired. Also, multiple electronic devices may be connected, with each device providing portions of the necessary operations (e.g., as a server array, a group of blade servers, or a multi-processor system). In fig. 5, one processor 501 is taken as an example.
The memory 502, which is a non-transitory computer readable storage medium, may be used to store non-transitory software programs, non-transitory computer executable programs, and modules, such as program instructions/modules (e.g., the prior probability prediction module 301, the output module 303 shown in fig. 3) corresponding to the output method of the inquiry question in the embodiment of the present application. The processor 501 executes various functional applications of the server and data processing, i.e., implementing the inquiry question output method in the above-described method embodiments, by running non-transitory software programs, instructions, and modules stored in the memory 502.
The memory 502 may include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store data created according to use of the electronic device displayed by the keyboard, and the like. Further, the memory 502 may include high speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, memory 502 may optionally include memory located remotely from processor 501, which may be connected to keyboard display electronics over a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The electronic device of the inquiry question output method may further include: an input device 503 and an output device 504. The processor 501, the memory 502, the input device 503 and the output device 504 may be connected by a bus or other means, and fig. 5 illustrates the connection by a bus as an example.
The input device 503 may receive input numeric or character information and generate key signal inputs related to user settings and function controls of the electronic device displayed by the keyboard, such as a touch screen, keypad, mouse, track pad, touch pad, pointer stick, one or more mouse buttons, track ball, joystick, or other input device. The output devices 804 may include a display device, auxiliary lighting devices (e.g., LEDs), and haptic feedback devices (e.g., vibrating motors), among others. The display device may include, but is not limited to, a Liquid Crystal Display (LCD), a Light Emitting Diode (LED) display, and a plasma display. In some implementations, the display device can be a touch screen.
Various implementations of the systems and techniques described here can be realized in digital electronic circuitry, integrated circuitry, application specific ASICs (application specific integrated circuits), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, receiving data and instructions from, and transmitting data and instructions to, a storage system, at least one input device, and at least one output device.
These computing programs (also known as programs, software applications, or code) include machine instructions for a programmable processor, and may be implemented using procedural and/or object oriented programming languages, and/or assembly/machine languages. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and/or device (e.g., magnetic discs, optical disks, memory, Programmable Logic Devices (PLDs)) used to provide machine instructions and/or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal used to provide machine instructions and/or data to a programmable processor.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic, speech, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), Wide Area Networks (WANs), and the Internet.
The computer system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.
According to the technical scheme of the embodiment of the application, the first sub-keyboard and the second sub-keyboard can be generated under the condition that the width of the electronic equipment is larger than the first preset width, and the first sub-keyboard and the second sub-keyboard are displayed at intervals, namely, the first sub-keyboard and the second sub-keyboard have intervals, so that a user does not need to perform key operation in the intervals, the user can easily touch keys in the keyboard in the operation process, the operation path of the user on the keyboard can be shortened, and the input efficiency is further improved.
It should be understood that various forms of the flows shown above may be used, with steps reordered, added, or deleted. For example, the steps described in the present application may be executed in parallel, sequentially, or in different orders, and the present invention is not limited thereto as long as the desired results of the technical solutions disclosed in the present application can be achieved.
The above-described embodiments should not be construed as limiting the scope of the present application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may be made in accordance with design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims (14)
1. An inquiry question output method, comprising:
predicting disease prior probability according to a first symptom of a first object to obtain the disease prior probability of the first object aiming at the first symptom;
inputting a symptom state vector and the disease prior probability into a trained countermeasure generation network, and outputting an inquiry question of a second symptom, wherein a state value corresponding to the first symptom in the symptom state vector indicates that the first symptom exists in the first object.
2. The method of claim 1, wherein outputting the second symptom of the interrogation question further comprises:
receiving response information input by the first subject for the second symptom;
updating the symptom state vector according to the answer information, adding the second symptom into the first symptom under the condition that the answer information indicates that the first object has the second symptom, and returning to the step of predicting the prior disease probability according to the first symptom of the first object to obtain the prior disease probability of the first object aiming at the first symptom until the prior disease probability meets an inquiry stopping condition.
