A method and system for identifying the metabolic concentration of in vivo components

Abstract

This invention discloses a method for identifying the metabolic concentration of in vivo components, comprising: establishing a classification neural network model for identifying the metabolic concentration of in vivo components; acquiring multiple reagent reaction colors with different gradients as sample images, processing the sample images to obtain a training dataset corresponding to the classification neural network model for identifying the metabolic concentration of in vivo components; obtaining a numerical prediction neural network model for identifying the metabolic concentration of in vivo components based on the interaction between the classification neural network model for identifying the metabolic concentration of in vivo components and the training dataset; acquiring real-time reaction color images corresponding to the metabolic concentration of in vivo components and inputting them into the numerical prediction neural network model for identifying the metabolic concentration of in vivo components, and using the output of the numerical prediction neural network model for identifying the metabolic concentration of in vivo components as the metabolic concentration of in vivo components. The invention also discloses a system, electronic equipment, and computer-readable storage medium, elevating image classification detection to a numerical prediction task, realizing the transformation from a classification problem to a regression problem.

Description

Technical Field

[0001] This invention relates to the field of body composition detection technology, and in particular to a method and system for identifying the metabolic concentration of body components. Background Technology

[0002] Currently, the main methods for reagent reaction color recognition include color histograms, threshold segmentation, color moments, color space conversion, and image color classification using neural networks. Due to factors such as reflections, device errors, shadows, and reagent inhomogeneity in reagent images, each traditional algorithm has limitations. For example, when using color histograms, there are multiple ways to set weights when converting RGB colors to grayscale, and different weight settings will yield slightly different results. Threshold segmentation has limitations when segmenting pixel-level points, especially when the background and reaction colors are similar. Converting the RGB color space to HSV (Hue, Saturation, Value) space also presents challenges in threshold segmentation. Neural network image color recognition, such as Yolov8, uses BCE loss for category detection and DFL Loss + CIOU Loss for bounding box regression. It can be seen that most current neural network image color recognition is categorical detection, with fewer models and applications for numerical prediction. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for identifying the metabolic concentration of in vivo components. This method identifies the reaction color of bodily fluids tested by a multi-well reagent and predicts the concentration of the fluid in each well based on the color intensity. The method involves training a classification model and then performing corresponding numerical predictions. Specifically, a deep neural network model is used to fine-tune the training images. The last layer of the convolutional layer is extracted as a feature. The convolutional layer is divided according to different well positions to extract the features corresponding to each well. The model weights are convolved with the corresponding features to reduce the data to one-dimensional data. The model weights are obtained by linearly fitting the one-dimensional data with different gradients of different samples to their corresponding concentration values. This weight replaces the last layer of the original deep neural network model, thereby realizing the transformation of the deep neural network model from classification prediction to numerical prediction. This allows the model to upgrade image classification detection to image numerical prediction, transforming the problem from classification to regression.

[0004] The first aspect of the present invention is to provide a method for identifying the metabolic concentration of in vivo components, used to identify the reaction color of a body fluid tested by a multi-well reagent, and to predict the concentration of the body fluid in each well based on the intensity of the color, comprising:

[0005] S1, Establish a neural network model for identifying and classifying metabolic concentrations of in vivo components;

[0006] S2, obtain multiple reagent reaction colors with different gradients as sample images, and process the sample images to obtain a training dataset corresponding to the in vivo component metabolic concentration recognition and classification neural network model;

[0007] S3. Based on the interaction between the in vivo component metabolic concentration identification classification neural network model and the training dataset, a numerical prediction neural network model for in vivo component metabolic concentration identification is obtained. The last layer of the convolutional layer of the in vivo component metabolic concentration identification classification neural network model is extracted as a feature. The convolutional layer is divided according to different hole positions to extract the features corresponding to each hole. The model weights are convolved with the corresponding features to reduce the data to one-dimensional data. The one-dimensional data with different gradients for different samples is numerically trained and linearly fitted with the corresponding concentration values ​​to obtain weight values. The weight values ​​are used to replace the last layer of the original deep network model, thereby realizing the transformation of the deep neural network model from classification prediction to numerical prediction.

[0008] S4, acquire the real-time reaction color image corresponding to the metabolic concentration of the body component and input it into the numerical prediction neural network model for identifying the metabolic concentration of the body component. The output of the numerical prediction neural network model for identifying the metabolic concentration of the body component is used as the metabolic concentration of the body component.

[0009] Preferably, the in vivo component metabolic concentration identification and classification neural network model in S1 is a deep neural network model.

[0010] Preferably, S2 includes:

[0011] S21, Obtain multiple reagent reaction colors with different gradients as initial sample images;

[0012] S22, perform preprocessing operations on the initial sample image, the preprocessing operations including: performing one or more operations on the initial sample image such as image removal, noise reduction and removal of invalid regions to obtain the sample image to be labeled;

[0013] S23, label the different pores and different colors in the sample image to be labeled with multi-pore index and concentration, and finally form the training dataset.

[0014] Preferably, step S3, obtaining a numerical prediction neural network model for identifying metabolic concentrations of body components based on the interaction between the classification neural network model for identifying metabolic concentrations of body components and the training dataset, includes:

[0015] S31, based on the in vivo component metabolic concentration identification and classification neural network model and the preprocessing operation, the training dataset is fine-tuned multiple times to obtain multiple fine-tuning results;

[0016] S32, verify the recognition and classification effect of the fine-tuning result, extract the first key model parameters of the in vivo component metabolic concentration recognition and classification neural network model corresponding to the preset recognition and classification effect, and the second key model parameters of the in vivo component metabolic concentration recognition and classification neural network model in S1; the first key model parameters include the last convolutional layer of the model; the second key model parameters include the activation function layer;

[0017] S33, the last convolutional layer of the model is used as the feature layer, and features of different pores are extracted based on the feature layer. The features of different pores are weighted with the in vivo component metabolic concentration recognition and classification neural network model to obtain multidimensional data; the multidimensional data in the multidimensional data is reduced to one-dimensional data.

[0018] S34, numerical training will be performed on the one-dimensional data of different samples and different gradients, and the weight values ​​will be obtained by linear fitting with their corresponding concentration values.

[0019] S35, based on the weight values ​​and their corresponding relationships, the network is updated, and the weight values ​​are used to replace the activation function layer to form the in vivo component metabolic concentration identification numerical prediction neural network model.

[0020] Preferably, the features of the different aperture positions in S33 are the relative position features of each aperture obtained after the last convolutional layer is divided.

[0021] Preferably, the numerical training and linear fitting in S34 is linear regression fitting.

[0022] Preferably, S4 includes:

[0023] S41, acquire real-time reaction color images corresponding to the metabolic concentration of in vivo components and input them into the in vivo component metabolic concentration recognition numerical prediction neural network model.

[0024] S42, Based on the reaction color image, identify the reaction color of the body fluid tested by the porous reagent; Based on the color intensity in the identification result and the metabolic concentration of the body components in the body, a numerical prediction neural network model is used to predict the metabolic concentration of the body components in each well;

[0025] S43, output and display the metabolic concentration of the in vivo component in each well.

[0026] A second aspect of the present invention provides a system for identifying the metabolic concentration of in vivo components, used to implement the method for identifying the metabolic concentration of in vivo components of the first aspect, comprising:

[0027] The first model building module (101) is used to build a neural network model for identifying and classifying metabolic concentrations of in vivo components.

[0028] The training dataset acquisition module (102) is used to acquire multiple reagent reaction colors with different gradients as sample images, and to obtain a training dataset corresponding to the in vivo component metabolic concentration identification and classification neural network model after processing the sample images.

[0029] The second model building module (103) is used to obtain a numerical prediction neural network model for identifying metabolic concentrations of in vivo components based on the interaction between the classification neural network model for identifying metabolic concentrations of in vivo components and the training dataset; the last layer of the convolutional layer of the classification neural network model for identifying metabolic concentrations of in vivo components is extracted as a feature, the convolutional layer is divided according to different hole positions to extract the features corresponding to each hole, the model weights are convolved with the corresponding features to reduce the data to one-dimensional data, the one-dimensional data with different gradients for different samples are numerically trained, and the weight values ​​are obtained by linear fitting with their corresponding concentration values; the weight values ​​are used to replace the last layer of the original deep network model, thereby realizing the transformation of the deep neural network model from classification prediction to numerical prediction;

[0030] The metabolic concentration recognition module (104) is used to acquire real-time reaction color images corresponding to the metabolic concentration of in vivo components and input them into the in vivo component metabolic concentration recognition numerical prediction neural network model. The output of the in vivo component metabolic concentration recognition numerical prediction neural network model is used as the metabolic concentration of in vivo components.

