CN115132303A - Physiological label prediction method, model training method, device, equipment and medium - Google Patents
Physiological label prediction method, model training method, device, equipment and medium Download PDFInfo
- Publication number
- CN115132303A CN115132303A CN202210469503.8A CN202210469503A CN115132303A CN 115132303 A CN115132303 A CN 115132303A CN 202210469503 A CN202210469503 A CN 202210469503A CN 115132303 A CN115132303 A CN 115132303A
- Authority
- CN
- China
- Prior art keywords
- physiological
- symptom
- label
- target
- prediction
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 124
- 238000012549 training Methods 0.000 title claims abstract description 57
- 208000024891 symptom Diseases 0.000 claims abstract description 677
- 239000013589 supplement Substances 0.000 claims abstract description 36
- 238000012545 processing Methods 0.000 claims description 51
- 230000004962 physiological condition Effects 0.000 claims description 28
- 230000002159 abnormal effect Effects 0.000 claims description 27
- 230000015654 memory Effects 0.000 claims description 18
- 238000003860 storage Methods 0.000 claims description 16
- 230000000153 supplemental effect Effects 0.000 claims description 15
- 230000001502 supplementing effect Effects 0.000 claims description 11
- 238000004590 computer program Methods 0.000 claims description 9
- 230000003993 interaction Effects 0.000 abstract description 9
- 230000008569 process Effects 0.000 abstract description 8
- 238000013473 artificial intelligence Methods 0.000 abstract description 3
- 238000013461 design Methods 0.000 description 39
- 238000010586 diagram Methods 0.000 description 23
- 230000006870 function Effects 0.000 description 10
- 238000003062 neural network model Methods 0.000 description 8
- 230000005856 abnormality Effects 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 6
- 230000009469 supplementation Effects 0.000 description 6
- 239000011159 matrix material Substances 0.000 description 4
- 230000033228 biological regulation Effects 0.000 description 3
- 238000012512 characterization method Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 201000010099 disease Diseases 0.000 description 3
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 3
- 206010011878 Deafness Diseases 0.000 description 2
- 230000000295 complement effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000010370 hearing loss Effects 0.000 description 2
- 231100000888 hearing loss Toxicity 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 208000034693 Laceration Diseases 0.000 description 1
- AZFKQCNGMSSWDS-UHFFFAOYSA-N MCPA-thioethyl Chemical compound CCSC(=O)COC1=CC=C(Cl)C=C1C AZFKQCNGMSSWDS-UHFFFAOYSA-N 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013475 authorization Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 230000005055 memory storage Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000009877 rendering Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
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
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
- G16H10/20—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for electronic clinical trials or questionnaires
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The application discloses a physiological label prediction method, a model training method, a device, equipment and a medium, and belongs to the technical field of artificial intelligence. The method comprises the following steps: acquiring an input first physiological symptom; determining a second physiological symptom associated with the first physiological symptom; predicting the first physiological symptom and the second physiological symptom to obtain a target physiological label and a prediction probability; under the condition that the prediction probability is larger than a target threshold value, outputting a target physiological label; and determining a symptom supplement guide statement according to the target physiological label under the condition that the prediction probability is not greater than the target threshold. Under the condition that the first physiological symptom of the input target object is insufficient, the prediction basis in the process of predicting the physiological label is expanded, and the target physiological label is utilized to pertinently guide the user to input the physiological symptom; under the condition of ensuring the accuracy of the physiological label prediction, the human-computer interaction efficiency of the physiological label prediction is improved.
Description
Technical Field
The application relates to the technical field of artificial intelligence, in particular to a physiological label prediction method, a model training method, a device, equipment and a medium.
Background
With the development of computer technology, the application scenarios of physiological label prediction are expanding continuously, and the corresponding physiological label needs to be predicted accurately according to physiological symptoms.
In the related art, if a user has limited input of a chief physiological symptom of a target subject, it is necessary to supplement a plurality of physiological symptoms of the target subject by filling in a questionnaire template corresponding to the chief physiological symptom. Furthermore, the neural network model is utilized to predict the chief complaint physiological symptoms and a plurality of physiological symptoms input by the user, and corresponding physiological labels are obtained. In the prediction process, the neural network model is trained through a large amount of sample data, so that the prediction accuracy of the physiological label can be improved.
However, obtaining multiple physiological symptoms through the questionnaire template requires the user to fill all questions in the questionnaire one by one, and how to improve the efficiency of obtaining multiple physiological symptoms is a problem to be solved.
Disclosure of Invention
The application provides a physiological label prediction method, a model training method, a device, equipment and a medium, and the technical scheme is as follows:
according to an aspect of the present application, there is provided a physiological label prediction method, the method including:
acquiring an input first physiological symptom, wherein the first physiological symptom is used for describing a physiological abnormal state of a target object;
determining a second physiological symptom associated with the first physiological symptom, wherein the second physiological symptom comprises at least one physiological symptom having an association relation with the first physiological symptom;
predicting the first physiological symptom and the second physiological symptom to obtain a target physiological label and a prediction probability of the target physiological label, wherein the target physiological label is used for describing a physiological abnormity type of the target object;
outputting the target physiological label under the condition that the prediction probability of the target physiological label is greater than a target threshold value;
and under the condition that the prediction probability of the target physiological label is not greater than the target threshold value, determining a symptom supplement guide statement according to the target physiological label, wherein the symptom supplement guide statement is used for guiding supplement of the first physiological symptom.
In an alternative design of the subject application, the determining a second physiological condition associated with the first physiological condition includes: determining a second physiological symptom associated with the first physiological symptom based on symptom associated information;
wherein the symptom associated information is used for indicating the degree of correlation between different physiological symptoms.
In an optional design of the present application, the symptom correlation information includes a correlation between different physiological symptoms;
the determining a second physiological symptom associated with the first physiological symptom based on symptom associated information includes:
determining a second physiological symptom associated with the first physiological symptom based on an association relationship between different physiological symptoms.
In an alternative design of the present application, the determining a symptom supplementary guidance statement according to the target physiological label includes: and generating the symptom supplementing guide statement according to a third physiological symptom corresponding to the target physiological label.
In an alternative design of the present application, the symptom supplement guide sentence includes a guide question and a guide answer option, the guide answer option is used for providing supplementary content of the first physiological symptom, the guide question is used for prompting selection of the guide answer option, and the guide question is predefined;
the generating the symptom supplementing guide statement according to a third physiological symptom corresponding to the target physiological label comprises: determining the guide answer option according to the third physiological symptom.
In an alternative design of the present application, the third physiological symptom includes a common physiological symptom and a unique physiological symptom, the common physiological symptom corresponds to more than a physiological tags of the target physiological tags, and the unique physiological symptom corresponds to no more than a physiological tags of the target physiological tags; wherein a is a positive integer not exceeding the number of target physiological labels;
the determining the guide answer option according to the third physiological symptom comprises:
determining the leading answer option based on the unique physiological symptom.
In an alternative design of the present application, the method further includes:
reconstructing the first physiological symptom and the second physiological symptom according to prompt information to obtain reconstruction information;
the prompt information comprises a slot position and prompt content, the slot position is used for providing a filling position of the target physiological label, and the prompt content is used for prompting to fill the slot position;
the predicting the first physiological symptom and the second physiological symptom to obtain a target physiological label and a predicted probability of the target physiological label includes:
and predicting the reconstruction information to obtain a target physiological label and the prediction probability of the target physiological label, wherein the target physiological label is used for filling the slot position.
In an optional design of the present application, the reconstructing the first physiological symptom and the second physiological symptom according to the prompt information to obtain reconstruction information includes:
superposing the prompt information with the first physiological symptom and the second physiological symptom, and splicing to obtain the reconstruction information;
or, encoding the first physiological symptom and the second physiological symptom to obtain a second characteristic representation;
and calculating to obtain the reconstruction information according to the second feature representation and the prompt information.
In an optional design of the present application, the performing the prediction processing on the first physiological symptom and the second physiological symptom to obtain a target physiological label and a prediction probability of the target physiological label includes:
coding the first physiological symptom and the second physiological symptom to obtain a second characteristic representation;
and performing classification prediction on the second feature representation to obtain the target physiological label and the prediction probability of the target physiological label.
According to another aspect of the present application, there is provided a training method of a physiological label prediction model, the method comprising:
obtaining a sample physiological symptom and a sample physiological label, wherein the sample physiological symptom is used for describing a physiological abnormal state of a sample object, and the sample physiological label is used for describing a physiological abnormal type of the sample object;
determining a supplemental physiological symptom associated with the sample physiological symptom, the supplemental physiological symptom comprising at least one physiological symptom having an association relationship with the sample physiological symptom;
calling an initial physiological label prediction model to perform prediction processing on the sample physiological symptoms and the supplementary physiological symptoms to obtain a predicted physiological label and a predicted probability of the predicted physiological label;
and carrying out backward propagation training on the initial physiological label prediction model based on the prediction error between the predicted physiological label and the sample physiological label to obtain the physiological label prediction model.
According to another aspect of the present application, there is provided a physiological signature prediction apparatus, the apparatus comprising:
the acquisition module is used for acquiring an input first physiological symptom, and the first physiological symptom is used for describing a physiological abnormal state of a target object;
a determination module for determining a second physiological symptom associated with the first physiological symptom, wherein the second physiological symptom comprises at least one physiological symptom having an association relation with the first physiological symptom;
the prediction module is used for performing prediction processing on the first physiological symptom and the second physiological symptom to obtain a target physiological label and a prediction probability of the target physiological label, wherein the target physiological label is used for describing a physiological abnormity type of the target object;
an output module, configured to output the target physiological label if the predicted probability of the target physiological label is greater than a target threshold;
the output module is further configured to determine, according to the target physiological label, a symptom supplement guidance statement for guiding supplement of the first physiological symptom when the prediction probability of the target physiological label is not greater than the target threshold.
In an alternative design of the present application, the determining module is further configured to:
determining a second physiological symptom associated with the first physiological symptom based on symptom associated information;
wherein the symptom associated information is used for indicating the degree of correlation between different physiological symptoms.
In an optional design of the present application, the symptom association information includes association relationships between different physiological symptoms belonging to the same physiological label and the physiological label;
the determining module is further configured to:
performing prediction processing on the first physiological symptom to obtain an auxiliary physiological label;
determining a second physiological symptom associated with the first physiological symptom based on an association relationship between a different physiological symptom belonging to the secondary physiological label and the secondary physiological label.
In an alternative design of the present application, the determining module is further configured to:
coding the first physiological symptom to obtain a first characteristic representation of the first physiological symptom;
and performing classification prediction on the first feature representation to obtain the auxiliary physiological label.
In an optional design of the present application, the symptom correlation information includes a correlation between different physiological symptoms;
the determining module is further configured to:
determining a second physiological symptom associated with the first physiological symptom based on an association relationship between different physiological symptoms.
In an optional design of the present application, the output module is further configured to: and generating the symptom supplementing and guiding statement according to a third physiological symptom corresponding to the target physiological label.
In an optional design of the present application, the symptom supplement guide sentence includes a guide question sentence and a guide answer option, the guide answer option is used for providing supplement content of the first physiological symptom, the guide question sentence is used for prompting selection of the guide answer option, and the guide question sentence is predefined;
the output module is further configured to: determining the guide answer option according to the third physiological symptom.
In an alternative design of the present application, the third physiological symptom includes a common physiological symptom and a unique physiological symptom, the common physiological symptom corresponds to more than a physiological tags of the target physiological tags, and the unique physiological symptom corresponds to no more than a physiological tags of the target physiological tags; wherein a is a positive integer not exceeding the number of target physiological labels;
the output module is further configured to: determining the leading answer option based on the unique physiological symptom.
In an alternative design of the present application, the apparatus further includes:
the construction module is used for reconstructing the first physiological symptom and the second physiological symptom according to prompt information to obtain reconstruction information; the prompt information comprises a slot position and prompt content, the slot position is used for providing a filling position of the target physiological label, and the prompt content is used for prompting to fill the slot position;
the prediction module is further configured to: and predicting the reconstruction information to obtain a target physiological label and the prediction probability of the target physiological label, wherein the target physiological label is used for filling the slot position.
In an alternative design of the present application, the building module is further configured to:
superposing the prompt information with the first physiological symptom and the second physiological symptom, and splicing to obtain the reconstruction information;
or, encoding the first physiological symptom and the second physiological symptom to obtain a second characteristic representation;
and calculating to obtain the reconstruction information according to the second feature representation and the prompt information.
In an alternative design of the present application, the prediction module is further configured to:
coding the first physiological symptom and the second physiological symptom to obtain a second characteristic representation;
and performing classification prediction on the second feature representation to obtain the target physiological label and the prediction probability of the target physiological label.
According to another aspect of the present application, there is provided a training apparatus for a physiological label prediction model, the apparatus including:
the acquisition module is used for acquiring a sample physiological symptom and a sample physiological label, wherein the sample physiological symptom is used for describing a physiological abnormal state of a sample object, and the sample physiological label is used for describing a physiological abnormal type of the sample object;
a determination module for determining a supplemental physiological symptom associated with the sample physiological symptom, the supplemental physiological symptom comprising at least one physiological symptom having an association relationship with the sample physiological symptom;
the prediction module is used for calling an initial physiological label prediction model to carry out prediction processing on the sample physiological symptoms and the supplementary physiological symptoms to obtain a predicted physiological label and a prediction probability of the predicted physiological label;
and the training module is used for carrying out backward propagation training on the initial physiological label prediction model based on the prediction error between the predicted physiological label and the sample physiological label to obtain the physiological label prediction model.
According to another aspect of the present application, there is provided a computer device comprising a processor and a memory, the memory having stored therein at least one instruction, at least one program, a set of codes, or a set of instructions, which is loaded and executed by the processor to implement the physiological label prediction method and/or the training method of the physiological label prediction model as described above.
According to another aspect of the present application, there is provided a computer readable storage medium having stored therein at least one instruction, at least one program, set of codes or set of instructions, which is loaded and executed by a processor to implement the physiological label prediction method and/or the training method of the physiological label prediction model as described above.
According to another aspect of the present application, a computer program product is provided, the computer program product comprising computer instructions stored in a computer readable storage medium, which are read and executed by a processor to implement the physiological label prediction method and/or the training method of the physiological label prediction model as described above.
The beneficial effect that technical scheme that this application provided brought includes at least:
under the condition that the input first physiological symptom of the target object is insufficient, the second physiological symptom associated with the first physiological symptom is determined, so that the prediction basis in the physiological label prediction is expanded, and the description range of the predicted target physiological label is expanded; the supplementary physiological symptoms are linked with the target physiological label through the symptom supplementary guide statement, and the target physiological label is utilized to pertinently guide the user to input the physiological symptoms, so that the first physiological symptom is supplemented, and the efficiency of obtaining the effective first physiological symptom is improved; under the condition of ensuring the accuracy of the physiological label prediction, the human-computer interaction efficiency of the physiological label prediction is improved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a block diagram of a computer system provided in an exemplary embodiment of the present application;
FIG. 2 is a schematic diagram of a predictive model using physiological signatures as provided by an exemplary embodiment of the present application;
FIG. 3 is a schematic diagram of a predictive model using physiological signatures as provided by an exemplary embodiment of the present application;
FIG. 4 is a flow chart of a physiological signature prediction method provided by an exemplary embodiment of the present application;
FIG. 5 is a flow chart of a physiological signature prediction method provided by an exemplary embodiment of the present application;
FIG. 6 is a flow chart of a physiological signature prediction method provided by an exemplary embodiment of the present application;
FIG. 7 is a flow chart of a method for physiological signature prediction as provided by an exemplary embodiment of the present application;
FIG. 8 is a flow chart of a physiological signature prediction method provided by an exemplary embodiment of the present application;
FIG. 9 is a flow chart of a method for physiological signature prediction as provided by an exemplary embodiment of the present application;
FIG. 10 is an interface diagram of physiological label prediction provided by an exemplary embodiment of the present application;
FIG. 11 is an interface diagram of physiological label prediction provided by an exemplary embodiment of the present application;
FIG. 12 is a flow chart of a method for physiological signature prediction as provided by an exemplary embodiment of the present application;
FIG. 13 is a flow chart of a physiological signature prediction method provided by an exemplary embodiment of the present application;
FIG. 14 is a flow chart of a method for physiological signature prediction as provided by an exemplary embodiment of the present application;
FIG. 15 is a schematic diagram of a trained physiological label prediction model provided by an exemplary embodiment of the present application;
FIG. 16 is a flow chart of a method for training a physiological label prediction model provided by an exemplary embodiment of the present application;
FIG. 17 is a schematic diagram of a training aided prediction network provided by an exemplary embodiment of the present application;
FIG. 18 is a block diagram of a physiological signature prediction device provided in an exemplary embodiment of the present application;
FIG. 19 is a block diagram of a training apparatus for a physiological label prediction model provided in an exemplary embodiment of the present application;
fig. 20 is a block diagram of a server according to an exemplary embodiment of the present application.
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the application and, together with the description, serve to explain the principles of the application.
Detailed Description
To make the objects, technical solutions and advantages of the present application more clear, embodiments of the present application will be described in further detail below with reference to the accompanying drawings.
Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. The following description refers to the accompanying drawings in which the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.
The terminology used in the present disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. As used in this disclosure and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.
It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) referred to in the present application are information and data authorized by the user or fully authorized by each party, and the collection, use and processing of the related data need to comply with the relevant laws and regulations and standards of the relevant countries and regions. For example, a first physiological symptom, a first physiological label, etc. referred to in this application are obtained with sufficient authorization. It will be understood that, although the terms first, second, etc. may be used herein to describe various information, these information should not be limited by these terms. These terms are only used to distinguish one type of information from another. For example, a first parameter may also be referred to as a second parameter, and similarly, a second parameter may also be referred to as a first parameter, without departing from the scope of the present disclosure. The word "if," as used herein, may be interpreted as "at … …" or "when … …" or "in response to a determination," depending on the context.
FIG. 1 illustrates a block diagram of a computer system provided in an exemplary embodiment of the present application. The computer system may implement a system architecture that becomes a training method for a physiological label prediction model and/or a physiological label prediction method. The computer system may include: a terminal 100 and a server 200.
The terminal 100 may be an electronic device such as a mobile phone, a tablet Computer, a car terminal (car machine), a wearable device, a PC (Personal Computer), and the like. The terminal 100 may have a client installed therein for running a target application, which may be a game application or another application providing a training function of a physiological label prediction model and/or a physiological label prediction function, and the present application is not limited thereto. The form of the target Application is not limited in the present Application, and includes, but is not limited to, an Application (App) installed in the terminal 100, an applet, and the like, and may be a web page form.
The server 200 may be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server providing cloud computing services. The server 200 may be a background server of the target application program, and is configured to provide a background service for a client of the target application program.
According to the training method and/or the physiological label prediction method for the physiological label prediction model provided by the embodiment of the application, the execution subject of each step can be computer equipment, and the computer equipment refers to electronic equipment with data calculation, processing and storage capabilities. Taking the embodiment environment shown in fig. 1 as an example, the terminal 100 may execute the training method and/or the physiological label prediction method of the physiological label prediction model (for example, a client installed and running in the terminal 100 of the target application program executes the training method and/or the physiological label prediction method of the physiological label prediction model), the server 200 may execute the training method and/or the physiological label prediction model, or the terminal 100 and the server 200 cooperate with each other to execute the training method and/or the physiological label prediction method, which is not limited in this application.
In addition, the technical scheme of the application can be combined with the block chain technology. For example, some data involved in the training method of the physiological label prediction model and/or the physiological label prediction method disclosed in the present application (such as data of physiological symptoms, physiological labels, etc.) may be saved on the blockchain. The terminal 100 and the server 200 may communicate with each other through a network, such as a wired or wireless network.
Next, the physiological label prediction model in the present application is introduced:
fig. 2 is a diagram illustrating a prediction model using physiological signatures according to an exemplary embodiment of the present application.
Acquiring a first physiological symptom 352, and calling an auxiliary prediction network 310a to perform prediction processing on the first physiological symptom 352; the auxiliary prediction network 310a includes a first encoder 310b and a first classifier 310 c; specifically, the first encoder 310b is invoked to encode the first physiological symptom 352, so as to obtain a first feature representation 352c corresponding to the first physiological symptom 352; calling a first classifier 310c to perform classification prediction on the first feature representation 352c to obtain an auxiliary physiological label 354 a;
determining a second physiological symptom 354 associated with the first physiological symptom 352 based on the symptom association information 342; reconstructing the first physiological symptom 352 and the second physiological symptom 354 according to the prompt message 344 to obtain reconstruction information 354 b;
calling a physiological label prediction model 330a to perform prediction processing on the first physiological symptom 352 and the second physiological symptom 354; the physiological label prediction model 330a includes a second encoder 330b and a second classifier 330 c; specifically, the second encoder 330b is invoked to encode the reconstruction information 354b, so as to obtain a second feature representation 354c corresponding to the reconstruction information 354 b; the second classifier 330c is invoked to perform classification prediction on the second feature representation 354c, so as to obtain a target physiological label 356a and a prediction probability 356b of the target physiological label.
In the event that the predicted probability 356b of the target physiological label is greater than the target threshold, outputting a first physiological symptom 352, a second physiological symptom 354, and a target physiological label 356 a; in case the predicted probability 356b of the target physiological label is not greater than the target threshold, a symptom supplement guide sentence 346 is determined according to the target physiological label 356a, wherein the symptom supplement guide sentence 346 includes a guide question 346a and a guide answer option 346 b.
Fig. 3 illustrates a schematic diagram of a predictive model using physiological signatures provided by an exemplary embodiment of the present application.
The acquired first physiological symptoms 352 include physiological symptom a and physiological symptom b;
calling an auxiliary prediction network 310a to perform prediction processing on the first physiological symptom 352 to obtain an auxiliary physiological label 354 a; the auxiliary physiological label 354a includes a physiological label a, a physiological label B, and a physiological label C;
symptom associated information 342 includes first information, second information, and third information;
the first information is used for describing the association relationship between different physiological symptoms belonging to the physiological label A and the physiological label A; the second information is used for describing the association relationship between different physiological symptoms belonging to the physiological label B and the physiological label B; the third information is used for describing the association relationship between different physiological symptoms belonging to the physiological label C and the physiological label C.
Determining a second physiological symptom 354 associated with the first physiological symptom 352 based on the first information, the second information, and the third information; the second physiological condition 354 includes physiological condition c, physiological condition d, and physiological condition e; it is noted that the second physiological condition 354 is generally different from the first physiological condition 352, but the same is not excluded.
Calling a physiological label prediction model 330a to perform prediction processing on the first physiological symptom 352 and the second physiological symptom 354 to obtain a target physiological label 356a and a prediction probability 356b of the target physiological label;
target physiological label 356a includes physiological label D, physiological label E, physiological label F, and physiological label G; the predicted probabilities 356b of the target physiological label include a probability D, a probability E, a probability F, and a probability G.
It should be noted that no limitation is imposed on the relationship between the target physiologic tag 356a and the auxiliary physiologic tag 354a in this application, i.e., the target physiologic tag 356a and the auxiliary physiologic tag 354a can be identical or different.
For example, the physiological symptoms corresponding to the physiological label D include physiological symptoms a to e;
the physiological symptoms corresponding to the physiological label E comprise physiological symptoms a to E and physiological symptoms x;
the physiological symptoms corresponding to the physiological label F comprise physiological symptoms a to e and physiological symptoms y;
the physiological symptoms corresponding to the physiological label G comprise physiological symptoms a to e and physiological symptoms z;
wherein the physiological symptoms a to e include: physiological symptom a, physiological symptom b, physiological symptom c, physiological symptom d, and physiological symptom e; the physiological symptoms a to e correspond to a plurality of physiological labels in the target physiological label 356a, and the physiological symptoms a to e are common physiological symptoms.
The physiological symptom x, the physiological symptom y, and the physiological symptom z correspond to one of the target physiological tags 356a, and the physiological symptom x, the physiological symptom y, and the physiological symptom z are unique physiological symptoms.
In the event that the predicted probability 356b of the target physiological label is not greater than the target threshold, determining a symptom supplementing lead statement 346 from the target physiological label 356 a; the symptom supplement guide sentence 346 includes a guide question sentence 346a and a guide answer choice 346 b. Illustratively, the guidance question 346a is predefined, such as: guide question 346a includes: please choose which physiological symptoms the following occur? Or, please choose which physiological symptoms do not appear?
The leading answer option 346b includes unique physiological symptoms, i.e., including physiological symptom x, physiological symptom y, and physiological symptom z.
Illustratively, the first physiological symptom 352 is supplemented by selecting one or more of the guide answer choices 346b or not selecting all of the guide answer choices 346b, as directed by the guide question 346 a.
It should be noted that the present embodiment is implemented by determining a second physiological symptom 354 associated with the first physiological symptom 352; the basis of the physiological label prediction model 330a in prediction processing is expanded; namely: the physiological symptoms a and b included by the first physiological symptom 352; the first and second physiological symptoms 352 and 354 include physiological symptoms a to e. The description range of the target physiological label 356a predicted by the physiological label prediction model 330a is expanded.
By constructing the guidance answer option 346b by using the unique physiological symptom in the physiological symptoms corresponding to the target physiological label 356a, the efficiency of obtaining the effective first physiological symptom 352 is improved, and the first physiological symptom 352 is supplemented; the user can be guided to supplement the physiological symptoms which are helpful for the physiological label prediction in a targeted manner. Under the condition of ensuring the accuracy of the physiological label prediction, the human-computer interaction efficiency of the physiological label prediction is improved.
Next, the physiological label prediction method will be described by the following embodiments.
Fig. 4 shows a flowchart of a physiological signature prediction method provided by an exemplary embodiment of the present application. The method may be performed by a computer device. The method comprises the following steps:
step 510: acquiring an input first physiological symptom;
the first physiological symptom is used for describing a physiological abnormal state of the target object; illustratively, the first physiological symptom includes one or more abnormal representations of the first sample subject being in a physiologically abnormal state; further, the physiological abnormal condition may be described from multiple dimensions by at least one of duration, severity, and characterization properties;
those skilled in the art will appreciate that in one implementation in the present application, the acquisition of the first physiological symptom and other data is subject to individual consent by authorized subjects to comply with relevant laws and regulations and standards in the relevant country and region.
Step 520: determining a second physiological symptom associated with the first physiological symptom;
the second physiological condition includes at least one physiological condition having an associative relationship with the first physiological condition.
Optionally, the second physiological condition belongs to at least one of candidate physiological conditions, the candidate physiological condition being a known at least one physiological condition; illustratively, the known at least one physiological symptom is obtained from a pool of physiological symptoms.
Illustratively, there is an associative relationship between the first physiological symptom and the second physiological symptom; the second physiological symptom can be determined in a statistical processing mode, and can also be predicted and determined through a neural network model; the manner in which the second physiological condition is determined is not subject to any limiting regulation by the present application.
Step 530: predicting the first physiological symptom and the second physiological symptom to obtain a target physiological label and a prediction probability of the target physiological label;
illustratively, the target physiological label is used to describe the type of physiological abnormality of the target subject; for example, the target physiological label is used to describe a disease of the target subject. In one implementation, the target physiological label describes a physiological abnormality site and a physiological abnormality cause of the target subject; for example, there may be no semantic association between different physiological tags, and there may also be a semantic association; that is, the physiological tags can be discretized or textual.
Illustratively, the predicted probability of the target physiological label is used to describe the confidence of the target physiological label; such as: the prediction probability of the target physiological label can take any value from 0 to 1.
Illustratively, the first physiological symptom and the second physiological symptom are subjected to prediction processing to obtain a target physiological label and a prediction probability of the target physiological label, which can be obtained by calling a physiological label prediction model for prediction or can be obtained by statistical processing calculation;
in one example, the computer device invokes a physiological label prediction model, the physiological label prediction model including an Artificial Neural Network (ANN);
step 540: under the condition that the prediction probability of the target physiological label is greater than a target threshold value, outputting the target physiological label;
illustratively, the prediction probability of the target physiological label is greater than the target threshold for indicating that the target physiological label satisfies the confidence condition, and the target physiological label is used as the physiological label prediction result of the first physiological symptom.
For example, the target threshold may be predefined, or may be determined according to the input first physiological symptom, and the present embodiment does not provide any limiting provisions for the determination manner of the target threshold; in one implementation, the target threshold is 0.8.
Step 550: determining a symptom supplement guide statement according to the target physiological label under the condition that the prediction probability of the target physiological label is not greater than the target threshold;
illustratively, the prediction probability of the target physiological label is not greater than the target threshold for indicating that the target physiological label does not satisfy the confidence condition, i.e., the target physiological label is not a physiological label prediction result for the first physiological symptom.
In the case where the predicted probability of the target physiological signature is not greater than the target threshold, the first physiological symptom needs to be supplemented. The symptom supplement guide sentence is used for guiding supplement of the first physiological symptom.
For example, the symptom supplementation guidance statement may directly provide a physiological symptom option for supplementing the first physiological symptom according to the target physiological label, or indirectly provide a supplementation prompt for supplementing the first physiological symptom according to the physiological abnormal part and the physiological abnormal reason of the target object described in the target physiological label.
In summary, in the method provided in this embodiment, when the first physiological symptom of the input target object is insufficient, the prediction basis in the prediction of the physiological label is expanded by determining the second physiological symptom associated with the first physiological symptom, and the description range of the predicted target physiological label is expanded; the supplementary physiological symptoms are associated with the target physiological label through the symptom supplementary guidance statement, and the target physiological label is utilized to pertinently guide the user to input the physiological symptoms, so that the first physiological symptom is supplemented, and the efficiency of obtaining the effective first physiological symptom is improved; under the condition of ensuring the accuracy of the physiological label prediction, the human-computer interaction efficiency of the physiological label prediction is improved.
Fig. 5 shows a flowchart of a physiological signature prediction method provided by an exemplary embodiment of the present application. The method may be performed by a computer device. That is, in an alternative design, in the embodiment shown in fig. 4, step 520 may be implemented as step 522:
step 522: determining a second physiological symptom associated with the first physiological symptom based on the symptom-associated information;
in this embodiment, the second physiological condition does not generally include the first physiological condition, but does not exclude the inclusion of the first physiological condition;
illustratively, symptom associated information is used to indicate the degree of correlation between different physiological symptoms. For example, the first physiological condition includes physiological condition a; in a case where the symptom correlation information indicates that there is a correlation between the physiological symptom a and the physiological symptom B, determining that the second physiological symptom includes the physiological symptom B; in a case where the symptom correlation information indicates that there is no correlation between the physiological symptom a and the physiological symptom C, determining that the second physiological symptom does not include the physiological symptom C;
it should be noted that the symptom association information may include association relationships between different physiological symptoms and physiological tags, and different physiological symptoms belong to the same physiological tag; associations between different physiological symptoms may also be included; namely, the symptom-related information may indirectly indicate the degree of correlation between different physiological symptoms by using the physiological label, or may directly indicate the degree of correlation between different physiological symptoms. For example, the symptom correlation information may indicate a degree of correlation between different physiological symptoms through at least one of a correlation coefficient, a correlation label, and a correlation word.
Taking the correlation coefficient as an example, the correlation coefficient includes Mutual Information of Points (PMI); further, taking the symptom-related information as an example of indirectly indicating the degree of correlation between different physiological symptoms by using the physiological label, the point-to-point information is as follows:
wherein, PMI (a,b) Representing mutual information of points, p a Representing the probability of occurrence of a physiological label, p b Representing the probability of occurrence of a first candidate physiological symptom, p (a,b) Representing the probability of the physiological label and the first candidate physiological symptom occurring simultaneously, and log representing a logarithmic operation; wherein the first candidate physiological symptom is any one of different physiological symptoms belonging to the same physiological label. As will be appreciated by those skilled in the art, the point-to-point information is used to describe the association between the first physiological symptom and a candidate physiological symptom, where the symptom association information directly indicates the degree of correlation between the different physiological symptoms, and the candidate physiological symptom is at least one of the different physiological symptoms.
Taking the relevance label as an example, the relevance label comprises a first label, a second label and a third label, and the incidence relations represented by the three labels are sequentially decreased; the correlation label may directly indicate the degree of correlation between different physiological symptoms, or may indirectly indicate the degree of correlation between different physiological symptoms.
Taking the relevant words as an example, the association relationship is described by one or more words, and the relevant words include at least one of the following: significant correlation, irrelevance, significant irrelevance; the correlation words can directly indicate the correlation degree between different physiological symptoms, and can also indirectly indicate the correlation degree between different physiological symptoms.
In an alternative implementation manner, the candidate physiological symptoms of which the symptom associated information meets the correlation condition are determined as the second physiological symptoms associated with the first physiological symptoms; illustratively, the candidate physiological condition is at least one of the different physiological conditions.
In summary, in the method provided in this embodiment, when the first physiological symptom of the input target object is insufficient, the second physiological symptom associated with the first physiological symptom is determined based on the symptom-associated information, so that the prediction basis in the prediction of the physiological label is expanded, and the description range of the predicted target physiological label is expanded; under the condition of ensuring the accuracy of the physiological label prediction, the human-computer interaction efficiency of the physiological label prediction is improved.
For step 522, there are at least two implementations as follows:
the implementation mode is as follows: the correlation degree between different physiological symptoms can be indirectly indicated by using the physiological label corresponding to the symptom correlation information;
fig. 6 shows a flowchart of a physiological signature prediction method provided by an exemplary embodiment of the present application. The method may be performed by a computer device. That is, in an alternative design, in the embodiment shown in fig. 5, step 522 may be implemented as steps 522a and 522 b:
step 522 a: performing prediction processing on the first physiological symptom to obtain an auxiliary physiological label;
illustratively, the prediction processing is performed on the first physiological symptom to obtain the auxiliary physiological label, which may be obtained by calling an auxiliary prediction network for prediction, or may be obtained by calculating through statistical processing;
in one example, a computer device invokes an auxiliary predictive Network, the auxiliary predictive Network including an Artificial Neural Network (ANN); the auxiliary prediction network is used for predicting the first physiological symptom to obtain an auxiliary physiological label;
it should be noted that, in an alternative design of the present application, the auxiliary prediction network and the physiological label prediction model have the same network architecture; those skilled in the art will appreciate that the auxiliary prediction network and the physiological label prediction model are not trained simultaneously, and the auxiliary physiological label and the physiological label prediction model obtained after training usually have different network parameters, but the situation that the auxiliary physiological label and the physiological label prediction model can have the same network parameters is not excluded.
The secondary physiological label is a predictive outcome of the first physiological symptom.
Optionally, in an optional implementation, an auxiliary prediction network is invoked to perform prediction processing on the first physiological symptom to obtain an auxiliary physiological label;
further alternatively, step 522a may be implemented as two substeps as follows:
substep 1: coding the first physiological symptom to obtain a first characteristic representation of the first physiological symptom;
illustratively, the first feature representation of the first physiological symptom may be implemented as at least one of a feature value, a feature vector, a feature matrix; the present embodiment does not set any limit thereto;
similarly, the first physiological symptom is coded to obtain a first characteristic representation of the first physiological symptom, which can be obtained by calling an auxiliary prediction network for prediction or by calculating through statistical processing;
optionally, the auxiliary prediction network comprises a first encoder and a first classifier; illustratively, the first encoder is for encoding a natural language into the feature representation; that is, a first physiological symptom is encoded as a sample feature representation; the first encoder includes, but is not limited to, at least one of the following network structures: convolutional Neural Networks (CNN), embedded Language Model structures (ELMo), Bidirectional and autoregressive translation Model structures (BART), Generative Pre-Training Model structures (GPT), Enhanced Representation structures (ERNIE) from kNowledge IntEgration, feed-Forward Neural Networks (FNN), and Long-Term Short-Term memory Networks (LSTM).
And substep 2: carrying out classification prediction on the first feature representation to obtain an auxiliary physiological label;
similarly, the first feature representation is classified and predicted to obtain an auxiliary physiological label, which can be obtained by calling an auxiliary prediction network for prediction or by statistical processing calculation;
optionally, the auxiliary prediction network comprises a first encoder and a first classifier;
illustratively, the first classifier is for classifying the feature representation; namely, classifying the first feature representation to obtain an auxiliary physiological label; the first classifier includes, but is not limited to, at least one of the following network structures: support Vector Machines (SVMs), normalized index classifiers (Softmax classifiers).
Step 522 b: determining a second physiological symptom associated with the first physiological symptom based on the association relationship between the different physiological symptoms belonging to the auxiliary physiological label and the auxiliary physiological label;
for example, the symptom correlation information may indirectly indicate the degree of correlation between different physiological symptoms by using the physiological label, that is, the symptom correlation information includes correlation between different physiological symptoms and physiological labels belonging to the same physiological label; the symptom association information determines a second physiological symptom by indicating an association between the secondary physiological marker and the secondary physiological label.
In summary, in the method provided in this embodiment, under the condition that the input first physiological symptom of the target object is insufficient, the second physiological symptom associated with the first physiological symptom is determined based on the symptom-associated information by predicting the auxiliary physiological label, so that the association relationship between the second physiological symptom and the first physiological symptom is ensured; the prediction basis in the process of predicting the physiological label is expanded, and the description range of the predicted target physiological label is expanded; under the condition of ensuring the accuracy of the physiological label prediction, the human-computer interaction efficiency of the physiological label prediction is improved.
The implementation mode two is as follows: the corresponding symptom associated information can directly indicate the degree of correlation between different physiological symptoms;
fig. 7 shows a flowchart of a physiological signature prediction method provided by an exemplary embodiment of the present application. The method may be performed by a computer device. That is, in an alternative design, in the embodiment shown in fig. 5, step 522 may be implemented as step 522 c:
step 522 c: determining a second physiological symptom associated with the first physiological symptom based on the association relationship between different physiological symptoms;
for example, the symptom associated information may directly indicate a degree of correlation between different physiological symptoms, i.e., the symptom associated information includes an association relationship between different physiological symptoms. The embodiment and the above related information indirectly indicate the correlation degree between different physiological symptoms by using the physiological label; in this embodiment, the first physiological symptom does not need to be predicted by invoking the auxiliary prediction network.
Regarding the construction of symptom associated information, in an alternative implementation, the physiological data is subjected to statistical processing to construct symptom associated information; illustratively, physiological data is typically data that includes physiological symptoms and physiological signatures.
The physiological data is usually a plurality of data, and the physiological symptoms and the occurrence probability of the corresponding physiological label can be obtained by performing statistical processing on the physiological data, and the symptom-related information can be constructed by the occurrence probability. Optionally, the physiological data further comprises: at least one of historical physiological symptoms, physiological test results, and physiological sensitivity types.
Optionally, the physiological data comprises physiological knowledge; the physiological knowledge includes all physiological symptoms corresponding to the same physiological label, but the occurrence probability of the physiological symptoms is not shown. Further optionally, the symptom associated information is checked through the physiological data to update the symptom associated information; by checking the symptom associated information, the symptom associated information is further ensured to correctly indicate the correlation degree between different physiological symptoms.
Illustratively, the symptom associated information is checked through physiological knowledge; the condition that typical physiological symptoms corresponding to recorded physiological labels easily appear in physiological data is fully considered; it is ensured that the symptom correlation information correctly indicates the degree of correlation between different physiological symptoms.
In another alternative implementation, the symptom associated information is predicted by a neural network model; illustratively, physiological symptoms in the physiological data are input into a neural network model to obtain symptom associated information; or, inputting the physiological symptoms and the physiological labels in the physiological data into the neural network model to obtain the symptom associated information.
Those skilled in the art will understand that the above-mentioned construction method of the symptom-related information is applicable to the case that the symptom-related information includes the association relationship between different physiological symptoms belonging to the same physiological label and the physiological label, and also applicable to the case that the symptom-related information includes the association relationship between different physiological symptoms.
In summary, in the method provided in this embodiment, when the first physiological symptom of the input target object is insufficient, the second physiological symptom associated with the first physiological symptom is determined according to the association relationship between different physiological symptoms included in the symptom association information, so that the association relationship between the second physiological symptom and the first physiological symptom is ensured; the prediction basis in the process of predicting the physiological label is expanded, and the description range of the predicted target physiological label is expanded; under the condition of ensuring the accuracy of the physiological label prediction, the human-computer interaction efficiency of the physiological label prediction is improved.
Fig. 8 shows a flowchart of a physiological signature prediction method provided by an exemplary embodiment of the present application. The method may be performed by a computer device. That is, in an alternative design, in the embodiment shown in fig. 4, step 550 may be implemented as step 552:
step 552: under the condition that the prediction probability of the target physiological label is not greater than the target threshold value, generating a symptom supplement guide statement according to a third physiological symptom corresponding to the target physiological label;
illustratively, the symptom supplementation guide statement directly provides a physiological symptom option for supplementing the first physiological symptom according to the target physiological label. Specifically, a symptom supplement guide sentence is generated according to the third physiological symptom; because the third physiological symptom is a physiological symptom corresponding to the target physiological label, a connection is established between the symptom supplement guide statement and the target physiological label;
the third physiological symptom corresponding to the target physiological label may be determined through statistical processing, or may be determined through prediction of a neural network model.
Optionally, as shown in fig. 9, step 552 in this embodiment may be implemented as step 552 a:
step 552 a: determining a guiding answer option according to a third physiological symptom under the condition that the prediction probability of the target physiological label is not greater than a target threshold value;
illustratively, the symptom supplement guide statement comprises a guide question and a guide answer, wherein the guide answer is used for providing supplement content of the first physiological symptom, and the guide question is used for prompting selection of the guide answer; the guide question is predefined.
In this embodiment, the third physiological symptom corresponds to the target physiological label, and the third physiological symptom includes one or more physiological symptoms; the guide answer option comprises all symptoms or part of symptoms in the third physiological symptom; for example, the introductory question includes a request to select which physiological symptom is present? Or, please choose which physiological symptoms do not appear?
Illustratively, the first physiological symptom 352 is supplemented by performing selection on one or more of the guide answer choices or performing no selection on all of the guide answer choices according to the guidance of the guide question sentence.
In an alternative embodiment, the guidance question is determined based on the predicted probability of the physiological label.
For example, in the case that the maximum value of the difference between the prediction probabilities included in the prediction probabilities of the target physiological label is greater than the difference threshold, the guidance question is used to prompt the physiological symptom that does not appear in the guidance answer option selection, such as: the introductory question includes a request to select which of the following physiological symptoms did not occur? For example, in the case that the maximum value of the difference between the prediction probabilities included in the prediction probabilities of the target physiological label is not greater than the difference threshold, the guidance question is used to prompt that the physiological symptom that has appeared in the guidance answer option selection, such as: the introductory question includes a request to select which physiological symptoms are present?
It should be noted that, under the condition that the maximum value of the difference between the prediction probabilities is not greater than the difference threshold, the first physiological symptom is supplemented by selecting the physiological symptom that has already occurred, so that the prediction probability of at least one physiological label in the target physiological labels is improved, and the efficiency of obtaining the effective first physiological symptom is improved;
under the condition that the maximum value of the difference between the prediction probabilities is larger than the difference threshold, the first physiological symptom is supplemented by selecting the physiological symptom which does not appear, and the physiological label prediction is carried out through the supplemented first physiological symptom, so that the accuracy of the physiological label prediction is ensured.
Illustratively, FIG. 10 shows an interface diagram of physiological label prediction provided by an exemplary embodiment of the present application; in the interface diagram of the physiological label prediction, the target physiological label 704 is determined as the physiological label prediction result of the first physiological symptom 702, and the target physiological label 704 is output and displayed; in an alternative design, the interface diagram also shows a first physiological symptom 702;
for example, fig. 11 illustrates an interface diagram of physiological label prediction provided by an exemplary embodiment of the present application; in the interface graph predicted by the physiological label, determining a guide answer option 714b according to a third physiological symptom corresponding to the target physiological label; the guidance question 714a includes: please choose which physiological symptoms the following occur? In an alternative design, a first physiological symptom 712 is also displayed in the interface diagram.
In summary, in the method provided in this embodiment, under the condition that the first physiological symptom of the input target object is insufficient, the symptom supplementation guide statement is associated with the target physiological label through the third physiological symptom corresponding to the target physiological label, and the third physiological symptom is used to determine the symptom supplementation guide statement, so as to specifically guide the user to input the physiological symptom, thereby implementing supplementation of the first physiological symptom, and improving the efficiency of obtaining an effective first physiological symptom; under the condition of ensuring the accuracy of the physiological label prediction, the human-computer interaction efficiency of the physiological label prediction is improved.
Fig. 12 shows a flowchart of a physiological signature prediction method provided by an exemplary embodiment of the present application. The method may be performed by a computer device. That is, in an alternative design, in the embodiment shown in fig. 9, step 552a may be implemented as step 552 b:
step 552 b: determining a guidance answer option according to the unique physiological symptom under the condition that the prediction probability of the target physiological label is not greater than the target threshold value;
illustratively, the third physiological symptom corresponding to the target physiological signature includes a common physiological symptom corresponding to more than a physiological signatures in the target physiological signature and a unique physiological symptom corresponding to no more than a physiological signatures in the target physiological signature; a is a positive integer not exceeding the number of target physiological tags.
In one example, the target physiological label includes physiological label a, physiological label B, physiological label C, and physiological label D;
the physiological label A corresponds to a physiological symptom a, a physiological symptom b, a physiological symptom m and a physiological symptom x;
the physiological label B corresponds to a physiological symptom a, a physiological symptom B, a physiological symptom m and a physiological symptom y;
the physiological label C corresponds to a physiological symptom a, a physiological symptom b, a physiological symptom x and a physiological symptom y;
the physiological label D corresponds to a physiological symptom a, a physiological symptom b, a physiological symptom m and a physiological symptom z.
It can be understood that the physiological symptom a and the physiological symptom b both correspond to 4 physiological labels in the target physiological labels; the physiological symptom m corresponds to 3 physiological labels in the target physiological labels; the physiological symptoms x and y correspond to 2 physiological labels in the target physiological labels; the physiological symptom z corresponds to 1 of the target physiological labels.
Illustratively, the common physiological symptom corresponds to more than 2 of the target physiological tags, and the unique physiological symptom corresponds to no more than 2 of the target physiological tags; that is, common physiological symptoms include: physiological symptom a, physiological symptom b, and physiological symptom m; the unique physiological symptoms include: physiological symptom x, physiological symptom y, and physiological symptom z; the guide answer options are determined based on the unique physiological symptoms.
In summary, according to the method provided by this embodiment, when the input first physiological symptom of the target object is insufficient, the guidance answer option is constructed by using the unique physiological symptom in the third physiological symptom corresponding to the target physiological label, so that the efficiency of obtaining an effective first physiological symptom is improved, and the first physiological symptom is supplemented by associating the symptom supplement guidance statement with the target physiological label; the user can be guided in a targeted way to supplement the physiological symptoms which are helpful for the physiological label prediction. Under the condition of ensuring the accuracy of the physiological label prediction, the human-computer interaction efficiency of the physiological label prediction is improved.
Next, the reconstruction information will be described by an embodiment: fig. 13 shows a flowchart of a physiological signature prediction method provided by an exemplary embodiment of the present application. The method may be performed by a computer device. That is, in an alternative design, on the basis of the embodiment shown in fig. 4, step 525 is further included; step 530 may be implemented as step 532:
step 525: reconstructing the first physiological symptom and the second physiological symptom according to the prompt information to obtain reconstruction information;
illustratively, the prompt message includes a slot position and prompt content, the slot position is used for providing a filling position of the target physiological label, and the prompt content is used for prompting to fill the slot position; it should be noted that the presentation form of the prompt message may be text message, or may be represented by coded prompt features, such as at least one of a feature value, a feature vector, and a feature matrix; illustratively, the prompt information is constructed according to physiological knowledge, specifically, one or more prompt information pre-constructed according to physiological knowledge, or predicted prompt information encoded by an artificial neural network;
taking the example that the prompt information is text information, the prompt information at least includes at least one of the following:
f (x) is related to a typical physiological symptom as [ MASK ];
f (x) may be [ MASK ];
f (x) appears [ MASK ], denying [ MASK ];
f (x) not admitting having [ MASK ], possibly [ MASK ];
wherein, f (x) represents prompt information, and [ MASK ] represents a slot position; f, (x) the text information except the slot position is the prompt content.
In an alternative design, step 525 may be implemented as at least any one of the following two substeps:
substep 1: superposing the prompt information with the first physiological symptom and the second physiological symptom, and splicing to obtain reconstruction information;
illustratively, under the condition that the implementation mode of the prompt information is character information, the prompt information is superposed with the first physiological symptom and the second physiological symptom, and the superposed prompt information is spliced to obtain reconstructed information;
such as: the first and second physiological symptoms are: indirect pain in the left ear, no hearing loss; the prompt information is as follows: (x) appearance [ MASK ], deny [ MASK ]; the prompt information is superposed on the first physiological symptom and the second physiological symptom to obtain reconstruction information; the reconstruction information is: indirect pain in the left ear, no hearing loss, appearance [ MASK ], negative inhibition [ MASK ];
substep 2: coding the first physiological symptom and the second physiological symptom to obtain a second characteristic representation; calculating to obtain reconstruction information according to the second feature representation and the prompt information;
exemplarily, in the case that the implementation manner of the prompt information is prompt feature representation, first, the first physiological symptom and the second physiological symptom are encoded to obtain a corresponding second feature representation; the reconstruction information is a coded representation of the reconstruction characteristics; the reconstruction characteristic representation is calculated according to the association characteristic representation and the prompt information;
illustratively, the calculation method can be convolution operation through other matrix, vector or numerical operation, and can also be directly obtained through superposition; no restrictive provisions are made for this.
Step 532: calling a physiological label prediction model to carry out prediction processing on the reconstruction information to obtain a target physiological label and the prediction probability of the target physiological label;
in this embodiment, the physiological label prediction model is used for performing physiological label prediction according to the reconstruction information to obtain a target physiological label; wherein the target physiological label is used to fill the slot.
In summary, according to the method provided in this embodiment, the first physiological symptom and the second physiological symptom are reconstructed through the prompt information, so that the training process of the physiological label prediction model is adapted to the training process of the physiological label prediction model, the complexity of the physiological label prediction performed by the physiological label prediction model is reduced, the use efficiency of the physiological label prediction model for performing prediction by using the first physiological symptom and the second physiological symptom is improved, and the accuracy of the physiological label prediction is ensured.
Next, a physiological label prediction model will be introduced by one embodiment:
fig. 14 shows a flowchart of a method for predicting a physiological signature provided in an exemplary embodiment of the present application. The method may be performed by a computer device. That is, in an alternative design, in the embodiment shown in fig. 4, step 530 is implemented as steps 534 and 536:
step 534: coding the first physiological symptom and the second physiological symptom to obtain a second characteristic representation;
similarly, the first physiological symptom and the second physiological symptom are coded, and the coding process can be obtained by calling a physiological label prediction model for prediction or can be obtained by statistical processing calculation;
optionally, the physiological label prediction model comprises a second encoder and a second classifier;
in this embodiment, a case of calling a physiological label prediction model to perform prediction processing to obtain a target physiological label and a prediction probability of the target physiological label is taken as an example for explanation; the physiological label prediction model comprises a second encoder and a second classifier; illustratively, the second feature representation of the first physiological symptom and the second physiological symptom may be implemented as at least one of a feature value, a feature vector, a feature matrix; the present embodiment does not set any limit thereto;
illustratively, the second encoder is for encoding the natural language into the feature representation; that is, the first physiological symptom and the second physiological symptom are encoded as a second characteristic representation; the second encoder includes, but is not limited to, at least one of the following network structures: CNN, ELMo, BART, GPT, ERNIE.
Step 536: classifying and predicting the second feature representation to obtain a target physiological label and the prediction probability of the target physiological label;
similarly, the classification prediction of the second feature representation can be obtained by calling a physiological label prediction model for prediction, or can be obtained by statistical processing calculation;
optionally, the physiological label prediction model comprises a second encoder and a second classifier; illustratively, the second classifier is for classifying the feature representation; namely, classifying the second feature representation to obtain a target physiological label and a prediction probability of the target physiological label; the second classifier includes, but is not limited to, at least one of the following network structures: SVM, Softmax classifier.
It should be noted that, in the case of invoking the auxiliary prediction network, the second encoder and the second classifier in the physiological label prediction model are independent from the auxiliary prediction network, that is, the physiological label prediction model and the auxiliary prediction network may have the same network structure; i.e. each comprising an encoder and a classifier, the training of the physiological label prediction model and the auxiliary prediction network is not performed simultaneously.
In summary, in the method provided in this embodiment, the physiological label prediction model is implemented as the second encoder and the second classifier, so that the accuracy of the physiological label prediction model in predicting the first physiological symptom and the second physiological symptom is ensured; by determining the second physiological symptom associated with the first physiological symptom, the prediction basis in the physiological label prediction is expanded, and the accuracy of the physiological label prediction is ensured.
Next, a training method of the physiological label prediction model will be described by the following embodiments.
FIG. 15 illustrates a schematic diagram of training a physiological label prediction model provided by an exemplary embodiment of the present application.
Obtaining a sample physiological symptom 332 and a sample physiological label 332 a; calling an auxiliary prediction network 310a to perform prediction processing on the sample physiological symptoms 332 to obtain auxiliary sample physiological labels 334 a;
determining a supplemental physiological symptom 334 associated with the sample physiological symptom 332 based on the association between the different physiological symptoms belonging to the auxiliary sample physiological label 310a and the auxiliary physiological label 334a, i.e. based on the symptom association information 342;
reconstructing the sample physiological symptoms 332 and the complementary physiological symptoms 334 according to the prompt information 344 to obtain reconstructed sample information 334 b;
calling an initial physiological label prediction model 330 to perform prediction processing on the reconstructed sample information 334b to obtain a predicted physiological label 336a and a predicted probability 336b of the predicted physiological label; based on the prediction error 338 between the predicted physiological label 336a and the sample physiological label 332a, the initial physiological label prediction model 330 is trained backward to obtain the physiological label prediction model 330 a.
Fig. 16 shows a flowchart of a training method of a physiological label prediction model provided by an exemplary embodiment of the present application. The method may be performed by a computer device. The method comprises the following steps:
step 610: obtaining a sample physiological symptom and a sample physiological label;
the sample physiological symptoms are used for describing physiological abnormal states of the sample objects; illustratively, the sample physiological condition includes one or more abnormal characterizations that the sample subject is in a physiologically abnormal state; further, the physiological abnormal condition may be described from multiple dimensions by at least one of duration, severity, and characterization properties; such as:
duration of time; features for describing physiological symptoms in the time dimension, such as: the treatment lasts for three days, appears intermittently within two days and appears sporadically within one week.
Degree of violence; an abnormality degree for describing a physiological abnormal state; such as: slight discomfort and severe pain.
Characterizing the property; a property that describes a physiological abnormal condition; such as: stabbing pain, laceration pain, distending pain.
The sample physiological label is used for describing the physiological abnormality type of the sample object; for example, the target physiological label is used to describe a disease of the target subject.
Step 620: determining a supplemental physiological symptom associated with the sample physiological symptom;
the supplemental physiological symptom includes at least one of the candidate physiological symptoms having an association relationship with the sample physiological symptom.
Illustratively, there is an association between the sample physiological condition and the supplemental physiological condition; the supplementary physiological symptoms can be determined in a statistical processing mode, and can also be predicted and determined through a neural network model;
step 630: calling an initial physiological label prediction model to perform prediction processing on the sample physiological symptoms and the supplementary physiological symptoms to obtain a prediction physiological label and a prediction probability of the prediction physiological label;
illustratively, the predicted physiological label is a prediction result of the initial physiological label prediction model for predicting the sample physiological symptoms and the complementary physiological symptoms, and the predicted physiological label is used for predicting the physiological abnormality type of the sample object; for example, the predictive physiological signature is used to describe a disease of the sample object.
Illustratively, the prediction probability of the predicted physiological signature is used to describe the confidence of the predicted physiological signature; such as: the prediction probability of the predicted physiological label can take any value from 0 to 1.
Step 640: based on a prediction error between the predicted physiological label and the sample physiological label, carrying out backward propagation training on the initial physiological label prediction model to obtain a physiological label prediction model;
illustratively, the prediction error includes, but is not limited to, at least one of: cross Entropy Loss function (Cross-Encopy Loss), 0-1 Loss function (Zero-One Loss), and Dis Loss function (Dice Loss).
Optionally, the prediction error is determined based on the predicted physiological signature, the prediction probability of the predicted physiological signature, and the sample physiological signature.
Illustratively, the prediction error comprises a cross-entropy loss error, the loss error describing a difference between the predicted physiological signature and the sample physiological signature; specifically, the prediction error is:
wherein Loss denotes the prediction error, y ik The sample physiological label is used for indicating that the ith sample physiological symptom corresponds to the kth sample physiological label; p is a radical of ik Representing a prediction probability of the predicted physiological label, wherein the prediction probability of the predicted physiological label represents a prediction probability that the ith sample physiological symptom corresponds to the kth sample physiological label; log represents a logarithmic operation, i represents the ith sample physiological symptom, and N represents the number of sample physiological symptoms; k denotes the kth sample physiological label and M denotes the number of sample physiological labels.
The objective of the backward error propagation training of the initial physiological label prediction model is to minimize the first predicted physiological label and the first prediction error between the first predicted physiological labels. For example, the physiological label prediction model trained by the present embodiment may be applied to any embodiment of the above physiological label prediction method.
In summary, in the method provided in this embodiment, when the sample physiological symptoms of the input sample object are insufficient, the supplementary physiological symptoms associated with the sample physiological symptoms are determined, so that the prediction basis in the prediction of the physiological label is expanded, and the description range of the predicted physiological label obtained by prediction is expanded; the information quantity of physiological symptoms used in the process of training the initial physiological label prediction model is improved, the dimensionality for describing the physiological symptoms of the sample object is expanded, and the training effect of the initial physiological label prediction model is effectively guaranteed.
Fig. 17 is a diagram illustrating a training aided prediction network provided by an exemplary embodiment of the present application.
Obtaining a sample physiological symptom 312 and a sample physiological label 312 a; calling an initial auxiliary prediction network 310 to perform prediction processing on the sample physiological symptoms 312 to obtain a predicted physiological label 314 a; the initial auxiliary prediction network 310 is back-propagated trained based on the prediction error 316 between the predicted physiological label 314a and the sample physiological label 312a, resulting in an auxiliary prediction network 310 a.
It should be noted that the auxiliary prediction network obtained by training in this embodiment may be applied to any embodiment of the above physiological label prediction method in the case that the auxiliary prediction network is called in the embodiment of the physiological label prediction method.
It should be further noted that, in the present embodiment, the sample physiological symptoms 312 and the sample physiological labels 312 a; the sample physiological symptoms 332 and sample physiological labels 332a obtained in the embodiment shown in fig. 15 may be the same or different; those skilled in the art will appreciate that the training of the auxiliary prediction network is independent of the training of the physiological label prediction model; taking the case of obtaining the same sample physiological symptoms and sample physiological labels as an example:
the auxiliary prediction network directly obtains a predicted physiological label through sample physiological symptom prediction;
the physiological label prediction model obtains a predicted physiological label through predicting reconstructed sample information, wherein the reconstructed sample information is constructed according to sample physiological symptoms, supplementary physiological symptoms and prompt information;
even in the case where the sample physiological symptoms and sample physiological signatures are the same; the auxiliary prediction network and the physiological label prediction model have different prediction inputs; the acquired sample physiological symptoms and sample physiological signatures typically have different network parameters.
Those skilled in the art will understand that the above embodiments can be implemented independently, or the above embodiments can be freely combined to combine a new embodiment to implement the training method and/or the physiological label prediction method of the physiological label prediction model of the present application.
