CN111553412A - Method, device and equipment for training precocious puberty classification model - Google Patents
Method, device and equipment for training precocious puberty classification model Download PDFInfo
- Publication number
- CN111553412A CN111553412A CN202010344479.6A CN202010344479A CN111553412A CN 111553412 A CN111553412 A CN 111553412A CN 202010344479 A CN202010344479 A CN 202010344479A CN 111553412 A CN111553412 A CN 111553412A
- Authority
- CN
- China
- Prior art keywords
- classification
- image
- training
- information
- classification model
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The present application relates to a method, apparatus, computer device and storage medium for training a sexual precocity classification model. The method for training the sexual precocity classification model comprises the following steps: acquiring text information and image information for representing the type of precocity; extracting a classification parameter value corresponding to a preset precocious classification parameter from the text information; carrying out feature extraction processing on the image information to obtain image features; and training the precocious puberty classification model by using the classification parameter values and the image characteristics. By adopting the method, the sexual precocity classification model can be trained by utilizing the classification parameter values and the image characteristics, the data dimension of the training data of the sexual precocity classification model is increased, and the classification performance of the sexual precocity classification model is improved.
Description
Technical Field
The present application relates to the field of computer technologies, and in particular, to a method, an apparatus, a computer device, and a storage medium for training a sexual precocity classification model.
Background
With the development of computer technology, artificial intelligence technology is widely applied in various industries, and one of the key links of artificial intelligence technology is the training of models to obtain a model with better performance. The artificial intelligence technology can be applied to the field of classifying sexual precocity, but the performance of the traditional training sexual precocity classification model is not optimized. The inventors found that the reason for the suboptimal classification performance is the single dimension of data of the traditional training early-maturing classification model, for example, the training data are only the volume, long diameter, transverse diameter and the like of the uterus/ovary, and the data are only some structured data reflecting the development of a single secondary sex character. Therefore, how to train the sexual precocity classification model is a problem to be solved in order to improve the classification performance of the sexual precocity classification model.
Disclosure of Invention
In view of the above, it is desirable to provide a method, an apparatus, a computer device and a storage medium for training a precocious puberty classification model, which can improve the classification performance of the precocious puberty classification model.
A method of training a precocious puberty classification model, comprising:
acquiring text information and image information for representing the type of precocity;
extracting a classification parameter value corresponding to a preset precocious classification parameter from the text information;
carrying out feature extraction processing on the image information to obtain image features;
and training the precocious puberty classification model by using the classification parameter values and the image characteristics.
In one embodiment, the step of performing feature extraction processing on the image information to obtain an image feature includes:
inputting the image information into a pre-trained convolutional neural network, controlling the convolutional neural network to extract bone morphological characteristics from the image information, and taking the bone morphological characteristics as the image characteristics; the bone morphology features are image features related to bone age.
In one embodiment, before the step of inputting the image information into the pre-trained convolutional neural network, the method further includes:
acquiring an image training sample, inputting the image training sample into a convolutional neural network constructed based on a DenseNet model, and controlling the convolutional neural network to finish training according to the image training sample.
In one embodiment, the precocious puberty classification parameter comprises at least one of the level of development of the second sexual characteristics, bone age, size of the reproductive organs.
In one embodiment, the classification parameter value and the image feature are multiple;
the step of training the precocious puberty classification model by using the classification parameter values and the image features comprises:
classifying the classification parameter values and the image characteristics of the same sample into the same group;
and training the precocious puberty classification model according to the classification parameter values and the image characteristics of each group.
In one embodiment, the precocious classification model is an XGBoost classifier;
the step of training the precocious puberty classification model according to the classification parameter values and the image features of each group comprises the following steps:
and inputting the classification parameter values and the image characteristics of each group into the XGboost classifier, and controlling the XGboost classifier to train according to a ten-fold cross validation mode.
In one embodiment, the method further comprises:
performing reason analysis processing on the classification result output by the sexual precocity classification model to obtain corresponding classification reason information;
and sending the classification reason information and the classification result to a target terminal so that the target terminal outputs the classification reason information and the classification result.
An apparatus for training a precocious puberty classification model, comprising:
the information acquisition module is used for acquiring text information and image information for representing the sexual precocity category;
the classification parameter value extraction module is used for extracting a classification parameter value corresponding to a preset precocious classification parameter from the text information;
the image feature extraction module is used for carrying out feature extraction processing on the image information to obtain image features;
and the model training module is used for training the precocious classification model by using the classification parameter values and the image characteristics.
A computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
acquiring text information and image information for representing the type of precocity;
extracting a classification parameter value corresponding to a preset precocious classification parameter from the text information;
carrying out feature extraction processing on the image information to obtain image features;
and training the precocious puberty classification model by using the classification parameter values and the image characteristics.
A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, carries out the steps of:
acquiring text information and image information for representing the type of precocity;
extracting a classification parameter value corresponding to a preset precocious classification parameter from the text information;
carrying out feature extraction processing on the image information to obtain image features;
and training the precocious puberty classification model by using the classification parameter values and the image characteristics.
According to the method, the device, the computer equipment and the storage medium for training the sexual precocity classification model, after the server obtains the text information and the image information for representing the sexual precocity category, the server extracts the classification parameter value corresponding to the preset sexual precocity classification parameter from the text information, performs feature extraction processing on the image information to obtain the image feature, trains the sexual precocity classification model by using the classification parameter value and the image feature, increases the data dimension of the training data of the sexual precocity classification model, and improves the classification performance of the sexual precocity classification model.
Drawings
FIG. 1 is a diagram illustrating an internal structure of a computer device according to an embodiment;
FIG. 2 is a schematic flow chart diagram illustrating a method for training a precocious puberty classification model in one embodiment;
FIG. 3 is a flowchart illustrating the steps of training a precocious puberty classification model in one embodiment;
FIG. 4 is a block diagram of an apparatus for training a precocious puberty classification model in one embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
Reference in the specification to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the specification. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. It is explicitly and implicitly understood by one skilled in the art that the embodiments described herein can be combined with other embodiments.
The method for training the precocious sexual classification model provided by the application can be applied to a server shown in fig. 1, and the internal structure diagram of the server can be shown in fig. 1. The computer device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is used for storing data for training a sexual precocity classification model. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is for, when executed by a processor, implementing a method of training a precocious sexual classification model.
The server may be implemented by an independent server or a server cluster composed of a plurality of servers.
Those skilled in the art will appreciate that the architecture shown in fig. 1 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, as shown in fig. 2, a method for training a precocious sexual classification model is provided, which is illustrated by taking the method as an example applied to the server in fig. 1, and includes the following steps:
step S202, acquiring text information and image information for representing the sexual precocity category.
Wherein, the precocious categories may include central precocious and non-central precocious; the text information and the image information can both reflect the growth and development conditions, so as to characterize the sexual precocity category. Further, the text information may include information about parameters (which may be characterized in the form of chinese characters, english characters, etc., such as "bone age", "bone age") and corresponding parameter values (which may be characterized in the form of numerical values, such as "8") obtained from the medical record; the server can perform structured processing on the text information to obtain structured text information; the structured text information can be understood as structured data, which is equivalent to text information arranged according to a preset format; according to a preset format, parameters and corresponding parameter values can be extracted from the text information after the structured processing. For example, if the text information is "Alice is 8 years old in bone age and 3mL in unilateral ovarian volume", the text information may be structured by using a natural language processing technique, and information on the type of precocious puberty (for example, "bone age", "8", "unilateral ovarian volume", and "3") may be extracted from the structured text information. The image information may correspond to image forms such as pictures and videos, and since the image is represented in a pixel value form (further, may be a matrix formed by pixel values), and cannot be subjected to text structuring, the image information may be regarded as unstructured data; the server may perform image feature extraction processing on the unstructured image information to obtain corresponding image features, for example, the original image information is represented in a 256 × 256 pixel value matrix form, and the server may perform feature extraction processing on the pixel value matrix according to a category to which the image belongs or an image report conclusion to obtain a 64 × 64 feature matrix, where the feature matrix carries the image features. Further, the image information may correspond to a bone age radiograph and/or a pelvic ultrasound image.
Specifically, the manner of acquiring the text information and the image information by the server may be: the server can send the sexual precocity information acquisition instruction to the electronic medical record database, and after the electronic medical record database receives the sexual precocity information acquisition instruction, the electronic medical record database sends text information and image information related to the sexual precocity category to the server from the electronic medical records of one or more samples; the electronic medical record can comprise past medical history information, an age of bone X-ray film and a pelvic cavity ultrasonic image, wherein the past medical history information is text information, and the age of bone X-ray film and the pelvic cavity ultrasonic image are image information. In one embodiment, the server may also obtain structured data characterized in the form of a data table, such as hormone test values; specifically, the electronic medical record database can extract the hormone test value from the electronic medical record and send the hormone test value to the server.
Step S204, extracting a classification parameter value corresponding to the preset precocious puberty classification parameter from the text information.
The precocious puberty classification parameter may be a parameter related to a precocious puberty category, and may include one or more of a development grade (e.g., breast and pubic hair development Tanner stage grades I-V, etc.), a bone age, and a size of a reproductive organ (e.g., unilateral ovarian volume), and the classification parameter value may be a numerical value corresponding to the precocious puberty classification parameter, for example, in the description of step S202, the classification parameter value corresponding to the bone age is "8" and the parameter value corresponding to the unilateral ovarian volume is "3".
In this step, after obtaining the text information, the server may extract a classification parameter value corresponding to the precocious classification parameter from the text information, where the server may perform a structural processing on the text information by using a natural language processing technique, and extract the classification parameter value from the text information after the structural processing.
In step S206, feature extraction processing is performed on the image information to obtain image features.
Because the image information is information corresponding to image forms such as pictures and videos, after the server acquires the image information, feature extraction processing can be performed on the image information corresponding to the picture forms to obtain image features, wherein the image features carry information related to precocious puberty categories. For example, Alice has a bone morphology (bone morphology is related to sexual precocity) corresponding to a bone age X-ray film, and since the bone morphology is related to bone age and bone age is related to a category of sexual precocity, a bone morphology feature (belonging to one of image features) obtained by performing feature extraction processing on image information corresponding to the bone age X-ray film is related to a category of sexual precocity. If the image information is information corresponding to a video format, a corresponding frame image may be extracted from the video first, and feature extraction processing may be performed on the frame image.
And S208, training the precocious puberty classification model by using the classification parameter values and the image characteristics.
The server processes the text information and the image information to obtain corresponding classification parameter values and image characteristics, and trains the precocious classification model by taking the obtained classification parameters and the obtained image characteristics as training data. The precocious early-maturing classification model may be an XGBoost classifier, an AdaBoost classifier, a GBDT classifier (iterative decision tree classifier), or the like.
In one embodiment, in order to further improve the classification performance of the sexual precocity classification model, the sexual precocity classification model may be an XGBoost classifier, and after obtaining corresponding classification parameter values and image features, the server inputs the obtained classification parameter values and image features as training data into the XGBoost classifier, and trains the XGBoost classifier in a ten-fold cross validation manner.
In another embodiment, to further improve the classification accuracy of the precocious puberty classification model, the precocious puberty classification model may be trained according to the classification parameter values and the image features of a plurality of samples, specifically: the method comprises the steps that a server collects text information and image information of a plurality of samples, extracts corresponding classification parameter values and image features from the text information and the image information respectively, and then associates the classification parameter values and the image features of the same sample, namely, the classification parameter values and the image features are classified into the same group; and then the server inputs the related classification parameter values and image features (namely the classification parameter values and the image features of the same group) into the XGboost classifier as training data, and the XGboost classifier is trained in a ten-fold cross validation mode.
In the method for training the sexual precocity classification model, after acquiring text information and image information for representing the sexual precocity category, a server extracts classification parameter values corresponding to preset sexual precocity classification parameters from the text information, performs feature extraction processing on the image information to obtain image features, trains the sexual precocity classification model by using the classification parameter values and the image features, increases the data dimension of training data of the sexual precocity classification model, and improves the classification performance of the sexual precocity classification model; if the difference between the variable numbers of the text information and the image information is too large, for example, if the variable number of the text information is only 30 (bone age, size of reproductive organ, etc.), the variable number of the image information is thousands of (since the image information of different patients is different and the image information is represented by pixel values, each pixel value is equivalent to one variable, that is, the variable number included in the image information may be thousands of), at this time, the text information and the image information are directly input into the precocious puberty classification model for training, which may cause the image information to cover the text information, the precocious puberty classification model is over-fitted, and the classification accuracy is reduced; therefore, in the method for training the sexual precocity classification model, the image information is firstly subjected to feature extraction processing, the acquired image features are used for training the sexual precocity classification model, the number of the obtained variables of the image features is less than that of the variables of the image information before feature extraction processing, the condition that the sexual precocity classification model is over-fitted due to overlarge difference of the number of the data variables can be avoided, and the accuracy of classifying the sexual precocity classification model is improved.
In one embodiment, in order to improve the training speed of the sexual precocity classification model, the image information can be input into a convolutional neural network which is trained in advance; specifically, after acquiring the image information, the server inputs the image information into a convolutional neural network trained in advance, and the convolutional neural network extracts bone morphology features related to bone age (or precocious puberty category) from the image information according to parameters trained in advance.
In order to further guarantee the classification performance of the sexual precocity classification model, a convolutional neural network can be constructed based on a DenseNet model, and an image training sample is used for training the convolutional neural network, wherein the image training sample can be obtained from an image database and can comprise a bone age X-ray film, a pelvic cavity ultrasonic image and the like; specifically, after the image training samples are obtained by the server, the image training samples are input into a convolutional neural network constructed based on a DenseNet model, the convolutional neural network is trained according to the image training samples to obtain a trained convolutional neural network model, and the trained convolutional neural network model is used as an image feature extraction model.
In some scenarios, the precocious classification model only outputs the precocious classification result after completing classification, and the reason for influencing the precocious classification model to output the classification result is unknown; therefore, in order to further enable the reason of the classification result output by the precocious early-maturing classification model to have the knowability and the interpretability, after the server obtains the classification result output by the precocious early-maturing classification model, the server can also perform reason analysis processing on the classification result output by the precocious early-maturing classification model (analyze the reason of the classification result output by the precocious early-maturing classification model), further obtain corresponding classification reason information, send the classification reason information and the classification result to a target terminal (such as a terminal device corresponding to an electronic medical record system), and the target terminal outputs the received classification reason information and the classification result, wherein the output mode can be display on a screen or voice playing. The server may perform cause analysis processing on the classification result output by the precocious puberty classification Model by using a Local interim Model-aggregation extensions (LIME) Model.
For better understanding of the above method, an application example of the method for training the sexual precocity classification model according to the present application is described in detail with reference to fig. 3:
step S302, acquiring text information and image information: text information (such as past medical history information), a hormone test value, a bone age X-ray film and a pelvic cavity ultrasonic image are obtained from a database (equivalent to an electronic medical record database) of the electronic medical record system and are sent to a server.
Step S304, processing text information and image information:
processing the text information by using a natural language technology to structure the text information, and extracting classification parameter values from the structured text information, wherein the classification parameter values respectively correspond to the development grade, the bone age and the reproductive organ size of the second sex characteristics; training a convolutional neural network constructed based on a DenseNet model by using an image training sample, inputting image information corresponding to the bone age X-ray film into the trained convolutional neural network, and controlling the convolutional neural network to perform feature extraction processing on the image information to obtain bone morphological features corresponding to the bone age X-ray film. The core of the DenseNet model is to establish dense connection between all layers in the front and the layers behind, the connection mode is that each layer is connected with the layer in the front in the channel dimension, and the whole is used as the input of the next layer, so that the feature reuse is realized, and the efficiency of extracting image features by the convolutional neural network can be improved.
Step S306, construction of a precocious puberty classification model: classifying parameter values and bone morphological characteristics of the same sample into the same group, performing class labeling (labeling of central precocity class and non-central precocity class) on each group according to a precocity classification standard, inputting the labeled classification parameter values and bone morphological characteristics in each group into an XGboost classifier, and training the XGboost classifier by using a ten-fold cross validation mode for 20 times to obtain the trained XGboost classifier (i.e. a precocity classification model). The algorithm in the XGboost classifier is a lifting tree model of an integrated algorithm, and a plurality of CART regression tree models are integrated together to form a strong classifier; the XGboost classifier adds a regular term in the objective function for controlling the complexity of the model; the regular term comprises the number of leaf nodes of the tree, the target function of the regular term is added, and the variance of the sexual precocity classification model is reduced, so that the classification result of the sexual precocity classification model is more stable, and overfitting can be prevented; and the XGboost classifier supports parallel processing of calculation, reduces the calculation amount and improves the classification speed of the sexual precocity classification model.
Step S308, performing reason analysis processing on the classification result: the server utilizes the LIME model to perform reason analysis processing on the classification result output by the sexual precocity classification model to obtain classification reason information, so that the reason of the classification result output by the sexual precocity classification model can be explained (the important characteristic can be understood to cause the sexual precocity classification model to output a corresponding classification result).
Step S310, transmission of classification results and classification reason information: and the server sends the classification result and the classification reason information to the target terminal for display and output, and sends the classification result and the classification reason information back to the database of the electronic medical record system for storage.
In the traditional technology, a classification decision tree and BP neural network technology can be utilized to construct a central sexual precocity classification model of the girl based on pelvic cavity ultrasonic report information of the sexual precocity sick child, but the model only utilizes data of one source/dimension of the sexual precocity sick child, the used sample size is limited, and the reliability of the result is still to be verified.
The method combines the previous data of multiple sources/dimensions such as medical history information, physical examination data, sex hormone values, growth hormone values, bone age X-ray films, pelvic cavity ultrasonic images and the like, and solves the technical problem that the classification performance of a sexual precocity classification model is not high because single-dimension data cannot fully cover the development progress information of a patient; the multidimensional data fusion is realized, the corresponding processing is carried out on the data with different dimensions, and the training speed and the classification performance of the sexual precocity classification model can be improved.
It should be understood that although the above embodiments refer to the sexual precocity classification parameters of female organs such as "unilateral ovarian volume", this should not be understood as the data source (such as text information and image information) for training the sexual precocity classification model is only related to female, and those skilled in the art can select the data source according to actual situations. Although the respective steps in the flowcharts of fig. 2 to 3 are sequentially shown as indicated by arrows, the steps are not necessarily performed sequentially in the order indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in fig. 2 to 3 may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of performing the steps or stages is not necessarily sequential, but may be performed alternately or alternately with other steps or at least some of the other steps or stages.
In one embodiment, as shown in fig. 4, there is provided an apparatus 400 for training a precocious sexual classification model, comprising: an information acquisition module 402, a classification parameter value extraction module 404, an image feature extraction module 406, and a model training module 408, wherein:
an information obtaining module 402, configured to obtain text information and image information for characterizing a sexual precocity category;
a classification parameter value extraction module 404, configured to extract a classification parameter value corresponding to a preset precocious classification parameter from the text information;
the image feature extraction module 406 is configured to perform feature extraction processing on the image information to obtain an image feature;
and the model training module 408 is configured to train the precocious puberty classification model by using the classification parameter values and the image features.
In an embodiment, the image feature extraction module 406 is further configured to input the image information into a pre-trained convolutional neural network, control the convolutional neural network to extract bone shape features from the image information, and use the bone shape features as image features; bone morphology features are image features related to bone age.
In an embodiment, the image feature extraction module 406 is further configured to obtain an image training sample, input the image training sample into a convolutional neural network constructed based on a DenseNet model, and control the convolutional neural network to complete training according to the image training sample.
In one embodiment, the precocious puberty classification parameter comprises at least one of development grade, bone age, size of reproductive organs of the second sex characteristics.
In one embodiment, the classification parameter value and the image feature are multiple; the model training module 408 is further configured to classify the classification parameter values and the image features of the same sample into the same group; and training the precocious puberty classification model according to the classification parameter values and the image characteristics of each group.
In one embodiment, the precocious classification model is an XGBoost classifier; the model training module 408 is further configured to input the classification parameter values and the image features of each group into the XGBoost classifier, and control the XGBoost classifier to perform training in a ten-fold cross validation manner.
In one embodiment, the apparatus 400 for training a precocious puberty classification model further comprises: the system comprises a reason analysis module and a classification result sending module; the reason analysis module is used for carrying out reason analysis processing on the classification result output by the precocious puberty classification model to obtain corresponding classification reason information; and the classification result sending module is used for sending the classification reason information and the classification result to the target terminal so as to enable the target terminal to output the classification reason information and the classification result.
For the specific limitation of the apparatus for training the precocious puberty classification model, reference may be made to the above limitation on the method for training the precocious puberty classification model, and details are not repeated here. The modules in the apparatus for training the precocious sexual classification model can be wholly or partially implemented by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:
acquiring text information and image information for representing the type of precocity;
extracting a classification parameter value corresponding to a preset precocious classification parameter from the text information;
carrying out feature extraction processing on the image information to obtain image features;
and training the sexual precocity classification model by using the classification parameter values and the image characteristics.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
inputting the image information into a pre-trained convolutional neural network, controlling the convolutional neural network to extract bone morphological characteristics from the image information, and taking the bone morphological characteristics as image characteristics; bone morphology features are image features related to bone age.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
acquiring an image training sample, inputting the image training sample into a convolutional neural network constructed based on a DenseNet model, and controlling the convolutional neural network to finish training according to the image training sample.
In one embodiment, the classification parameter value and the image feature are multiple;
the processor, when executing the computer program, further performs the steps of:
classifying the classification parameter values and the image characteristics of the same sample into the same group;
training the precocious puberty classification model according to the classification parameter values and the image characteristics of each group
In one embodiment, the precocious classification model is an XGBoost classifier;
the processor, when executing the computer program, further performs the steps of:
and inputting the classification parameter values and the image characteristics of each group into the XGboost classifier, and controlling the XGboost classifier to train according to a ten-fold cross validation mode.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
performing reason analysis processing on the classification result output by the sexual precocity classification model to obtain corresponding classification reason information;
and sending the classification reason information and the classification result to the target terminal so that the target terminal outputs the classification reason information and the classification result.
In one embodiment, the precocious puberty classification parameter comprises at least one of development grade, bone age, size of reproductive organs of the second sex characteristics.
It should be noted that, the steps executed by the processor in the computer device correspond to the method for training the sexual precocity classification model in the present application, and the contents and the corresponding technical effects described in the embodiment of the method for training the sexual precocity classification model are all applicable to the embodiment of the computer device, and are not described herein again.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
acquiring text information and image information for representing the type of precocity;
extracting a classification parameter value corresponding to a preset precocious classification parameter from the text information;
carrying out feature extraction processing on the image information to obtain image features;
and training the sexual precocity classification model by using the classification parameter values and the image characteristics.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
inputting the image information into a pre-trained convolutional neural network, controlling the convolutional neural network to extract bone morphological characteristics from the image information, and taking the bone morphological characteristics as image characteristics; bone morphology features are image features related to bone age.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
acquiring an image training sample, inputting the image training sample into a convolutional neural network constructed based on a DenseNet model, and controlling the convolutional neural network to finish training according to the image training sample.
In one embodiment, the classification parameter value and the image feature are multiple;
the processor, when executing the computer program, further performs the steps of:
classifying the classification parameter values and the image characteristics of the same sample into the same group;
training the precocious puberty classification model according to the classification parameter values and the image characteristics of each group
In one embodiment, the precocious classification model is an XGBoost classifier;
the processor, when executing the computer program, further performs the steps of:
and inputting the classification parameter values and the image characteristics of each group into the XGboost classifier, and controlling the XGboost classifier to train according to a ten-fold cross validation mode.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
performing reason analysis processing on the classification result output by the sexual precocity classification model to obtain corresponding classification reason information;
and sending the classification reason information and the classification result to the target terminal so that the target terminal outputs the classification reason information and the classification result.
In one embodiment, the precocious puberty classification parameter comprises at least one of development grade, bone age, size of reproductive organs of the second sex characteristics.
It should be noted that, the steps executed by the processor in the computer device correspond to the method for training the sexual precocity classification model in the present application, and the contents and the corresponding technical effects described in the embodiment of the method for training the sexual precocity classification model are all applicable to the embodiment of the computer device, and are not described herein again.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a non-volatile computer-readable storage medium, and can include the processes of the embodiments of the methods described above when executed. Any reference to memory, storage, database or other medium used in the embodiments provided herein can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above examples only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1. A method of training a precocious puberty classification model, comprising:
acquiring text information and image information for representing the type of precocity;
extracting a classification parameter value corresponding to a preset precocious classification parameter from the text information;
carrying out feature extraction processing on the image information to obtain image features;
and training the precocious puberty classification model by using the classification parameter values and the image characteristics.
2. The method according to claim 1, wherein the step of performing feature extraction processing on the image information to obtain image features comprises:
inputting the image information into a pre-trained convolutional neural network, controlling the convolutional neural network to extract bone morphological characteristics from the image information, and taking the bone morphological characteristics as the image characteristics; the bone morphology features are image features related to bone age.
3. The method of claim 2, further comprising, prior to the step of inputting the image information into a pre-trained convolutional neural network:
acquiring an image training sample, inputting the image training sample into a convolutional neural network constructed based on a DenseNet model, and controlling the convolutional neural network to finish training according to the image training sample.
4. The method of claim 1, wherein said precocious puberty classification parameters comprise at least one of development grade, bone age, size of reproductive organs of secondary sexual characteristics.
5. The method according to any one of claims 1 to 4, wherein the classification parameter value and the image feature are plural;
the step of training the precocious puberty classification model by using the classification parameter values and the image features comprises:
classifying the classification parameter values and the image characteristics of the same sample into the same group;
and training the precocious puberty classification model according to the classification parameter values and the image characteristics of each group.
6. The method of claim 5, wherein the precocious classification model is an XGboost classifier;
the step of training the precocious puberty classification model according to the classification parameter values and the image features of each group comprises the following steps:
and inputting the classification parameter values and the image characteristics of each group into the XGboost classifier, and controlling the XGboost classifier to train according to a ten-fold cross validation mode.
7. The method of claim 1, 2, 3, 4, or 6, further comprising:
performing reason analysis processing on the classification result output by the sexual precocity classification model to obtain corresponding classification reason information;
and sending the classification reason information and the classification result to a target terminal so that the target terminal outputs the classification reason information and the classification result.
8. An apparatus for training a precocious puberty classification model, comprising:
the information acquisition module is used for acquiring text information and image information for representing the sexual precocity category;
the classification parameter value extraction module is used for extracting a classification parameter value corresponding to a preset precocious classification parameter from the text information;
the image feature extraction module is used for carrying out feature extraction processing on the image information to obtain image features;
and the model training module is used for training the precocious classification model by using the classification parameter values and the image characteristics.
9. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor implements the steps of the method of any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010344479.6A CN111553412A (en) | 2020-04-27 | 2020-04-27 | Method, device and equipment for training precocious puberty classification model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010344479.6A CN111553412A (en) | 2020-04-27 | 2020-04-27 | Method, device and equipment for training precocious puberty classification model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111553412A true CN111553412A (en) | 2020-08-18 |
Family
ID=72004104
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010344479.6A Pending CN111553412A (en) | 2020-04-27 | 2020-04-27 | Method, device and equipment for training precocious puberty classification model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111553412A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113256579A (en) * | 2021-05-19 | 2021-08-13 | 扬州大学 | Pulmonary tuberculosis recognition system based on pre-training model |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1839391A (en) * | 2003-06-25 | 2006-09-27 | 美国西门子医疗解决公司 | Systems and methods for automated diagnosis and decision support for breast imaging |
| US20140030253A1 (en) * | 2012-07-27 | 2014-01-30 | Vancouver Biotech Ltd. | HUMANIZED FORMS OF MONOCLONAL ANTIBODIES TO HUMAN GnRH RECEPTOR |
| CN104699928A (en) * | 2013-12-09 | 2015-06-10 | 奥美之路(北京)技术顾问有限公司 | Physique development assessment model for 3-15 years-old children |
| CN107595248A (en) * | 2017-08-31 | 2018-01-19 | 郭淳 | A kind of method and system for detecting and evaluating upgrowth and development of children situation |
| CN108056786A (en) * | 2017-12-08 | 2018-05-22 | 浙江大学医学院附属儿童医院 | A kind of stone age detection method and device based on deep learning |
| CN109146879A (en) * | 2018-09-30 | 2019-01-04 | 杭州依图医疗技术有限公司 | A kind of method and device detecting the stone age |
| CN109243618A (en) * | 2018-09-12 | 2019-01-18 | 腾讯科技（深圳）有限公司 | Construction method, disease label construction method and the smart machine of medical model |
| CN109544518A (en) * | 2018-11-07 | 2019-03-29 | 中国科学院深圳先进技术研究院 | A kind of method and its system applied to the assessment of skeletal maturation degree |
| CN110265119A (en) * | 2018-05-29 | 2019-09-20 | 中国医药大学附设医院 | Bone age assessment and prediction of height model, its system and its prediction technique |
| CN110275880A (en) * | 2019-05-21 | 2019-09-24 | 阿里巴巴集团控股有限公司 | Data analysing method, device, server and readable storage medium storing program for executing |
| CN110555846A (en) * | 2019-10-13 | 2019-12-10 | 浙江德尚韵兴医疗科技有限公司 | full-automatic bone age assessment method based on convolutional neural network |
-
2020
- 2020-04-27 CN CN202010344479.6A patent/CN111553412A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1839391A (en) * | 2003-06-25 | 2006-09-27 | 美国西门子医疗解决公司 | Systems and methods for automated diagnosis and decision support for breast imaging |
| US20140030253A1 (en) * | 2012-07-27 | 2014-01-30 | Vancouver Biotech Ltd. | HUMANIZED FORMS OF MONOCLONAL ANTIBODIES TO HUMAN GnRH RECEPTOR |
| CN104699928A (en) * | 2013-12-09 | 2015-06-10 | 奥美之路(北京)技术顾问有限公司 | Physique development assessment model for 3-15 years-old children |
| CN107595248A (en) * | 2017-08-31 | 2018-01-19 | 郭淳 | A kind of method and system for detecting and evaluating upgrowth and development of children situation |
| CN108056786A (en) * | 2017-12-08 | 2018-05-22 | 浙江大学医学院附属儿童医院 | A kind of stone age detection method and device based on deep learning |
| CN110265119A (en) * | 2018-05-29 | 2019-09-20 | 中国医药大学附设医院 | Bone age assessment and prediction of height model, its system and its prediction technique |
| CN109243618A (en) * | 2018-09-12 | 2019-01-18 | 腾讯科技（深圳）有限公司 | Construction method, disease label construction method and the smart machine of medical model |
| CN109146879A (en) * | 2018-09-30 | 2019-01-04 | 杭州依图医疗技术有限公司 | A kind of method and device detecting the stone age |
| CN109544518A (en) * | 2018-11-07 | 2019-03-29 | 中国科学院深圳先进技术研究院 | A kind of method and its system applied to the assessment of skeletal maturation degree |
| CN110275880A (en) * | 2019-05-21 | 2019-09-24 | 阿里巴巴集团控股有限公司 | Data analysing method, device, server and readable storage medium storing program for executing |
| CN110555846A (en) * | 2019-10-13 | 2019-12-10 | 浙江德尚韵兴医疗科技有限公司 | full-automatic bone age assessment method based on convolutional neural network |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113256579A (en) * | 2021-05-19 | 2021-08-13 | 扬州大学 | Pulmonary tuberculosis recognition system based on pre-training model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112101190B (en) | Remote sensing image classification method, storage medium and computing device | |
| CN109344908B (en) | Method and apparatus for generating a model | |
| CN111754596B (en) | Editing model generation method, device, equipment and medium for editing face image | |
| EP3913542A2 (en) | Method and apparatus of training model, device, medium, and program product | |
| WO2022022154A1 (en) | Facial image processing method and apparatus, and device and storage medium | |
| CN110188829B (en) | Neural network training method, target recognition method and related products | |
| US11314970B1 (en) | Reinforcement learning techniques for automated video summarization | |
| WO2021169364A1 (en) | Semantic emotion analysis method and apparatus, device, and storage medium | |
| US20220101121A1 (en) | Latent-variable generative model with a noise contrastive prior | |
| CN112395979A (en) | Image-based health state identification method, device, equipment and storage medium | |
| CN113821668A (en) | Data classification identification method, device, equipment and readable storage medium | |
| CN112786160A (en) | Multi-image input multi-label gastroscope image classification method based on graph neural network | |
| US20220101122A1 (en) | Energy-based variational autoencoders | |
| CN111553412A (en) | Method, device and equipment for training precocious puberty classification model | |
| CN115115552B (en) | Image correction model training method, image correction device and computer equipment | |
| CN112464924A (en) | Method and device for constructing training set | |
| TWI803243B (en) | Method for expanding images, computer device and storage medium | |
| CN115690276A (en) | Video generation method and device of virtual image, computer equipment and storage medium | |
| CN111445545B (en) | Text transfer mapping method and device, storage medium and electronic equipment | |
| US20220101145A1 (en) | Training energy-based variational autoencoders | |
| CN112801994B (en) | Bone age assessment method and system | |
| CN112948582B (en) | Data processing method, device, equipment and readable medium | |
| CN116129158A (en) | Power transmission line iron tower small part image recognition method and device | |
| CN114283301A (en) | Self-adaptive medical image classification method and system based on Transformer | |
| CN115525740A (en) | Method and device for generating dialogue response sentence, electronic 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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200818 |
|
| RJ01 | Rejection of invention patent application after publication |