3. The method of claim 1, wherein the predicting a prior probability of disease based on the first symptom of the first subject, prior to obtaining the prior probability of disease for the first symptom, further comprises:
receiving first content input by the first object;
performing word segmentation processing on the first content to obtain medical related words;
and obtaining the first symptom according to the medical related word segmentation.
4. The method of claim 1, wherein the interrogation cessation condition comprises an inclusion of a target probability in the disease prior probability that is greater than a preset probability threshold.
5. The method of claim 2, further comprising, until the disease prior probability satisfies an interrogation stop condition: outputting an interrogation sequence for the first subject, wherein the interrogation sequence comprises interrogation questions of the second symptoms ordered in output time order.
6. The method of claim 1, wherein the trained challenge generating network is obtained by:
obtaining M sample inquiry sequences and a disease diagnosis label of each sample inquiry sequence, wherein M is a positive integer;
obtaining a plurality of sample symptom state vectors corresponding to each sample inquiry sequence respectively, and training disease prior probability aiming at sample symptoms existing in each sample symptom state vector;
training the confrontation generation network according to a plurality of sample symptom state vectors respectively corresponding to each sample inquiry sequence and the training disease prior probability aiming at the sample symptom indicated to exist in each sample symptom state vector to obtain the trained confrontation generation network.
7. An inquiry question output apparatus, characterized in that the apparatus comprises:
the prior probability prediction module is used for predicting disease prior probability according to a first symptom of a first object to obtain the disease prior probability of the first object aiming at the first symptom;
and the output module is used for inputting the symptom state vector and the disease prior probability into the trained confrontation generation network and outputting the inquiry question of the second symptom, wherein the state value corresponding to the first symptom in the symptom state vector indicates that the first symptom exists in the first object.
8. The apparatus of claim 7, further comprising:
an answer receiving module for receiving answer information input by the first object for the question of the second symptom;
and an updating module, configured to update the symptom state vector according to the answer information, add the second symptom to the first symptom when the answer information indicates that the first subject has the second symptom, and return to the step of predicting the disease prior probability according to the first symptom of the first subject to obtain the disease prior probability of the first subject for the first symptom until the disease prior probability satisfies an inquiry stop condition.
9. The apparatus of claim 7, further comprising:
the content receiving module is used for receiving first content input by the first object;
the word segmentation module is used for carrying out word segmentation processing on the first content to obtain medical related words;
a first needle-shaped acquisition module for obtaining the first symptom according to the medical-related segmentation.
10. The apparatus of claim 7, wherein the interrogation cessation condition comprises an inclusion of a target probability in the disease prior probability that is greater than a preset probability threshold.
11. The apparatus of claim 7, further comprising:
and the sequence output module is used for outputting an inquiry sequence aiming at the first object until the disease prior probability meets an inquiry stopping condition, wherein the inquiry sequence comprises inquiry questions of a second symptom which are sequenced in an output time sequence.
12. The apparatus of claim 7, further comprising:
the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring M sample inquiry sequences and a disease diagnosis label of each sample inquiry sequence, and M is a positive integer;
a second obtaining module, configured to obtain a plurality of sample symptom state vectors corresponding to each sample inquiry sequence, and a priori probability of a training disease indicating existing sample symptoms in each sample symptom state vector;
training the confrontation generation network according to a plurality of sample symptom state vectors respectively corresponding to each sample inquiry sequence and the training disease prior probability aiming at the sample symptom indicated to exist in each sample symptom state vector to obtain the trained confrontation generation network.
13. An electronic device, comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein the content of the first and second substances,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-6.