Fig. 18 shows a block diagram of a physiological signature prediction apparatus provided in an exemplary embodiment of the present application. The device includes:
an obtaining module 810, configured to obtain an input first physiological symptom, where the first physiological symptom is used to describe a physiological abnormal state of a target subject;
a determining module 820 for determining a second physiological symptom associated with the first physiological symptom, wherein the second physiological symptom comprises at least one physiological symptom having an association relation with the first physiological symptom;
a prediction module 830, configured to perform prediction processing on the first physiological symptom and the second physiological symptom to obtain a target physiological label and a prediction probability of the target physiological label, where the target physiological label is used to describe a physiological abnormality type of the target object;
an output module 840, configured to output the target physiological label if the predicted probability of the target physiological label is greater than a target threshold;
the output module 840 is further configured to determine, according to the target physiological label, a symptom supplement guidance statement for guiding to supplement the first physiological symptom when the prediction probability of the target physiological label is not greater than the target threshold.
In an alternative design of the present application, the determining module 820 is further configured to:
determining a second physiological symptom associated with the first physiological symptom based on symptom associated information;
wherein the symptom associated information is used for indicating the degree of correlation between different physiological symptoms.
In an optional design of the present application, the symptom association information includes association relationships between different physiological symptoms belonging to the same physiological label and the physiological label;
the determining module 820 is further configured to:
performing prediction processing on the first physiological symptom to obtain an auxiliary physiological label;
determining a second physiological symptom associated with the first physiological symptom based on an association between a different physiological symptom belonging to the secondary physiological label and the secondary physiological label.
In an alternative design of the present application, the determining module 820 is further configured to:
coding the first physiological symptom to obtain a first characteristic representation of the first physiological symptom;
and performing classification prediction on the first feature representation to obtain the auxiliary physiological label.
In an optional design of the present application, the symptom correlation information includes a correlation between different physiological symptoms;
the determining module 820 is further configured to:
determining a second physiological symptom associated with the first physiological symptom based on an association relationship between different physiological symptoms.
In an alternative design of the present application, the output module 840 is further configured to: and generating the symptom supplementing guide statement according to a third physiological symptom corresponding to the target physiological label.
In an optional design of the present application, the symptom supplement guide sentence includes a guide question sentence and a guide answer option, the guide answer option is used for providing supplement content of the first physiological symptom, the guide question sentence is used for prompting selection of the guide answer option, and the guide question sentence is predefined;
the output module 840 is further configured to: determining the guide answer option according to the third physiological symptom.
In an alternative design of the present application, the third physiological symptom includes a common physiological symptom and a unique physiological symptom, the common physiological symptom corresponds to more than a physiological tags of the target physiological tags, and the unique physiological symptom corresponds to no more than a physiological tags of the target physiological tags; wherein a is a positive integer not exceeding the number of target physiological labels;
the output module 840 is further configured to: determining the leading answer option based on the unique physiological symptom.
In an alternative design of the present application, the apparatus further includes:
the constructing module 850 is configured to reconstruct the first physiological symptom and the second physiological symptom according to the prompt information, so as to obtain reconstruction information; the prompt information comprises a slot position and prompt content, the slot position is used for providing a filling position of the target physiological label, and the prompt content is used for prompting to fill the slot position;
the prediction module 830 is further configured to: and performing prediction processing on the reconstruction information to obtain a target physiological label and a prediction probability of the target physiological label, wherein the target physiological label is used for filling the slot position.
In an alternative design of the present application, the building module 850 is further configured to:
superposing the prompt information with the first physiological symptom and the second physiological symptom, and splicing to obtain the reconstruction information;
or, encoding the first physiological symptom and the second physiological symptom to obtain a second characteristic representation;
and calculating to obtain the reconstruction information according to the second feature representation and the prompt information.
In an alternative design of the present application, the prediction module 830 is further configured to:
coding the first physiological symptom and the second physiological symptom to obtain a second characteristic representation;
and performing classification prediction on the second feature representation to obtain the target physiological label and the prediction probability of the target physiological label.
Fig. 19 is a block diagram of a training apparatus for a physiological label prediction model according to an exemplary embodiment of the present application. The device includes:
an obtaining module 860, configured to obtain a sample physiological symptom and a sample physiological label, where the sample physiological symptom is used to describe a physiological abnormal state of a sample object, and the sample physiological label is used to describe a physiological abnormal type of the sample object;
a determination module 870 for determining a supplemental physiological symptom associated with the sample physiological symptom, the supplemental physiological symptom including at least one physiological symptom having an association relationship with the sample physiological symptom;
a prediction module 880, configured to invoke an initial physiological label prediction model to perform prediction processing on the sample physiological symptoms and the supplemental physiological symptoms, so as to obtain a predicted physiological label and a prediction probability of the predicted physiological label;
a training module 890, configured to perform backward propagation training on the initial physiological label prediction model based on the prediction error between the predicted physiological label and the sample physiological label, so as to obtain the physiological label prediction model.
It should be noted that, when the apparatus provided in the foregoing embodiment implements the functions thereof, only the division of the above functional modules is illustrated, and in practical applications, the above functions may be distributed by different functional modules according to actual needs, that is, the content structure of the device is divided into different functional modules, so as to complete all or part of the functions described above.
With regard to the apparatus in the above-described embodiment, the specific manner in which the respective modules perform operations has been described in detail in the embodiment related to the method; the technical effects achieved by the operations performed by the respective modules are the same as those in the embodiments related to the method, and will not be described in detail here.
An embodiment of the present application further provides a computer device, where the computer device includes: a processor and a memory, the memory having stored therein a computer program; the processor is configured to execute the computer program in the memory to implement the physiological label prediction method and/or the training method of the physiological label prediction model provided in the above method embodiments.
Optionally, the computer device is a server. Illustratively, fig. 20 is a block diagram of a server according to an exemplary embodiment of the present application. In general, the server 900 includes: a processor 901 and a memory 902.
In some embodiments, the server 900 may further optionally include: an input interface 903 and an output interface 904. The processor 901, the memory 902, and the input interface 903 and the output interface 904 may be connected by buses or signal lines. Each peripheral device may be connected to the input interface 903 and the output interface 904 via a bus, a signal line, or a circuit board. The Input interface 903 and the Output interface 904 can be used to connect at least one peripheral device related to Input/Output (I/O) to the processor 901 and the memory 902. In some embodiments, the processor 901, the memory 902 and the input interface 903, the output interface 904 are integrated on the same chip or circuit board; in some other embodiments, the processor 901, the memory 902, and any one or both of the input interface 903 and the output interface 904 may be implemented on a single chip or circuit board, which is not limited in this application.
Those skilled in the art will appreciate that the above-described illustrative architecture is not intended to be limiting of the server 900, and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components may be used.
In an exemplary embodiment, there is also provided a chip comprising programmable logic circuits and/or program instructions for implementing the physiological signature prediction method and/or the training method of the physiological signature prediction model according to the above aspects when the chip is run on a computer device.
In an exemplary embodiment, a computer program product is also provided that includes computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor reads the computer instructions from the computer-readable storage medium and executes the computer instructions to implement the physiological label prediction method and/or the training method of the physiological label prediction model provided by the above method embodiments.
In an exemplary embodiment, a computer-readable storage medium is further provided, in which a computer program is stored, the computer program being loaded and executed by a processor to implement the physiological label prediction method and/or the training method of the physiological label prediction model provided in the above method embodiments.
It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program instructing relevant hardware, where the program may be stored in a computer-readable storage medium, and the storage medium may be a read-only memory, a magnetic disk or an optical disk. Those skilled in the art will recognize that, in one or more of the examples described above, the functions described in the embodiments of the present application may be implemented in hardware, software, firmware, or any combination thereof. When implemented in software, the functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A storage media may be any available media that can be accessed by a general purpose or special purpose computer.
The above description is only exemplary of the present application and should not be taken as limiting, as any modification, equivalent replacement, or improvement made within the spirit and principle of the present application should be included in the protection scope of the present application.
Claims (15)
1. A method of physiological signature prediction, the method comprising:
acquiring an input first physiological symptom, wherein the first physiological symptom is used for describing a physiological abnormal state of a target object;
determining a second physiological symptom associated with the first physiological symptom, the second physiological symptom including at least one physiological symptom having an association relationship with the first physiological symptom;
predicting the first physiological symptom and the second physiological symptom to obtain a target physiological label and a prediction probability of the target physiological label, wherein the target physiological label is used for describing a physiological abnormity type of the target object;
outputting the target physiological label under the condition that the prediction probability of the target physiological label is greater than a target threshold value;
and under the condition that the prediction probability of the target physiological label is not greater than the target threshold, determining a symptom supplement guide statement according to the target physiological label, wherein the symptom supplement guide statement is used for guiding supplement of the first physiological symptom.
2. The method of claim 1, wherein determining a second physiological condition associated with the first physiological condition comprises:
determining a second physiological symptom associated with the first physiological symptom based on symptom associated information;
wherein the symptom associated information is used for indicating the degree of correlation between different physiological symptoms.
3. The method according to claim 2, wherein the symptom association information includes association relationships between different physiological symptoms belonging to the same physiological label and the physiological label;
the determining a second physiological symptom associated with the first physiological symptom based on symptom associated information includes:
performing prediction processing on the first physiological symptom to obtain an auxiliary physiological label;
determining a second physiological symptom associated with the first physiological symptom based on an association between a different physiological symptom belonging to the secondary physiological label and the secondary physiological label.
4. The method of claim 3, wherein the predictively processing the first physiological symptom to obtain the secondary physiological signature comprises:
coding the first physiological symptom to obtain a first characteristic representation of the first physiological symptom;
and performing classification prediction on the first feature representation to obtain the auxiliary physiological label.
5. The method of claim 2, wherein the symptom association information includes an association between different physiological symptoms;
the determining a second physiological symptom associated with the first physiological symptom based on symptom associated information includes:
determining a second physiological symptom associated with the first physiological symptom based on an association relationship between different physiological symptoms.
6. The method of any one of claims 1 to 5, wherein determining a symptom supplementing guidance statement according to the target physiological label comprises:
and generating the symptom supplementing guide statement according to a third physiological symptom corresponding to the target physiological label.
7. The method according to claim 6, wherein the symptom supplement guide sentence comprises a guide question sentence and a guide answer choice, the guide answer choice is used for providing supplement content of the first physiological symptom, the guide question sentence is used for prompting selection of the guide answer choice, and the guide question sentence is predefined;
the generating the symptom supplementing guidance statement according to the third physiological symptom corresponding to the target physiological label comprises:
determining the guide answer option according to the third physiological symptom.
