An intelligent body component metabolism detection system

Abstract

The present application relates to an intelligent body component metabolism detection system, which belongs to the technical field of medical systems. The system includes a mobile terminal, a detection device and a server; the mobile terminal is used to input and store the patient's characteristic information; the detection device is bound to the mobile terminal, and is used to obtain the detection data after the body fluid to be tested is incubated at a constant temperature for a specified time and upload the detection data to the server; the server is configured to: obtain the detection data; obtain the index concentration value corresponding to the body fluid to be tested after processing the detection data based on a preset analysis model; obtain the analysis result after processing the index concentration value, the historical detection data of the same patient at different detection times, and the patient's characteristic information based on a preset evaluation model; send the analysis result to the mobile terminal; the mobile terminal is used to obtain and display the analysis result. The present application can obtain the analysis result by integrating the detection data, historical detection data and characteristic information, so that the obtained analysis result is more comprehensive and accurate.

Description

Technical Field

[0001] The present application relates to the technical field of medical systems, and in particular to an intelligent body component metabolism detection system. Background Art

[0002] In vivo component metabolism detection technology can be used to detect indicators of chronic diseases such as medical weight loss, sarcopenia, hypertension, and intestinal barrier dysfunction.

[0003] In the related technology, when conducting metabolic testing of body components, a single test indicator can be tested by collecting samples from patients through testing equipment, and a single analysis result can be issued. Since a single analysis result can only analyze a single sample collected, the resulting report lacks systematicity. In order to make up for the lack of systematicity, doctors can review past analysis results and then make a more comprehensive analysis and judgment based on personal experience. Since doctors are highly subjective when making judgments based on personal experience, there will also be problems with inaccurate diagnostic results. Not only that, patients need to go to the hospital for testing every time, which is a huge waste of social manpower, financial resources and material resources under the premise of limited medical resources. Summary of the invention

[0004] In order to make the analysis results more comprehensive and accurate, the present application provides an intelligent body component metabolism detection system.

[0005] The present application provides an intelligent body component metabolism detection system. The system includes a mobile terminal, a detection device and a server;

[0006] The mobile terminal is used to input and store the patient's characteristic information;

[0007] The detection device is bound to the mobile terminal and is used to obtain detection data after constant temperature incubation of the body fluid to be detected for a specified time and upload the detection data to the server, wherein the body fluid to be detected includes a colorimetric reagent and a patient's body fluid;

[0008] The server is configured to:

[0009] Acquiring the detection data;

[0010] Processing the detection data based on a preset analysis model to obtain an indicator concentration value corresponding to the body fluid to be tested;

[0011] Based on a preset evaluation model, the index concentration value corresponding to the body fluid to be tested, the historical test data of the same patient at different test times, and the patient's characteristic information are processed to obtain an analysis result;

[0012] Sending the analysis result to the mobile terminal;

[0013] The mobile terminal is used to obtain and display the analysis result.

[0014] By adopting the above technical solution, the detection equipment can incubate the body fluid to be tested at a constant temperature to obtain the detection data, and the server comprehensively analyzes the detection data, historical detection data and characteristic information to obtain the analysis results, so that the analysis results obtained by the patient are more comprehensive and accurate. At the same time, the patient can also intuitively view the analysis results through the mobile terminal, reducing manpower consumption.

[0015] Optionally, the detection device is provided in a plurality of groups, each of the detection devices incubates a plurality of types of body fluids to be tested at a constant temperature, and the concentrations of the colorimetric reagents corresponding to the same type of body fluids to be tested in different detection devices are different.

[0016] By adopting the above technical solution, multiple groups of detection equipment are set up and the concentrations of colorimetric reagents corresponding to the same type of body fluid to be tested in different detection equipment are different, so as to better observe the reaction of patient body fluids in colorimetric reagents of different concentrations and types, and ensure the accuracy of the obtained analysis results.

[0017] Optionally, the server is configured as:

[0018] Acquire multiple groups of body fluid images, each group of body fluid images includes multiple body fluid images of the same type but with different concentrations of color developing reagents;

[0019] Determining the horizontal coordinate of the preset coordinate system according to the concentration of the color developing reagent of the plurality of body fluid images in each group of body fluid images;

[0020] Determine the ordinate in the preset coordinate system according to the color depth of the color developing reagent of the plurality of body fluid images in each group of body fluid images;

[0021] Combining the abscissa and ordinate determined from the same body fluid image into a standard coordinate;

[0022] A plurality of the standard coordinates corresponding to each group of body fluid images are connected in series to obtain a standard curve formula, and each group of body fluid images corresponds to a standard curve formula.

[0023] By adopting the above technical solution, after the standard curve formula is established based on multiple body fluid images in each group of body fluid images, it is convenient to directly call the established standard curve formula in the future, and quickly obtain the index concentration value of each body fluid image through the mapping relationship between the concentration of the color developing reagent contained in the standard curve formula and the gray value. In addition, when establishing the standard curve formula, it is obtained by setting multiple body fluid images, in which the concentration of the color developing reagent in each body fluid image is different, so the accuracy of the established standard curve formula is guaranteed.

[0024] Optionally, the server is further configured to:

[0025] The vertical coordinate in the preset coordinate system is determined according to the color depth of the color developing reagent of the multiple body fluid images in each group of body fluid images, wherein the vertical coordinate of any body fluid image in each group of body fluid images is determined as follows:

[0026] Preprocessing the body fluid image;

[0027] Determining whether the color depth of the preprocessed body fluid image meets a preset uniformity;

[0028] If yes, determining the corresponding ordinate according to the gray value of the body fluid image;

[0029] If not, a computational model is used to predict the vertical coordinate of the body fluid image.

[0030] By adopting the above technical solution, the present application is provided with a plurality of algorithms capable of determining the vertical coordinate of each body fluid image in different scenarios, so as to broaden the application scenarios of the present application.

[0031] Optionally, the server is further configured to:

[0032] The using a computational model to predict the ordinate of the body fluid image comprises:

[0033] Dividing the body fluid image into a plurality of sub-regions according to color depth;

[0034] Dividing the sub-region into different chromaticity levels according to the color depth of the sub-region, where different chromaticity levels correspond to different grayscale values;

[0035] Calculating a plurality of classification probabilities according to the division results of the sub-regions of the body fluid image, wherein one of the classification probabilities corresponds to one of the chromaticity levels, and the classification probability is used to characterize the probability that the body fluid image belongs to the corresponding chromaticity level;

[0036] Select the classification probability with the largest classification probability as the target probability;

[0037] The corresponding ordinate is obtained by matching the gray value corresponding to the target probability.

[0038] By adopting the above technical solution, the server can further analyze the body fluid image, specifically by splitting the body fluid image into multiple sub-areas, and then dividing the sub-areas into different chromaticity levels, and finally selecting the one with the largest classification probability as the target probability to characterize the grayscale value of the body fluid image, thereby ensuring the accuracy of the obtained standard curve formula.

[0039] Optionally, the server is further configured to:

[0040] The calculating and obtaining a plurality of classification probabilities according to the division result of the sub-regions of the body fluid image includes:

[0041] Calculating the area of ​​the sub-region falling within each of the chroma levels;

[0042] The ratio of each area to the total area of ​​the body fluid image is calculated, and the ratio is used as the classification probability that the body fluid image belongs to the chromaticity level.

[0043] Optionally, the server is further configured to:

[0044] Determining the ordinate in the preset coordinate system according to the color depth of the color developing reagent of the multiple body fluid images in each group of body fluid images includes: using a neural network model to predict the ordinate of the body fluid image in the preset coordinate system.

[0045] By adopting the above technical solution, the present application provides a variety of algorithms to predict the vertical coordinate of each body fluid image, which can be selected by patients in different scenarios.

[0046] Optionally, the server is further configured to:

[0047] The determining whether the color depth of the preprocessed body fluid image meets a preset uniformity level includes:

[0048] Calculating grayscale value differences of regions with different color depths in the body fluid image;

[0049] When the maximum value of the gray value difference is less than a first preset value, determining that the color depth of the body fluid image meets a preset uniformity; or

[0050] Counting the number of regions with different color depths in the body fluid image;

[0051] When the number is less than a second preset value, it is determined that the color depth of the body fluid image meets a preset uniformity.

[0052] By adopting the above technical solution, the body fluid image that does not meet the preset uniformity is determined based on the large difference in color depth or severe color zoning, so that the selected algorithm for determining the horizontal coordinate is more adapted to the body fluid image, thereby ensuring the accuracy of the obtained standard curve formula.

[0053] Optionally, the server is further configured to:

[0054] After obtaining the standard curve formula, it includes:

[0055] According to the type of patient body fluid in the body fluid to be tested, the type and concentration of the color developing reagent, using a preset analysis model to match a standard curve formula corresponding to each of the body fluid images from a plurality of standard curve formulas;

[0056] The index concentration value corresponding to the body fluid image is obtained according to the standard curve formula.

[0057] Optionally, the detection device includes a detection card and an incubator;

[0058] The test card is provided with a plurality of test holes for accommodating body fluids to be tested;

[0059] The incubator is provided with a positioning slot, an imaging hole and a heating module, the imaging hole is located at the bottom of the positioning slot, a CMOS camera is provided at the bottom of the imaging hole and the imaging sight line of the CMOS camera is arranged toward the opening end of the imaging hole, and the detection card is connected with the positioning slot;

[0060] The heating module is arranged close to the positioning groove, and is used to perform constant temperature incubation on the body fluid to be tested for a specified time. After the constant temperature incubation for the specified time, the body fluid to be tested is imaged by the CMOS camera to obtain detection data.

[0061] By adopting the above technical solution, different types of colorimetric reagents can be placed in the multiple detection holes on the detection card, and the same patient body fluids can be placed in different colorimetric reagents, thereby achieving the effect of analyzing multiple body fluids to be tested at a single time, which is conducive to a more comprehensive analysis of the detection data.

[0062] In summary, the present application includes at least one of the following beneficial technical effects:

[0063] On the one hand, the detection equipment can incubate the body fluid to be tested at a constant temperature to obtain the detection data, and the server can comprehensively analyze the detection data, historical detection data and characteristic information to obtain the analysis results, so that the analysis results obtained by the patient are more comprehensive and accurate. At the same time, the patient can also intuitively view the analysis results through the mobile terminal to reduce manpower consumption. On the other hand, after the standard curve formula is established based on the multiple body fluid images in each group of body fluid images, it is convenient to directly call the established standard curve formula in the future. Through the mapping relationship between the concentration of the color developing reagent contained in the standard curve formula and the grayscale value, the indicator concentration value of each body fluid image can be quickly obtained, thereby realizing rapid analysis and obtaining analysis results. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 It is a system schematic diagram of the intelligent body component metabolism detection system in an embodiment of the present application.

[0065] Figure 2It is a schematic diagram of the structure of the detection card of the embodiment of the present application.

[0066] Figure 3 It is a schematic diagram of the structure of the incubator of an embodiment of the present application.

[0067] Figure 4 This is an example diagram of classifying body fluid images according to an embodiment of the present application.

[0068] Figure 5 This is an example diagram of an embodiment of the present application in which a plurality of standard coordinates are connected in series to obtain a standard curve formula.

[0069] Explanation of the accompanying drawings: 1. Box body; 2. Positioning slot; 3. Imaging hole; 4. Upper cover; 5. Detection card; 6. Detection hole; 7. Detection equipment; 8. Server; 9. Mobile terminal. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solutions and advantages of this application more clear, the following Figure 1-5 It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0071] The in vivo component metabolism detection technology can be used to detect the indicators of chronic diseases such as medical weight loss, sarcopenia, hypertension, and intestinal barrier dysfunction. In the actual detection process, it is necessary to sample a certain body fluid of the patient and analyze the collected samples. The doctor can give symptomatic treatment to the patient according to the analysis results. However, since the analysis results can only analyze the samples collected once, the resulting report lacks systematicity. Therefore, it is often necessary to regularly monitor the patient's in vivo component metabolism indicators, and then determine the patient's next treatment plan based on the changing trends of various combined indicators within a certain period of time and the clinical significance, combined with the patient's clinical indications. At present, doctors can only analyze and judge based on past analysis results and personal experience. However, different doctors have different experiences and make different judgments, which can easily lead to inaccurate diagnosis results, thereby delaying the patient's condition.

[0072] The embodiment of the present application discloses an intelligent body component metabolism detection system, which can make the analysis results more comprehensive and accurate.

[0073] Reference Figure 1-Figure 3 The intelligent body component metabolism detection system includes a mobile terminal 9, a detection device 7 and a server 8.

[0074] Among them, the detection device 7 is used to bind the mobile terminal 9, and is used to obtain detection data after constant temperature incubation of the body fluid to be tested for a specified time. The body fluid to be tested includes a colorimetric reagent and a patient's body fluid. The patient's body fluid is one of urine, serum, whole blood, cerebrospinal fluid, plasma, and saliva. The display reagent is a reagent such as potassium permanganate, phenolphthalein, etc. that can react with the patient's body fluid and develop color. The detection device 7 is also used to upload the obtained detection data to the server 8.

[0075] The detection device 7 includes two parts: a detection card 5 and an incubator.

[0076] The test card 5 is provided with a plurality of test holes 6 for accommodating the body fluid to be tested. The body fluid of the patient is the same in all the test holes 6 on the same test card 5, but the type of the color developing reagent is different. For example, if there are six test holes 6, the color developing reagent placed in each test hole 6 is different. In order to better observe the reaction of the patient's body fluid in different concentrations and different types of color developing reagents, this example is also provided with eight groups of test cards 5. The type of display reagent placed in each of the eight groups of test cards 5 is the same, but the concentration of the color developing reagent placed in the same position is different. For example, each test hole 6 on the eight groups of test cards 5 is given a number, and the color developing reagent type of the test holes 6 with the same number is the same but the concentration is different. Therefore, when the patient is tested, he needs to drop his own body fluid on the test hole 6.

[0077] The incubator includes a box body 1 and an upper cover 4. The upper cover 4 is covered on the box body 1. A positioning groove 2 is provided in the box body 1 to be engaged with the detection card 5. An imaging hole 3 is provided at the bottom of the positioning groove 2. The opening end of the imaging hole 3 coincides with the bottom of the positioning groove 2. A CMOS camera is provided at the bottom of the imaging hole 3. The imaging line of sight of the CMOS camera is set toward the opening end of the imaging hole 3. The CMOS camera is not shown in the accompanying drawings, and it is assumed that it can be fixedly installed at the bottom of the imaging hole 3. In addition, a heating module is also provided in the incubator. The heating module is arranged close to the positioning groove 2. The heating module, such as a hot air blower, an infrared heating lamp, etc., can realize a device for constant temperature incubation of the body fluid to be tested for a specified time. Similarly, the heating module is not shown in the accompanying drawings.

[0078] After the patient drips the body fluid to be tested into the test card 5, the test card 5 needs to be placed in the positioning slot 2. Usually, one test card 5 occupies one positioning slot 2, and the test card 5 is connected to the positioning slot 2 and the upper cover 4 is closed. At this time, the heating module can incubate the body fluid to be tested at a constant temperature, and the incubation time needs to reach a specified time, which is a time set in advance. After completing the constant temperature incubation, the CMOS camera can image the body fluid to be tested through the imaging hole 3 to obtain the detection data, so the detection data reflects the state of the body fluid to be tested in each detection hole 6. It can be understood that since the types of color developing reagents dripped into different detection holes 6 are different, after constant temperature incubation, the color presented by each body fluid to be tested may be different.

[0079] The server 8 is communicatively connected with the detection device 7. The server 8 is used to use a preset analysis model to process the detection data to obtain the index concentration value corresponding to the body fluid to be tested. The server 8 also uses a preset evaluation model to process the index concentration value corresponding to the body fluid to be tested, the historical detection data of the same patient at different detection times, and the patient's characteristic information to obtain the analysis results, and sends the analysis results to the mobile terminal 9.

[0080] Specifically, the process of server 8 analyzing the detection data is as follows.

[0081] First, the detection data is obtained. The detection data is uploaded to the server 8 by the detection device 7.

[0082] Then, the area where each detection hole 6 contained in the detection data is located is divided to obtain a body fluid image. In this example, eight groups of detection cards 5 are provided, one detection card 5 is a group, and after the eight groups of detection cards 5 obtain six body fluid images corresponding to each other in the above manner, a standard curve formula is established based on the obtained body fluid images. Specifically, the process of establishing the standard curve formula is shown in the following steps S1 to S5.

[0083] Step S1, obtaining multiple groups of body fluid images, each group of body fluid images includes multiple body fluids to be tested of the same type but different concentrations of color developing reagents. As in the above example, assuming that each detection hole 6 on the eight groups of detection cards 5 is numbered, the color developing reagents of the detection holes 6 with the same number are of the same type but different concentrations, so each group of body fluid images is the body fluid image of the detection holes 6 with the same number from the eight groups of detection cards 5. Therefore, each group of body fluid images includes eight body fluid images.

[0084] This example is to establish a standard curve formula corresponding to each group of body fluid images, and steps S2 to S5 are the process of establishing the standard curve formula of one group of body fluid images.

[0085] Step S2, determine the horizontal coordinate in a preset coordinate system according to the concentration of the colorimetric reagent of the multiple body fluid images in each group of body fluid images. Specifically, a coordinate system is pre-set, the horizontal coordinate of the coordinate system is the concentration of the colorimetric reagent, and the vertical coordinate is the grayscale value of the body fluid image. Then, the colorimetric reagent concentration corresponding to each body fluid image in the group of body fluid images is retrieved. The colorimetric reagent concentration can be measured before being placed in the patient's body fluid, or it can be manually recorded when the colorimetric reagent and the patient's body fluid are mixed. In other words, the corresponding colorimetric reagent concentration in the group of body fluid images can be directly determined in the coordinate system, that is, the horizontal coordinate is determined.

[0086] Step S3: determining a vertical coordinate in a preset coordinate system according to the color depth of the color developing reagent of the multiple body fluid images in each group of body fluid images.

[0087] First, each body fluid image in the group of body fluid images is preprocessed to determine the invalid area in the body fluid image. In a specific example, the invalid area is the edge shadow area of ​​each detection hole 6, and the invalid area can be determined by image recognition, and image recognition is a neural network model trained in advance. In other embodiments, the invalid area can also be a reflective area and a light spot area caused by reflection and light spot, respectively, such as identifying the reflective area and the light spot area in the body fluid image, and even in some images, the flocculent area must be identified, and the flocculent area refers to the invalid area caused by the flocculent generated due to insufficient reaction of the body fluid to be tested. Afterwards, the identified invalid area is removed.

[0088] Then, it is determined whether the color depth of the pre-processed body fluid image meets the preset uniformity, such as calculating the gray value difference of different color depth areas in the body fluid image. The different color depth areas are also identified by the neural network model trained in advance, and the different areas are divided according to the gray value identified. If the maximum value of the gray value difference is less than the first preset value, that is, the color difference between the darkest area and the lightest area of ​​the body fluid image is small, it means that the body fluid image meets the preset uniformity, otherwise the body fluid image does not meet the preset uniformity. Of course, the number of areas with different color depths in the body fluid image can also be counted. If the number of areas with different color depths is less than the second preset value, it means that the body fluid image meets the preset uniformity. For example, the second preset value is 5, and the area obtained by dividing the color from light to dark is less than five pieces, it means that the body fluid image meets the uniformity, otherwise it is considered that the body fluid image does not meet the uniformity. The above-mentioned first preset value and second preset value are both values ​​defined in advance.

[0089] If the body fluid image meets the preset uniformity, the gray value of the body fluid image is directly calculated, and then the calculated gray value is determined in the coordinate system, that is, the vertical coordinate is determined.

[0090] If the body fluid image does not meet the preset uniformity, the calculation model is used to predict the vertical coordinate of the body fluid image. The specific prediction process is shown in steps S31 to S34:

[0091] Step S31: Use an image segmentation algorithm to divide the body fluid image into multiple sub-regions according to color depth, and then divide the sub-regions into different chromaticity levels according to the colors in the sub-regions. In this example, multiple chromaticity levels are set in advance, and each chromaticity level corresponds to a different color depth. The higher the chromaticity level, the deeper the color depth and the larger the corresponding grayscale value. For the convenience of explaining the classification process, for example Figure 4 As shown, it is assumed that there is a body fluid image A, which is divided into 5 sub-areas according to color depth, represented by a1, a2, a3, a4, and a5 respectively, where the color depths of sub-areas a1 and a3 are consistent, and the color depths of sub-areas a2 and a5 are consistent. Figure 4 Different filled lines represent different color depths, and the same lines represent the same color depth. At the same time, the color depth of sub-areas a1 and a3>the color depth of sub-areas a2 and a5>the color depth of a4, and the color depth of sub-areas a1 and a3 corresponds to the first chromaticity level, the color depth of sub-areas a2 and a5 corresponds to the second chromaticity level, and the sub-area a4 corresponds to the third chromaticity level, then sub-areas a1 and a3 are classified into the first chromaticity level, sub-areas a2 and a5 are classified into the second chromaticity level, and sub-area a4 is classified into the third chromaticity level.