14. A non-transitory computer readable storage medium having stored thereon computer instructions for causing the computer to perform the method of any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010091503.XA CN111326251B (en) | 2020-02-13 | 2020-02-13 | Question output method and device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010091503.XA CN111326251B (en) | 2020-02-13 | 2020-02-13 | Question output method and device and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111326251A true CN111326251A (en) | 2020-06-23 |
| CN111326251B CN111326251B (en) | 2023-08-29 |
Family
ID=71172647
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010091503.XA Active CN111326251B (en) | 2020-02-13 | 2020-02-13 | Question output method and device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111326251B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112750529A (en) * | 2020-12-31 | 2021-05-04 | 平安科技（深圳）有限公司 | Intelligent medical inquiry device, equipment and medium |
| CN112768064A (en) * | 2021-01-26 | 2021-05-07 | 北京搜狗科技发展有限公司 | Disease prediction device, disease prediction apparatus, symptom information processing method, symptom information processing device, and symptom information processing apparatus |
| CN113223648A (en) * | 2021-05-08 | 2021-08-06 | 北京嘉和海森健康科技有限公司 | Pre-diagnosis information acquisition method and device |
| CN113409907A (en) * | 2021-07-19 | 2021-09-17 | 广州方舟信息科技有限公司 | Intelligent pre-inquiry method and system based on Internet hospital |
| WO2023029502A1 (en) * | 2021-08-30 | 2023-03-09 | 康键信息技术（深圳）有限公司 | Method and apparatus for constructing user portrait on the basis of inquiry session, device, and medium |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2001085021A1 (en) * | 2000-05-08 | 2001-11-15 | Schmitt Armand J | System and method for determining the probable existence of disease |
| EP3107061A1 (en) * | 2014-02-12 | 2016-12-21 | Shimura, Akiyoshi | Disease detection system and disease detection method |
| CN107103201A (en) * | 2017-05-10 | 2017-08-29 | 北京大数医达科技有限公司 | Generation method, device and the medical path air navigation aid of medical guidance path |
| US20180226143A1 (en) * | 2017-02-08 | 2018-08-09 | Groundbreaking Technology Llc | Medical diagnostic platform |
| US20180365381A1 (en) * | 2017-06-16 | 2018-12-20 | Htc Corporation | Computer aided medical method and medical system for medical prediction |
| CN109192300A (en) * | 2018-08-17 | 2019-01-11 | 百度在线网络技术（北京）有限公司 | Intelligent way of inquisition, system, computer equipment and storage medium |
| CN109524108A (en) * | 2018-11-21 | 2019-03-26 | 上海依智医疗技术有限公司 | A kind of method for building up and device of inquiry model |
| WO2019086867A1 (en) * | 2017-10-31 | 2019-05-09 | Babylon Partners Limited | A computer implemented determination method and system |
| CN109817329A (en) * | 2019-01-21 | 2019-05-28 | 暗物智能科技(广州)有限公司 | A kind of medical treatment interrogation conversational system and the intensified learning method applied to the system |
| US20190180841A1 (en) * | 2017-10-31 | 2019-06-13 | Babylon Partners Limited | Computer implemented determination method and system |
| CN110246572A (en) * | 2019-05-05 | 2019-09-17 | 清华大学 | Method and system are examined in a kind of medical treatment based on term vector point |
-
2020
- 2020-02-13 CN CN202010091503.XA patent/CN111326251B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2001085021A1 (en) * | 2000-05-08 | 2001-11-15 | Schmitt Armand J | System and method for determining the probable existence of disease |
| EP3107061A1 (en) * | 2014-02-12 | 2016-12-21 | Shimura, Akiyoshi | Disease detection system and disease detection method |
| US20180226143A1 (en) * | 2017-02-08 | 2018-08-09 | Groundbreaking Technology Llc | Medical diagnostic platform |
| CN107103201A (en) * | 2017-05-10 | 2017-08-29 | 北京大数医达科技有限公司 | Generation method, device and the medical path air navigation aid of medical guidance path |
| US20180365381A1 (en) * | 2017-06-16 | 2018-12-20 | Htc Corporation | Computer aided medical method and medical system for medical prediction |
| WO2019086867A1 (en) * | 2017-10-31 | 2019-05-09 | Babylon Partners Limited | A computer implemented determination method and system |
| US20190180841A1 (en) * | 2017-10-31 | 2019-06-13 | Babylon Partners Limited | Computer implemented determination method and system |
| CN109192300A (en) * | 2018-08-17 | 2019-01-11 | 百度在线网络技术（北京）有限公司 | Intelligent way of inquisition, system, computer equipment and storage medium |