[0031] Application Examples:

[0032] This embodiment mainly uses a multi-well reagent in the urine six-item test to give the concentration value of each well based on the color intensity of the reagent reaction in each well.

[0033] The specific testing steps are as follows:

[0034] (1) Training sample preparation: Prepare reagent samples of different gradients for each well and obtain training sample images through an optical detection system, such as... Figure 4 As shown:

[0035] (2) Perform training preprocessing on the training sample images and label the image data with labelme;

[0036] (3) Select, but not limited to, a Yolov8 deep network model, fine-tune the training on the labeled training data, and save the model; such as Figure 5 As shown.

[0037] (4) Extract the last convolutional layer of the model and divide it to take the relative position features of each well as the features of this sample and this well. After being combined with the model weights, it is reduced to one-dimensional data. This data is then regressed and fitted with the concentration corresponding to this sample and this well to obtain its weight.

[0038] (5) Replace the last activation function layer in the model in step (3) with the weights obtained in step (4) and save the model. The model can be converted from a classification model to a numerical prediction model.

[0039] A third aspect of the present invention is to provide an electronic device, including a processor and a memory, the memory storing a plurality of instructions, the processor being configured to read the instructions and execute the method described in the first aspect.

[0040] A fourth aspect of the present invention is to provide a computer-readable storage medium storing a plurality of instructions which can be read by a processor and executed by the method described in the first aspect.

[0041] The beneficial effects of the method and system of the present invention are as follows:

[0042] (1) The last convolutional layer of the trained model is used as the feature layer, which reduces the impact of algorithm rule mining and different weights during color extraction.

[0043] (2) The feature layer is fitted with the known concentration, thereby realizing the transformation from classification prediction to numerical prediction. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the specific embodiments or related technologies of the present invention, the drawings used in the description of the specific embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0045] Figure 1 This is a flowchart of a method for identifying the metabolic concentration of in vivo components according to an embodiment of the present invention;

[0046] Figure 2 A flowchart illustrating the conversion of a classification model to a numerical prediction model according to an embodiment of the present invention;

[0047] Figure 3 This is a diagram illustrating the architecture of an in vivo component metabolic concentration identification system according to an embodiment of the present invention.

[0048] Figure 4 To prepare training samples according to embodiments of the present invention, a schematic diagram of the principle of preparing reagent samples of different gradients in each well and obtaining training sample images through an optical detection system is provided.

[0049] Figure 5 This is a diagram of the Yolov8 model architecture provided according to an embodiment of the present invention;

[0050] Figure 6 This is a schematic diagram of an electronic device provided according to an embodiment of the present invention. Detailed Implementation

[0051] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0053] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. Example 1

[0054] like Figure 1-2 As shown, this embodiment provides a method for identifying the metabolic concentration of in vivo components, used to identify the reaction color of a multi-well reagent test fluid, and predict the concentration of the fluid in each well based on the color intensity, including:

[0055] S1, Establish a neural network model for identifying and classifying metabolic concentrations of in vivo components;

[0056] S2, obtain multiple reagent reaction colors with different gradients as sample images, and process the sample images to obtain a training dataset corresponding to the in vivo component metabolic concentration recognition and classification neural network model;

[0057] S3. Based on the interaction between the in vivo component metabolic concentration identification classification neural network model and the training dataset, a numerical prediction neural network model for in vivo component metabolic concentration identification is obtained. The last layer of the convolutional layer of the in vivo component metabolic concentration identification classification neural network model is extracted as a feature. The convolutional layer is divided according to different hole positions to extract the features corresponding to each hole. The model weights are convolved with the corresponding features to reduce the data to one-dimensional data. The one-dimensional data with different gradients for different samples is numerically trained and linearly fitted with the corresponding concentration values ​​to obtain weight values. The weight values ​​are used to replace the last layer of the original deep network model, thereby realizing the transformation of the deep neural network model from classification prediction to numerical prediction.

[0058] S4, acquire the real-time reaction color image corresponding to the metabolic concentration of the body component and input it into the numerical prediction neural network model for identifying the metabolic concentration of the body component. The output of the numerical prediction neural network model for identifying the metabolic concentration of the body component is used as the metabolic concentration of the body component.

[0059] As a preferred embodiment, the in vivo component metabolic concentration identification and classification neural network model in S1 is a deep neural network model.

[0060] See Figure 5 As a preferred embodiment, the deep network model includes, but is not limited to, Yolov 8.

[0061] In a preferred embodiment, S2 includes:

[0062] S21, Obtain multiple reagent reaction colors with different gradients as initial sample images;

[0063] S22, perform preprocessing operations on the initial sample image, the preprocessing operations including: performing one or more operations on the initial sample image such as image removal, noise reduction and removal of invalid regions to obtain the sample image to be labeled;

[0064] S23, label the different pores and different colors in the sample image to be labeled with multi-pore index and concentration, and finally form the training dataset.

[0065] In a preferred embodiment, step S3, obtaining a numerical prediction neural network model for identifying metabolic concentrations of in vivo components based on the interaction between the classification neural network model for identifying metabolic concentrations of in vivo components and the training dataset, includes:

[0066] S31, based on the in vivo component metabolic concentration identification and classification neural network model and the preprocessing operation, the training dataset is fine-tuned multiple times to obtain multiple fine-tuning results;

[0067] S32, verify the recognition and classification effect of the fine-tuning result, extract the first key model parameters of the in vivo component metabolic concentration recognition and classification neural network model corresponding to the preset recognition and classification effect, and the second key model parameters of the in vivo component metabolic concentration recognition and classification neural network model in S1; the first key model parameters include the last convolutional layer of the model; the second key model parameters include the activation function layer;

[0068] S33, the last convolutional layer of the model is used as the feature layer, and features of different pores are extracted based on the feature layer. The features of different pores are weighted with the in vivo component metabolic concentration recognition and classification neural network model to obtain multidimensional data; the multidimensional data in the multidimensional data is reduced to one-dimensional data.

[0069] S34, numerical training will be performed on the one-dimensional data of different samples and different gradients, and the weight values ​​will be obtained by linear fitting with their corresponding concentration values.

[0070] S35, based on the weight values ​​and their corresponding relationships, the network is updated, and the weight values ​​are used to replace the activation function layer to form the in vivo component metabolic concentration identification numerical prediction neural network model.

[0071] In a preferred embodiment, the features of the different aperture positions in S33 are the relative position features of each aperture obtained after the last convolutional layer is divided.

[0072] In a preferred embodiment, the numerical training and linear fitting in S34 are linear regression fitting. Of course, those skilled in the art will recognize that other regression fitting methods or more advanced fitting methods can also be used, all of which are within the scope of this invention.

[0073] In a preferred embodiment, S4 includes:

[0074] S41, acquire real-time reaction color images corresponding to the metabolic concentration of in vivo components and input them into the in vivo component metabolic concentration recognition numerical prediction neural network model.

[0075] S42, Based on the reaction color image, identify the reaction color of the body fluid tested by the porous reagent; Based on the color intensity in the identification result and the metabolic concentration of the body components in the body, a numerical prediction neural network model is used to predict the metabolic concentration of the body components in each well;

[0076] S43, output and display the metabolic concentration of the in vivo component in each well. Example 2

[0077] like Figure 3As shown, this embodiment provides a system for identifying the metabolic concentration of in vivo components, used to implement the method for identifying the metabolic concentration of in vivo components in Embodiment 1, including:

[0078] The first model building module 101 is used to build a neural network model for identifying and classifying metabolic concentrations of in vivo components.

[0079] The training dataset acquisition module 102 is used to acquire multiple reagent reaction colors with different gradients as sample images, and to obtain a training dataset corresponding to the in vivo component metabolic concentration recognition and classification neural network model after processing the sample images.

[0080] The second model building module 103 is used to obtain a numerical prediction neural network model for identifying metabolic concentrations of in vivo components based on the interaction between the classification neural network model for identifying metabolic concentrations of in vivo components and the training dataset; it extracts the last layer of the convolutional layer of the classification neural network model for identifying metabolic concentrations of in vivo components as a feature, divides the convolutional layer according to different hole positions and extracts the features corresponding to each hole, convolves the model weights with the corresponding features to reduce the data to one-dimensional data, performs numerical training on the one-dimensional data with different gradients for different samples, and performs linear fitting with the corresponding concentration values ​​to obtain weight values; it replaces the last layer of the original deep network model with the weight values, thereby realizing the transformation of the deep neural network model from classification prediction to numerical prediction;

[0081] The metabolic concentration recognition module 104 is used to acquire real-time reaction color images corresponding to the metabolic concentration of in vivo components and input them into the in vivo component metabolic concentration recognition numerical prediction neural network model. The output of the in vivo component metabolic concentration recognition numerical prediction neural network model is used as the metabolic concentration of in vivo components.

[0082] Application Examples:

[0083] This embodiment mainly uses a multi-well reagent in the urine six-item test to give the concentration value of each well based on the color intensity of the reagent reaction in each well.

[0084] The specific testing steps are as follows:

[0085] (1) Training sample preparation: Prepare reagent samples of different gradients for each well and obtain training sample images through an optical detection system, such as... Figure 4 As shown:

[0086] (2) Perform training preprocessing on the training sample images and label the image data with labelme;

[0087] (3) Select, but not limited to, a Yolov8 deep network model, fine-tune the training on the labeled training data, and save the model; such as Figure 5 As shown.

[0088] (4) Extract the last convolutional layer of the model and divide it to take the relative position features of each well as the features of this sample and this well. After being combined with the model weights, it is reduced to one-dimensional data. This data is then regressed and fitted with the concentration corresponding to this sample and this well to obtain its weight.

[0089] (5) Replace the last activation function layer in the model in step (3) with the weights obtained in step (4) and save the model. The model can be converted from a classification model to a numerical prediction model.

[0090] The present invention also provides a memory that stores multiple instructions for implementing the method as described in Embodiment 1.

[0091] like Figure 6 As shown, the present invention also provides an electronic device, including a processor 301 and a memory 302 connected to the processor 301. The memory 302 stores a plurality of instructions, which can be loaded and executed by the processor to enable the processor to perform the method as described in Embodiment 1.

[0092] Through the above description of the embodiments, those skilled in the art can clearly understand that the above embodiments can be implemented by software, or by using software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions of the above embodiments can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, mobile hard drive, etc.), including several instructions to cause a computer device (such as a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0093] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying the metabolic concentration of in vivo components, characterized in that, include: S1, Establish a neural network model for identifying and classifying metabolic concentrations of in vivo components; S2, obtain multiple reagent reaction colors with different gradients as sample images, and process the sample images to obtain a training dataset corresponding to the in vivo component metabolic concentration recognition and classification neural network model; S3, obtaining a numerical prediction neural network model for identifying metabolic concentrations of in vivo components based on the interaction between the classification neural network model for identifying metabolic concentrations of in vivo components and the training dataset, includes: extracting the last layer of the convolutional layer of the classification neural network model for identifying metabolic concentrations of in vivo components as a feature; dividing the convolutional layer according to different hole positions to extract the features corresponding to each hole; convolving the model weights with the corresponding features to reduce the data to one-dimensional data; numerically training the one-dimensional data with different gradients for different samples and linearly fitting it with the corresponding concentration values ​​to obtain weight values; replacing the last layer of the classification neural network model for identifying metabolic concentrations of in vivo components with the weight values, thereby realizing the transformation of the classification neural network model for identifying metabolic concentrations of in vivo components from classification prediction to numerical prediction; S4, acquire the real-time reaction color image corresponding to the metabolic concentration of the body component and input it into the numerical prediction neural network model for identifying the metabolic concentration of the body component. The output of the numerical prediction neural network model for identifying the metabolic concentration of the body component is used as the metabolic concentration of the body component.

2. The method for identifying the metabolic concentration of in vivo components according to claim 1, characterized in that, The in vivo component metabolic concentration identification and classification neural network model in S1 is a deep neural network model.

3. The method for identifying the metabolic concentration of in vivo components according to claim 2, characterized in that, S2 includes: S21, Obtain multiple reagent reaction colors with different gradients as initial sample images; S22, perform preprocessing operations on the initial sample image, the preprocessing operations including: performing one or more operations on the initial sample image such as image removal, noise reduction and removal of invalid regions to obtain the sample image to be labeled; S23, label the different pores and different colors in the sample image to be labeled with multi-pore index and concentration, and finally form the training dataset.

4. The method for identifying the metabolic concentration of in vivo components according to claim 3, characterized in that, S3, obtaining a numerical prediction neural network model for identifying metabolic concentrations of in vivo components based on the interaction between the in vivo component metabolic concentration identification classification neural network model and the training dataset, includes: S31, based on the in vivo component metabolic concentration identification and classification neural network model and the preprocessing operation, the training dataset is fine-tuned multiple times to obtain multiple fine-tuning results; S32, verify the recognition and classification effect of the fine-tuning result, extract the first key model parameters of the in vivo component metabolic concentration recognition and classification neural network model corresponding to the preset recognition and classification effect, and the second key model parameters of the in vivo component metabolic concentration recognition and classification neural network model in S1; the first key model parameters include the last convolutional layer of the model; the second key model parameters include the activation function layer; S33, the last convolutional layer of the model is used as the feature layer, and features of different pores are extracted based on the feature layer. The features of different pores are weighted with the in vivo component metabolic concentration recognition and classification neural network model to obtain multidimensional data; the multidimensional data in the multidimensional data is reduced to one-dimensional data. S34, The weight values ​​are obtained by numerical training and linear fitting of the one-dimensional data; S35, based on the weight values ​​and their corresponding relationships, the network is updated, and the weight values ​​are used to replace the activation function layer to form the in vivo component metabolic concentration identification numerical prediction neural network model.

5. The method for identifying the metabolic concentration of in vivo components according to claim 4, characterized in that, The features of different aperture positions in S33 are the relative position features of each aperture obtained after the last convolutional layer is divided.

6. The method for identifying the metabolic concentration of in vivo components according to claim 5, characterized in that, The numerical training and linear fitting in S34 is linear regression fitting.

7. The method for identifying the metabolic concentration of in vivo components according to claim 6, characterized in that, S4 includes: S41, acquire real-time reaction color images corresponding to the metabolic concentration of in vivo components and input them into the in vivo component metabolic concentration recognition numerical prediction neural network model. S42, Based on the reaction color image, identify the reaction color of the body fluid tested by the porous reagent; Based on the color intensity in the identification result and the metabolic concentration of the body components in the body, a numerical prediction neural network model is used to predict the metabolic concentration of the body components in each well; S43, output and display the metabolic concentration of the in vivo component in each well.

8. A system for identifying the metabolic concentration of in vivo components, used to implement the method for identifying the metabolic concentration of in vivo components according to any one of claims 1-7, characterized in that, include: The first model building module (101) is used to build a neural network model for identifying and classifying metabolic concentrations of in vivo components. The training dataset acquisition module (102) is used to acquire multiple reagent reaction colors with different gradients as sample images, and to obtain a training dataset corresponding to the in vivo component metabolic concentration identification and classification neural network model after processing the sample images. The second model building module (103) is used to obtain a numerical prediction neural network model for identifying metabolic concentrations of in vivo components based on the interaction between the classification neural network model for identifying metabolic concentrations of in vivo components and the training dataset. This includes: extracting the last layer of the convolutional layer of the classification neural network model for identifying metabolic concentrations of in vivo components as a feature; dividing the convolutional layer according to different hole positions to extract the features corresponding to each hole; convolving the model weights with the corresponding features to reduce the data to one-dimensional data; numerically training the one-dimensional data with different gradients for different samples and linearly fitting it with their corresponding concentration values ​​to obtain weight values; and replacing the last layer of the classification neural network model for identifying metabolic concentrations of in vivo components with the weight values, thereby realizing the transformation of the classification neural network model for identifying metabolic concentrations of in vivo components from classification prediction to numerical prediction. The metabolic concentration recognition module (104) is used to acquire real-time reaction color images corresponding to the metabolic concentration of in vivo components and input them into the in vivo component metabolic concentration recognition numerical prediction neural network model. The output of the in vivo component metabolic concentration recognition numerical prediction neural network model is used as the metabolic concentration of in vivo components.

9. An electronic device, characterized in that, It includes a processor and a memory, the memory storing multiple instructions, and the processor being used to read the instructions and execute the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions, which can be read by a processor and executed as described in any one of claims 1-7.

Images