8. The method of claim 7, wherein the third physiological symptom comprises a common physiological symptom and a unique physiological symptom, wherein the common physiological symptom corresponds to more than a physiological tags of the target physiological tags, and the unique physiological symptom corresponds to no more than a physiological tags of the target physiological tags;
wherein a is a positive integer not exceeding the number of target physiological labels;
the determining the guide answer option according to the third physiological symptom comprises:
determining the leading answer option based on the unique physiological symptom.
9. The method of any of claims 1 to 5, further comprising:
reconstructing the first physiological symptom and the second physiological symptom according to prompt information to obtain reconstruction information;
the prompt information comprises a slot position and prompt content, the slot position is used for providing a filling position of the target physiological label, and the prompt content is used for prompting to fill the slot position;
the predicting the first physiological symptom and the second physiological symptom to obtain a target physiological label and a predicted probability of the target physiological label includes:
and predicting the reconstruction information to obtain a target physiological label and the prediction probability of the target physiological label, wherein the target physiological label is used for filling the slot position.
10. A method for training a physiological label prediction model, the method comprising:
obtaining a sample physiological symptom and a sample physiological label, wherein the sample physiological symptom is used for describing a physiological abnormal state of a sample object, and the sample physiological label is used for describing a physiological abnormal type of the sample object;
determining a supplemental physiological symptom associated with the sample physiological symptom, the supplemental physiological symptom comprising at least one physiological symptom having an associative relationship with the sample physiological symptom;
calling an initial physiological label prediction model to perform prediction processing on the sample physiological symptoms and the supplementary physiological symptoms to obtain a predicted physiological label and a predicted probability of the predicted physiological label;
and carrying out backward propagation training on the initial physiological label prediction model based on the prediction error between the predicted physiological label and the sample physiological label to obtain the physiological label prediction model.
11. A physiological signature prediction apparatus, characterized in that the apparatus comprises:
the acquisition module is used for acquiring an input first physiological symptom, and the first physiological symptom is used for describing a physiological abnormal state of a target object;
a determination module for determining a second physiological symptom associated with the first physiological symptom, wherein the second physiological symptom comprises at least one physiological symptom having an association relation with the first physiological symptom;
the prediction module is used for performing prediction processing on the first physiological symptom and the second physiological symptom to obtain a target physiological label and a prediction probability of the target physiological label, wherein the target physiological label is used for describing a physiological abnormity type of the target object;
an output module, configured to output the target physiological label if the predicted probability of the target physiological label is greater than a target threshold;
the output module is further configured to determine, according to the target physiological label, a symptom supplement guidance statement when the prediction probability of the target physiological label is not greater than the target threshold, where the symptom supplement guidance statement is used to guide supplement of the first physiological symptom.
12. An apparatus for training a physiological label prediction model, the apparatus comprising:
the system comprises an acquisition module, a data processing module and a data processing module, wherein the acquisition module is used for acquiring sample physiological symptoms and sample physiological labels, the sample physiological symptoms are used for describing physiological abnormal states of sample objects, and the sample physiological labels are used for describing physiological abnormal types of the sample objects;
a determination module for determining a supplemental physiological symptom associated with the sample physiological symptom, the supplemental physiological symptom comprising at least one physiological symptom having an association relationship with the sample physiological symptom;
the prediction module is used for calling an initial physiological label prediction model to carry out prediction processing on the sample physiological symptoms and the supplementary physiological symptoms to obtain a predicted physiological label and a prediction probability of the predicted physiological label;
and the training module is used for carrying out backward propagation training on the initial physiological label prediction model based on the prediction error between the predicted physiological label and the sample physiological label to obtain the physiological label prediction model.
13. A computer device, characterized in that the computer device comprises: a processor and a memory, wherein at least one program is stored in the memory; the processor is configured to execute the at least one program in the memory to implement the physiological label prediction method according to any one of claims 1 to 9 and/or the training method of the physiological label prediction model according to claim 10.
14. A computer-readable storage medium having stored thereon executable instructions that are loaded and executed by a processor to implement the method for predicting physiological signatures as set forth in any one of claims 1 to 9 and/or the method for training a physiological signature prediction model as set forth in claim 10.
15. A computer program product comprising computer instructions stored in a computer readable storage medium, which are read by a processor and executed to implement the method for physiological signature prediction as defined in any one of claims 1 to 9 and/or the method for training a physiological signature prediction model as defined in claim 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210469503.8A CN115132303A (en) | 2022-04-28 | 2022-04-28 | Physiological label prediction method, model training method, device, equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210469503.8A CN115132303A (en) | 2022-04-28 | 2022-04-28 | Physiological label prediction method, model training method, device, equipment and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115132303A true CN115132303A (en) | 2022-09-30 |
Family
ID=83376018
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210469503.8A Pending CN115132303A (en) | 2022-04-28 | 2022-04-28 | Physiological label prediction method, model training method, device, equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115132303A (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN109326352A (en) * | 2018-10-26 | 2019-02-12 | 腾讯科技（深圳）有限公司 | Disease forecasting method, apparatus, terminal and storage medium |
| CN110444259A (en) * | 2019-06-06 | 2019-11-12 | 昆明理工大学 | Traditional Chinese medical electronic case history entity relationship extracting method based on entity relationship mark strategy |
| CN111274373A (en) * | 2020-01-16 | 2020-06-12 | 山东大学 | Electronic medical record question-answering method and system based on knowledge graph |
| CN111723209A (en) * | 2020-06-28 | 2020-09-29 | 上海携旅信息技术有限公司 | Semi-supervised text classification model training method, text classification method, system, device and medium |
| US20200388396A1 (en) * | 2019-06-04 | 2020-12-10 | Dana-Farber Cancer Institute, Inc. | System and method of using machine learning for extraction of symptoms from electronic health records |
| WO2021151328A1 (en) * | 2020-09-04 | 2021-08-05 | 平安科技（深圳）有限公司 | Symptom data processing method and apparatus, and computer device and storage medium |
| CN113257412A (en) * | 2021-06-16 | 2021-08-13 | 腾讯科技（深圳）有限公司 | Information processing method, information processing device, computer equipment and storage medium |
| WO2021174728A1 (en) * | 2020-03-04 | 2021-09-10 | 平安国际智慧城市科技股份有限公司 | Triage data processing method and apparatus, computer device, and storage medium |
| CN113780454A (en) * | 2021-09-17 | 2021-12-10 | 平安科技（深圳）有限公司 | Model training and calling method and device, computer equipment and storage medium |
-
2022
- 2022-04-28 CN CN202210469503.8A patent/CN115132303A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN109326352A (en) * | 2018-10-26 | 2019-02-12 | 腾讯科技（深圳）有限公司 | Disease forecasting method, apparatus, terminal and storage medium |
| US20200388396A1 (en) * | 2019-06-04 | 2020-12-10 | Dana-Farber Cancer Institute, Inc. | System and method of using machine learning for extraction of symptoms from electronic health records |
| CN110444259A (en) * | 2019-06-06 | 2019-11-12 | 昆明理工大学 | Traditional Chinese medical electronic case history entity relationship extracting method based on entity relationship mark strategy |
| CN111274373A (en) * | 2020-01-16 | 2020-06-12 | 山东大学 | Electronic medical record question-answering method and system based on knowledge graph |
| WO2021174728A1 (en) * | 2020-03-04 | 2021-09-10 | 平安国际智慧城市科技股份有限公司 | Triage data processing method and apparatus, computer device, and storage medium |
| CN111723209A (en) * | 2020-06-28 | 2020-09-29 | 上海携旅信息技术有限公司 | Semi-supervised text classification model training method, text classification method, system, device and medium |
| WO2021151328A1 (en) * | 2020-09-04 | 2021-08-05 | 平安科技（深圳）有限公司 | Symptom data processing method and apparatus, and computer device and storage medium |
| CN113257412A (en) * | 2021-06-16 | 2021-08-13 | 腾讯科技（深圳）有限公司 | Information processing method, information processing device, computer equipment and storage medium |
| CN113780454A (en) * | 2021-09-17 | 2021-12-10 | 平安科技（深圳）有限公司 | Model training and calling method and device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021072852A1 (en) | Sequence labeling method and system, and computer device | |
| US11468239B2 (en) | Joint intent and entity recognition using transformer models | |
| JP2022177220A (en) | Method for training text recognition model, method for recognizing text, and device for recognizing text | |
| US10789942B2 (en) | Word embedding system | |
| CN114676704B (en) | Sentence emotion analysis method, device and equipment and storage medium | |
| CN110674303B (en) | Event statement processing method and device, computer equipment and readable storage medium | |
| CN114942984A (en) | Visual scene text fusion model pre-training and image-text retrieval method and device | |
| CN114021582B (en) | Spoken language understanding method, device, equipment and storage medium combined with voice information | |
| CN113240510A (en) | Abnormal user prediction method, device, equipment and storage medium | |
| CN112101042A (en) | Text emotion recognition method and device, terminal device and storage medium | |
| CN115795038A (en) | Intention identification method and device based on localization deep learning framework | |
| CN113626608B (en) | Semantic-enhancement relationship extraction method and device, computer equipment and storage medium | |
| CN113157941B (en) | Service characteristic data processing method, service characteristic data processing device, text generating method, text generating device and electronic equipment | |
| US20230070966A1 (en) | Method for processing question, electronic device and storage medium | |
| CN115132303A (en) | Physiological label prediction method, model training method, device, equipment and medium | |
| CN116821339A (en) | Misuse language detection method, device and storage medium | |
| CN112818688B (en) | Text processing method, device, equipment and storage medium | |
| CN115687136A (en) | Script program processing method, system, computer equipment and medium | |
| CN110780850B (en) | Requirement case auxiliary generation method and device, computer equipment and storage medium | |
| CN111199170B (en) | Formula file identification method and device, electronic equipment and storage medium | |
| CN111967253A (en) | Entity disambiguation method and device, computer equipment and storage medium | |
| CN113537560A (en) | Method, system, electronic device and storage medium for predicting user insurance application will | |
| CN113780454A (en) | Model training and calling method and device, computer equipment and storage medium | |
| CN113204964B (en) | Data processing method, system, electronic equipment and storage medium | |
| CN113515931B (en) | Text error correction method, device, computer equipment and storage medium |
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 |