| CN109524108A (en) * | 2018-11-21 | 2019-03-26 | 上海依智医疗技术有限公司 | A kind of method for building up and device of inquiry model |
| CN109817329A (en) * | 2019-01-21 | 2019-05-28 | 暗物智能科技(广州)有限公司 | A kind of medical treatment interrogation conversational system and the intensified learning method applied to the system |
| CN110246572A (en) * | 2019-05-05 | 2019-09-17 | 清华大学 | Method and system are examined in a kind of medical treatment based on term vector point |
Non-Patent Citations (2)
| Title |
|---|
| XIUZHANG LI，等: "The effects of an online inquiry service on gatekeeping systems with heterogeneous patients" * |
| 肖敏: "基于大数据的问诊推荐***的研究与实现" * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112750529A (en) * | 2020-12-31 | 2021-05-04 | 平安科技（深圳）有限公司 | Intelligent medical inquiry device, equipment and medium |
| CN112768064A (en) * | 2021-01-26 | 2021-05-07 | 北京搜狗科技发展有限公司 | Disease prediction device, disease prediction apparatus, symptom information processing method, symptom information processing device, and symptom information processing apparatus |
| CN113223648A (en) * | 2021-05-08 | 2021-08-06 | 北京嘉和海森健康科技有限公司 | Pre-diagnosis information acquisition method and device |
| CN113223648B (en) * | 2021-05-08 | 2023-10-24 | 北京嘉和海森健康科技有限公司 | Pre-diagnosis information acquisition method and device |
| CN113409907A (en) * | 2021-07-19 | 2021-09-17 | 广州方舟信息科技有限公司 | Intelligent pre-inquiry method and system based on Internet hospital |
| WO2023029502A1 (en) * | 2021-08-30 | 2023-03-09 | 康键信息技术（深圳）有限公司 | Method and apparatus for constructing user portrait on the basis of inquiry session, device, and medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111326251B (en) | 2023-08-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111428008B (en) | Method, apparatus, device and storage medium for training a model | |
| CN112560912B (en) | Classification model training method and device, electronic equipment and storage medium | |
| JP7317791B2 (en) | Entity linking method, device, apparatus and storage medium | |
| CN112507715B (en) | Method, device, equipment and storage medium for determining association relation between entities | |
| CN112560479B (en) | Abstract extraction model training method, abstract extraction device and electronic equipment | |
| CN111326251A (en) | Method and device for outputting inquiry questions and electronic equipment | |
| CN111079442B (en) | Vectorization representation method and device of document and computer equipment | |
| CN111753543A (en) | Medicine recommendation method and device, electronic equipment and storage medium | |
| CN111639710A (en) | Image recognition model training method, device, equipment and storage medium | |
| KR20210075036A (en) | Data set processing method, device, electronic equipment and storage medium | |
| CN107491655B (en) | Liver disease information intelligent consultation system based on machine learning | |
| US11361002B2 (en) | Method and apparatus for recognizing entity word, and storage medium | |
| CN110517767B (en) | Auxiliary diagnosis method, auxiliary diagnosis device, electronic equipment and storage medium | |
| CN112507700A (en) | Event extraction method and device, electronic equipment and storage medium | |
| CN111680145A (en) | Knowledge representation learning method, device, equipment and storage medium | |
| CN112530576A (en) | Online doctor-patient matching method and device, electronic equipment and storage medium | |
| CN111079945B (en) | End-to-end model training method and device | |
| CN111274397B (en) | Method and device for establishing entity relation detection model | |
| CN110674260B (en) | Training method and device of semantic similarity model, electronic equipment and storage medium | |
| CN111539209B (en) | Method and apparatus for entity classification | |
| CN112560505A (en) | Recognition method and device of conversation intention, electronic equipment and storage medium | |
| CN113220836A (en) | Training method and device of sequence labeling model, electronic equipment and storage medium | |
| CN112216359A (en) | Medical data verification method and device and electronic equipment | |
| CN112329453B (en) | Method, device, equipment and storage medium for generating sample chapter | |
| US20210311979A1 (en) | Method for retrieving multi-turn dialogue, storage medium, and electronic device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |