Antigen detection data management system based on machine learning and big data

Abstract

The application discloses an antigen detection data management system based on machine learning and big data, relates to the technical field of antigen detection, and comprises a data acquisition module, a data storage module, an image analysis module and a verification module; the data acquisition module is used for acquiring image data of an antigen detection card and illumination intensity information of a shooting location; the data storage module is used for storing image data of the antigen detection card with a time stamp; the image analysis module judges an antigen detection result based on color features of the antigen detection card; and the verification module is used for acquiring color features of C lines and T lines from the start of chromatography to the completion of chromatography; the color features under different illumination conditions are connected through the chromatography degree of the C lines and the T lines, time stamp analysis is performed on image data under a condition without light interference, output data under the condition with light interference is obtained, a model under the condition with light interference is obtained through training of the output data, and the accuracy of antigen detection of a machine is improved.

Description

Technical Field

[0001] This invention relates to the field of antigen detection technology, specifically to an antigen detection data management system based on machine learning and big data. Background Technology

[0002] Antigen test strips are test strips used to detect viral antigens, commonly used for rapid detection of the presence of a certain virus in the human body, such as the novel coronavirus. Automatic machine recognition and result reading of antigen test strips can quickly improve testing efficiency. However, the effectiveness of machine recognition is affected by the image of the antigen test strip; the intensity of light affects the display of the image. Machine recognition models are typically trained under stable lighting conditions, making it difficult to handle antigen detection tasks with light interference. Therefore, improving the versatility of machine recognition and enhancing its accuracy under lighting conditions has become a pressing issue. Summary of the Invention

[0003] The purpose of this invention is to provide an antigen detection data management system based on machine learning and big data to solve the problems raised in the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: an antigen detection data management system based on machine learning and big data, comprising a data acquisition module, a data storage module, an image analysis module, and a verification module; the output of the data acquisition module is connected to the input of the data storage module and the verification module, for acquiring image data of the antigen detection card and light intensity information of the shooting location; the output of the data storage module is connected to the input of the image analysis module, for storing image data of the antigen detection card with timestamps; the image analysis module judges the antigen detection result based on the color characteristics of the antigen detection card; the output of the verification module is connected to the input of the image analysis module, for performing timestamp verification on the antigen detection card under the light intensity of the shooting location, recording the image data of the C-line and T-line from the start of chromatography to the completion of chromatography, and acquiring the color characteristics of the C-line and T-line from the start of chromatography to the completion of chromatography.

[0005] Specifically, the data acquisition module further includes: an image acquisition unit, a photometric measurement unit, and a timer unit; the image acquisition unit is used to acquire image data of the antigen test card; the photometric measurement unit is used to acquire light intensity information of the shooting location; and the timer unit is used to record the timestamp of the antigen test card image data.

[0006] Specifically, the image analysis module further includes a color space conversion unit, a localization unit, a model training unit, an interference analysis unit, and an antigen detection unit. The color space conversion unit converts the image data of the antigen test card to a color space to obtain the color features of the antigen test card. The localization unit, based on the model information of the antigen test card, obtains the color features of the C-line and T-line at corresponding positions on the antigen test card. The model training unit trains a general model and a supplementary model. The general model establishes the relationship between light intensity, color features, and the tomography degree of the C-line and T-line. The supplementary model, using the general model to obtain the tomography degree of the C-line and T-line, establishes the relationship between light intensity, color features, and the tomography degree of the C-line and T-line at interference locations. The interference analysis unit determines the interference locations on the antigen test card. The antigen detection unit compares the color features of the C-line and T-line with the antigen detection criteria to determine the antigen detection result.

[0007] Specifically, the antigen detection unit acquires the color characteristics of the C-line and T-line through the following steps:

[0008] The process involves determining the extent to which the C-line or T-line on the antigen test card is covered by interference. If the C-line or T-line is not covered by interference, the chromatographic extent of the C-line or T-line is determined using a standard model. Once the C-line or T-line chromatography is complete, the antigen detection result of the C-line or T-line is verified. If the C-line or T-line is partially covered by interference, the chromatographic extent of the uncovered areas of the C-line or T-line is determined using a standard model and used as the chromatographic extent of the C-line or T-line. Once the C-line or T-line chromatography is complete, the antigen detection result of the C-line or T-line is verified.

[0009] Specifically, the determination of whether to use C-line or T-line antigen detection is based on the following steps:

[0010] If the C-line or T-line is completely covered by interference, the degree of chromatography at the covered area is obtained through a supplementary model. When the C-line or T-line chromatography is completed, the antigen detection results of the C-line or T-line are verified. The verification of the antigen detection results of the C-line or T-line also includes the following steps: obtaining the color feature data of the C-line or T-line when the C-line or T-line chromatography is completed through a supplementary model and recording it as reference data, and using the reference data as the basis for antigen detection.

[0011] If the C-line or T-line is not completely covered by the interference area, the color characteristics of the C-line and T-line at the time of chromatography are obtained through the verification module as the basis for antigen detection.

[0012] Specifically, the model training unit determines the lighting conditions of the C-line and T-line based on their color features. If there is no interference between the lighting of the C-line and T-line, a standard model is used to obtain the tomography of the C-line and T-line. If there is interference between the lighting of the C-line and T-line, a supplementary model is used at the location of the interference to obtain the tomography of the C-line and T-line.

[0013] Specifically, the model training unit trains a standard model through the following steps:

[0014] Acquire the light intensity information of the location where the antigen test card was photographed; perform timestamp verification on the antigen test card under the light intensity of the location, record the shooting time of the antigen test card, record the image data of the C-line and T-line from the start of chromatography to the completion of chromatography, and acquire the color features of the C-line and T-line from the start of chromatography to the completion of chromatography; quantify the degree of chromatography according to the chromatography time, and divide the image data into training set and test set; use the light intensity information of the location where the antigen test card was photographed and the color features of the C-line and T-line in the training set as input, and the degree of chromatography as output, train a general model through a machine learning model, and verify the general model through the image data in the test set.

[0015] Specifically, the model training unit trains the supplementary model through the following steps:

[0016] In the image data of antigen test cards where the light from the C-line and T-line is interfered with, image data of antigen test cards where the C-line or T-line is not completely covered by the interference is obtained; using a standard model, the light intensity and color features of the uninterrupted areas of the C-line or T-line are input into the standard model, and the tomography degree of the C-line or T-line is obtained from the output of the standard model.

[0017] Image data of antigen test cards with the C-line or T-line not completely covered by interference are divided into training set and test set. In the training set, the light intensity and color features of the C-line or T-line covered by interference are used as inputs, and the tomography of the C-line or T-line is obtained from the output of the ordinary model as the output. A supplementary model is trained by a machine learning model. The supplementary model is validated by the image data of the test set.

[0018] Specifically, the interference analysis unit determines the location of the interference through the following steps:

[0019] In the image data of the antigen test card where the light from the C-line and T-line is interfering, the color feature data of the pixels is obtained, and the pixels whose color feature data deviates from the normal range are identified. The closed area formed by the identified pixels is the interference location of the antigen test card. The part of the interference location of the antigen test card that coincides with the C-line and T-line is the interference location of the C-line and T-line.

[0020] Specifically, identifying pixels whose color feature data deviates from the normal range also includes the following steps:

[0021] For the i-th pixel in the image data of the antigen test card, find the other pixel that is closest to the i-th pixel; and calculate the difference in color feature data between the i-th pixel and the other pixel that is closest, with the difference in absolute value form.

[0022] Obtain the difference data of all pixels in the image data of the antigen test card, sort all the difference data of all pixels in ascending order, with one position number corresponding to two pixels, and use the position number of the difference in the sorted data as the independent variable and the difference itself as the dependent variable to fit the data and obtain the fitting function.

[0023] The first derivative of the fitted function is obtained by taking its first derivative. Based on this, the maximum value of the difference and its corresponding position number are determined. Two pixels are obtained from these position numbers, each corresponding to a color feature data point. The average of these two color feature data points is used to determine the boundary between the normal range and the interference position. If the light intensity at the interference position is greater than that at the non-interference position, the pixel whose color feature data is greater than the boundary point is considered to be outside the normal range. Conversely, if the light intensity at the interference position is less than that at the non-interference position, the pixel whose color feature data is less than the boundary point is considered to be outside the normal range.

[0024] Compared with the prior art, the beneficial effects of the present invention are: analyzing the tomography degree of C-line and T-line to obtain the antigen detection basis in complete tomography; connecting the color features under different lighting conditions by the tomography degree of C-line and T-line, it is only necessary to perform time stamp analysis on image data under conditions without light interference to obtain output data under conditions with light interference, and use the obtained output data to train a model under conditions with light interference, thereby improving the accuracy and versatility of the machine for antigen detection. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the antigen detection data management system based on machine learning and big data according to the present invention. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. 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.

[0027] Example: Figure 1As shown, this invention provides a technical solution: an antigen detection data management system based on machine learning and big data, including a data acquisition module, a data storage module, an image analysis module, and a verification module. The output of the data acquisition module is connected to the input of the data storage module and the verification module, and is used to acquire image data of the antigen detection card and light intensity information of the shooting location. The output of the data storage module is connected to the input of the image analysis module, and is used to store image data of the antigen detection card with timestamps. The image analysis module judges the antigen detection result based on the color characteristics of the antigen detection card. The output of the verification module is connected to the input of the image analysis module, and is used to perform timestamp verification on the antigen detection card under the light intensity of the shooting location, record the image data of the C-line and T-line from the start of chromatography to the completion of chromatography, and acquire the color characteristics of the C-line and T-line from the start of chromatography to the completion of chromatography.

[0028] The data acquisition module further includes: an image acquisition unit, a photometric measurement unit, and a timer unit; the image acquisition unit is used to acquire image data of the antigen test card; the photometric measurement unit is used to acquire light intensity information of the shooting location; and the timer unit is used to record the timestamp of the antigen test card image data.

[0029] The image analysis module further includes a color space conversion unit, a localization unit, a model training unit, an interference analysis unit, and an antigen detection unit. The color space conversion unit converts the image data of the antigen test card to a color space to obtain the color features of the antigen test card. The localization unit, based on the model information of the antigen test card, obtains the color features of the C-line and T-line at corresponding positions on the antigen test card. The model training unit trains a standard model and a supplementary model. The standard model establishes the relationship between light intensity, color features, and the tomography degree of the C-line and T-line. The supplementary model, using the standard model, obtains the tomography degree of the C-line and T-line and establishes the relationship between light intensity, color features, and the tomography degree of the C-line and T-line at interference locations. The interference analysis unit determines the interference locations on the antigen test card. The antigen detection unit compares the color features of the C-line and T-line with the antigen detection criteria to determine the antigen detection result.

[0030] Optionally, taking grayscale values ​​as an example, the image data of the antigen test card is converted to the RGB color space. The grayscale value of each pixel is obtained based on the red, green, and blue channel values ​​in the RGB color space, and this grayscale value is used as the color feature of the C-line and T-line. Changes in light intensity directly affect the grayscale value of the image; the stronger the light, the higher the grayscale value; the weaker the light, the lower the grayscale value. This variation leads to instability in the color features, affecting the machine's judgment of the antigen detection results. Alternatively, the image can be converted to HSV, Lab, or other color spaces, and other color features can be selected.

[0031] The process involves determining the extent to which the C-line or T-line on the antigen test card is covered by interference. If the C-line or T-line is not covered by interference, the chromatographic extent of the C-line or T-line is determined using a standard model. Once the C-line or T-line chromatography is complete, the antigen detection result of the C-line or T-line is verified. If the C-line or T-line is partially covered by interference, the chromatographic extent of the uncovered areas of the C-line or T-line is determined using a standard model and used as the chromatographic extent of the C-line or T-line. Once the C-line or T-line chromatography is complete, the antigen detection result of the C-line or T-line is verified.

[0032] The light intensity at the location where the antigen test card is taken can be tested using photoelectric measurement. By placing a photodiode at the location and measuring the induced current of the photodiode, the light intensity at the interference-free and interference-free locations can be obtained. Only the color characteristics of the C-line or T-line under interference-free conditions are validated and analyzed to obtain a standard model. For randomly generated interference, a supplementary model is generated based on the standard model without additional validation analysis.

[0033] The determination of whether to use C-line or T-line antigen detection is based on the following steps:

[0034] If the C-line or T-line is completely covered by interference, the degree of chromatography at the covered area is obtained through a supplementary model. When the C-line or T-line chromatography is completed, the antigen detection results of the C-line or T-line are verified. The verification of the antigen detection results of the C-line or T-line also includes the following steps: obtaining the color feature data of the C-line or T-line when the C-line or T-line chromatography is completed through a supplementary model and recording it as reference data, and using the reference data as the basis for antigen detection.

[0035] If the C-line or T-line is not completely covered by the interference area, the color characteristics of the C-line and T-line at the time of chromatography are obtained through the verification module as the basis for antigen detection.

[0036] Both the standard model and the supplementary model are trained for the conditions of C-line validity and T-line positivity. The color characteristics of the chromatography completed under different lighting conditions are used as the basis for antigen detection. The ultimate goal is to obtain the color characteristics of the chromatography completed under different lighting conditions and to make judgments based on different antigen detection criteria under different lighting conditions, thus avoiding the interference of different lighting conditions on antigen detection.

[0037] Specifically, the model training unit determines the lighting conditions of the C-line and T-line based on their color features. If there is no interference between the lighting of the C-line and T-line, a standard model is used to obtain the tomography of the C-line and T-line. If there is interference between the lighting of the C-line and T-line, a supplementary model is used at the location of the interference to obtain the tomography of the C-line and T-line.

[0038] The model training unit trains a standard model through the following steps:

[0039] Acquire the light intensity information of the location where the antigen test card was photographed; perform timestamp verification on the antigen test card under the light intensity of the location, record the shooting time of the antigen test card, record the image data of the C-line and T-line from the start of chromatography to the completion of chromatography, and acquire the color features of the C-line and T-line from the start of chromatography to the completion of chromatography; quantify the degree of chromatography according to the chromatography time, and divide the image data into training set and test set; use the light intensity information of the location where the antigen test card was photographed and the color features of the C-line and T-line in the training set as input, and the degree of chromatography as output, train a general model through a machine learning model, and verify the general model through the image data in the test set.

[0040] The model training unit trains and supplements the model through the following steps:

[0041] In the image data of antigen test cards where the light from the C-line and T-line is interfered with, image data of antigen test cards where the C-line or T-line is not completely covered by the interference is obtained; using a standard model, the light intensity and color features of the uninterrupted areas of the C-line or T-line are input into the standard model, and the tomography degree of the C-line or T-line is obtained from the output of the standard model;

[0042] Image data of antigen test cards with the C-line or T-line not completely covered by interference are divided into training set and test set. In the training set, the light intensity and color features of the C-line or T-line covered by interference are used as inputs, and the tomography of the C-line or T-line is obtained from the output of the ordinary model as the output. A supplementary model is trained by a machine learning model. The supplementary model is validated by the image data of the test set.

[0043] For C-line or T-line, although the color features of the image data will differ under different lighting conditions, there is an invariant: the tomography degree of C-line or T-line. Taking C-line as an example, if one part of C-line is under normal lighting conditions and another part is affected and occluded, the color features of these two parts will differ, but the tomography degree of these two parts will be the same. Using the tomography degree as a bridge, we can verify the relationship between color features and tomography degree under normal lighting conditions, thus obtaining a general model. Since there is a general model, for the image data of the antigen test card where the C-line is not completely covered by interference, the tomography degree of C-line can be obtained using the general model. With the tomography degree as the output, and with the input of lighting intensity and color features, we can then train a supplementary model using a machine learning model, without needing to perform additional verification at the interference locations. In this way, even if various interferences occur during the antigen test card imaging process, output data can be automatically generated for training.

[0044] The degree of tomography can be quantified based on the tomography time. For example, under the current environmental conditions, it takes 15 minutes to completely tomographically analyze the C line. Then, t / 15 is taken as the degree of tomography, where t represents the time that has been tomographically analyzed. If the environmental conditions change, only the tomography speed is affected, not the relationship between color features and the degree of tomography. There is no need to retrain the model.

[0045] Since the data is image data, model training can use a convolutional neural network model, which includes the following steps:

[0046] Call the neural network toolkit and define the package parameters:

[0047] Define the network structure as follows: in order of connection, it includes an input layer, a first fully connected layer, a convolutional layer, a pooling layer, a second fully connected layer, and an output layer.

[0048] The network weights and biases are randomly initialized;

[0049] Choose a loss function for the model, including but not limited to root mean square error or cross-entropy loss function, and select the best one through testing; select the ADM solver; set the learning rate and forgetting rate.

[0050] Determine the dataset:

[0051] 70% of the data is divided into the training set and 30% into the test set;

[0052] The model is trained using a neural network toolkit; if the model overfits, a dropout layer can be added to the network structure; based on the validation results on the test set, the network structure, learning rate, and forgetting rate can be adjusted.

[0053] Specifically, the interference analysis unit determines the location of the interference through the following steps:

[0054] In the image data of the antigen test card where the light from the C-line and T-line is interfering, the color feature data of the pixels is obtained, and the pixels whose color feature data deviates from the normal range are identified. The closed area formed by the identified pixels is the interference location of the antigen test card. The part of the interference location of the antigen test card that coincides with the C-line and T-line is the interference location of the C-line and T-line.

[0055] Specifically, identifying pixels whose color feature data deviates from the normal range also includes the following steps:

[0056] For the i-th pixel in the image data of the antigen test card, find the other pixel that is closest to the i-th pixel; and calculate the difference in color feature data between the i-th pixel and the other pixel that is closest, with the difference in absolute value form.

[0057] Obtain the difference data of all pixels in the image data of the antigen test card, sort all the difference data of all pixels in ascending order, with one position number corresponding to two pixels, and use the position number of the difference in the sorted data as the independent variable and the difference itself as the dependent variable to fit the data and obtain the fitting function.

[0058] The first derivative of the fitted function is obtained by taking its first derivative. Based on this, the maximum value of the difference and its corresponding position number are determined. Two pixels are obtained from these position numbers, each corresponding to a color feature data point. The average of these two color feature data points is used to determine the boundary between the normal range and the interference position. If the light intensity at the interference position is greater than that at the non-interference position, the pixel whose color feature data is greater than the boundary point is considered to be outside the normal range. Conversely, if the light intensity at the interference position is less than that at the non-interference position, the pixel whose color feature data is less than the boundary point is considered to be outside the normal range.

[0059] There are two types of interference with antigen test cards. The first is when the card is covered by a shadow from an obstruction. In this case, the grayscale value at the interference location will decrease, resulting in two different grayscale values ​​at the interference and normal locations. The difference in grayscale values ​​between the two types of pixels reaches its maximum at the boundary point. Within the interference and normal locations, the difference in grayscale values ​​between pixels is smaller. Therefore, the boundary point is determined based on the difference, thus identifying the interference location. The second type is when the antigen test card is illuminated by an external light source, such as sunlight shining through a gap. In this case, the grayscale value at the interference location will increase.

[0060] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. An antigen detection data management system based on machine learning and big data, characterized by, The application relates to an antigen detection device, which comprises a data acquisition module, a data storage module, an image analysis module and a verification module; the output end of the data acquisition module is connected with the input ends of the data storage module and the verification module, which are used for acquiring image data of an antigen detection card and illumination intensity information of a shooting site; the output end of the data storage module is connected with the input end of the image analysis module, which is used for storing image data of the antigen detection card with a time stamp; the image analysis module judges an antigen detection result based on color features of the antigen detection card; the output end of the verification module is connected with the input end of the image analysis module, which is used for time stamp verification of the antigen detection card under the illumination intensity of the shooting site, recording image data of C lines and T lines from starting chromatography to completion of chromatography, and acquiring color features of the C lines and the T lines from the starting chromatography to the completion of the chromatography. The image analysis module further comprises a color space conversion unit, a positioning unit, a model training unit, an interference analysis unit and an antigen detection unit. The interference analysis unit determines the interference position through the following steps: In the image data of the antigen detection card with light interference of the C lines and the T lines, color feature data of pixel points are acquired, and the pixel points deviating from a normal range are determined; a closed region formed by the determined pixel points is the interference position of the antigen detection card; a part of the interference position of the antigen detection card coinciding with the C lines and the T lines is the interference position of the C lines and the T lines. The step of determining the pixel points deviating from the normal range further comprises the following steps: For the i-th pixel point in the image data of the antigen detection card, the other pixel point with the minimum distance from the i-th pixel point is found; and the difference value of the color feature data between the i-th pixel point and the other pixel point with the minimum distance is calculated, and the difference value is in the form of an absolute value; The difference value data of all the pixel points in the image data of the antigen detection card are obtained, the difference value data of all the pixel points are arranged in ascending order, one position sequence number corresponds to two pixel points, the position sequence number in the arrangement is taken as the independent variable, and the difference value itself is taken as the dependent variable, fitting is conducted, and a fitting function is obtained; A first derivative function is obtained by taking the first derivative of the fitting function; the maximum value of the difference value and the position sequence number corresponding to the maximum value are determined according to the first derivative function, the two pixel points are obtained according to the position sequence number, the two pixel points correspond to two color feature data; the average value of the two color feature data is taken as the demarcation point between the normal range and the interference position; if the illumination intensity at the interference position is greater than the illumination intensity at the non-interference position, the pixel point with the color feature data greater than the demarcation point is the pixel point deviating from the normal range; if the illumination intensity at the interference position is less than the illumination intensity at the non-interference position, the pixel point with the color feature data less than the demarcation point is the pixel point deviating from the normal range. 2.The machine learning and big data based antigen detection data management system of claim 1, wherein, The data acquisition module further comprises an image acquisition unit, a photometric measurement unit and a timer unit; the image acquisition unit is used for acquiring image data of an antigen detection card; the photometric measurement unit is used for acquiring illumination intensity information of a shooting site; and the timer unit is used for recording a time stamp of the image data of the antigen detection card. 3.The machine learning and big data based antigen detection data management system of claim 2, wherein, The color space conversion unit is configured to convert image data of the antigen detection card to a color space to obtain color features of the antigen detection card; the positioning unit is configured to obtain color features of C lines and T lines from corresponding positions of the C lines and the T lines on the antigen detection card according to model information of the antigen detection card; the model training unit is configured to train a general model and a supplementary model; the general model is configured to establish a relationship among illumination intensity, color features, and chromatographic degrees of the C lines and the T lines; the supplementary model is configured to obtain the chromatographic degrees of the C lines and the T lines by the general model and establish a relationship among the illumination intensity, the color features, and the chromatographic degrees of the C lines and the T lines at an interference position; the interference analysis unit is configured to determine an interference position on the antigen detection card; and the antigen detection unit is configured to compare the color features of the C lines and the T lines with antigen detection criteria to determine an antigen detection result. 4.The machine learning and big data based antigen detection data management system of claim 3, wherein, The antigen detection unit obtains the color features of the C lines and the T lines by the following steps: If the C lines or the T lines are not covered by the interference position, the chromatographic degrees of the C lines or the T lines are obtained by the general model; when the chromatography of the C lines or the T lines is completed, the antigen detection result of the C lines or the T lines is verified; if the C lines or the T lines are partially covered by the interference position, the chromatographic degrees of the C lines or the T lines that are not covered by the interference position are obtained by the general model as the chromatographic degrees of the C lines or the T lines; when the chromatography of the C lines or the T lines is completed, the antigen detection result of the C lines or the T lines is verified. 5.The machine learning and big data based antigen detection data management system of claim 4, wherein, The antigen detection criteria of the C lines or the T lines are determined by the following steps: If the C lines or the T lines are completely covered by the interference position, the chromatographic degrees of the C lines or the T lines that are covered by the interference position are obtained by the supplementary model, and when the chromatography of the C lines or the T lines is completed, the antigen detection result of the C lines or the T lines is verified; the verification of the antigen detection result of the C lines or the T lines further includes the following steps: the color feature data of the C lines or the T lines when the chromatography is completed are obtained by the supplementary model and recorded as reference data, and the reference data are used as the antigen detection criteria; If the C lines or the T lines are not completely covered by the interference position, the color features of the C lines and the T lines when the chromatography is completed are obtained by the verification module as the antigen detection criteria. 6.The machine learning and big data based antigen detection data management system of claim 5, wherein, The model training unit determines the light conditions of the C lines and the T lines based on the color features of the C lines and the T lines; if there is no interference in the light of the C lines and the T lines, the chromatographic degrees of the C lines and the T lines are obtained by the general model; if there is interference in the light of the C lines and the T lines, the chromatographic degrees of the C lines and the T lines at the interference position are obtained by the supplementary model. 7.The machine learning and big data based antigen detection data management system of claim 6, wherein, The model training unit trains the general model by the following steps: The light intensity information of the antigen detection card shooting location is acquired; the antigen detection card is subjected to time stamp verification under the light intensity of the shooting location, the shooting time of the antigen detection card is recorded, the image data of the C line and the T line from the start of chromatography to the completion of chromatography is recorded, the color characteristics of the C line and the T line from the start of chromatography to the completion of chromatography are acquired; the chromatography degree is quantified according to the chromatography time, and the image data is divided into a training set and a test set; the light intensity information of the antigen detection card shooting location and the color characteristics of the C line and the T line in the training set are taken as inputs, and the chromatography degree is taken as output, and a general model is trained through a machine learning model, and the general model is verified through the image data in the test set. 8.The machine learning and big data based antigen detection data management system of claim 7, wherein, The model training unit trains the supplementary model through the following steps: In the image data of the antigen detection card in which the light of the C line and the T line is interfered, the image data of the antigen detection card in which the non-interfered position of the C line or the T line is completely covered is acquired; the light intensity and the color characteristics of the non-interfered position of the C line or the T line are input into the general model through the general model, and the chromatography degree of the C line or the T line is obtained from the output result of the general model; The image data of the antigen detection card in which the non-interfered position of the C line or the T line is completely covered is divided into a training set and a test set; in the data of the training set, the light intensity and the color characteristics of the interfered position of the C line or the T line are taken as inputs, and the chromatography degree of the C line or the T line is taken as output from the output result of the general model, and a supplementary model is trained through a machine learning model; the supplementary model is verified through the image data of the test set.

Images