[0092] Step S32: Calculate the classification probability of the body fluid image based on the result of the sub-region division of the body fluid image, specifically, the ratio of the area of ​​the sub-regions of the body fluid image falling into different chromaticity levels to the total area of ​​the body fluid image as the classification probability at the chromaticity level. Figure 4 In the example, assume that the areas of sub-areas a1, a2, a3, a4, and a5 are 2, 4, 2, 3, and 5, respectively, in mm. 2 , then the classification probability of body fluid image A at the first chromaticity level is (2+2) / (2+4+2+3+5)=25%, similarly, the classification probability of body fluid image A at the second chromaticity level is (4+5) / (2+4+2+3+5)=56.25%, and the classification probability of body fluid image A at the third chromaticity level is (3) / (2+4+2+3+5)=18.75%. Of course, if the color depth within the body fluid image is consistent, there is only one sub-region obtained by division, and at the same time, the classification probability corresponding to the chromaticity level of the body fluid image in its sub-region is 100%.

[0093] Step S33: Select a classification probability with the largest classification probability as the target probability, and match the corresponding standard curve formula through the target probability. Specifically, when the body fluid image has only one classification probability, that is, when the colors of all sub-regions of the body fluid image are consistent, the classification probability is directly used as the target probability. When the body fluid image has multiple classification probabilities, select the classification probability with the largest classification probability as the target probability. Then, the grayscale value of the chromaticity level corresponding to the target probability is used as the grayscale value of the body fluid image, and the grayscale value of the chromaticity level is defined in advance.

[0094] Step S34: determine the ordinate in the coordinate system according to the grayscale value of the body fluid image.

[0095] Therefore, each body fluid image can obtain a corresponding grayscale value through the above-mentioned steps S31 to S34, and then the vertical coordinate in the coordinate system can be locked according to the grayscale value.

[0096] Of course, in addition to using the above-mentioned steps S31 to S34 to predict the grayscale value of the body fluid image, a neural network model based on the YOLOv8 network structure can also be used to predict the grayscale value of the body fluid image. Specifically, the YOLOv8 network structure is used to fine-tune the body fluid image and remove invalid areas such as shadow areas, reflective areas, and light spot areas in the image, and then a multi-layer convolution layer is used to perform a convolution operation on the image, and the data output by the last convolution layer is extracted as the target image, and the grayscale value of the target image is used as the grayscale value of the body fluid image, and then the ordinate in the coordinate system is locked according to the grayscale value.

[0097] Finally, the body fluid image is reduced to one-dimensional data according to the weight of the neural network model, and then the standard curve formula is retrieved. The standard curve formula here is trained in advance and is used to reflect the mapping relationship between the indicator concentration value and the one-dimensional data. Therefore, after obtaining the one-dimensional data, the indicator concentration value corresponding to the one-dimensional data will be obtained through the mapping relationship of the standard curve formula, thereby obtaining the indicator concentration value corresponding to each body fluid image.

[0098] Step S4: The horizontal coordinate and the vertical coordinate obtained for the same body fluid image are combined into a standard coordinate.

[0099] Step S5: connect multiple standard coordinates corresponding to each group of body fluid images in series to obtain a standard curve formula.

[0100] For the sake of illustration, for example Figure 5As shown, assuming that there is a coordinate system with the abscissa being the concentration of the color reagent and the ordinate being the gray value, and the standard coordinates corresponding to the eight body fluid images in a group of body fluid images are point a, point b, point c, point d, point e, point f, point g, and point h, respectively, it means that the abscissa and ordinate of point a correspond to the same body fluid image, and similarly, point b, point c, point d, point e, point f, point g, and point h also correspond to a body fluid image respectively. And by sequentially connecting multiple standard coordinates in descending or ascending order of the color reagent concentration, that is, sequentially connecting point a, point b, point c, point d, point e, point f, point g, and point h, a curve is obtained, and this curve is the standard curve formula corresponding to the group of body fluid images.

[0101] It should be noted that Figure 5 In the example, the standard curve formula is a curve close to a straight line, but in actual applications, the standard curve formula may be a curved curve, which is obtained based on the actual standard coordinates.

[0102] Based on the obtained standard curve formula, in the subsequent use process, the server 8 can use a preset analysis model to match the standard curve formula corresponding to each body fluid image from multiple standard curve formulas according to the type of patient body fluid in the body fluid to be tested, the type and concentration of the color developing reagent, that is, according to the concentration of the color developing reagent corresponding to the body fluid image, lock the abscissa of the standard curve formula, and the ordinate corresponding to the abscissa is the indicator concentration value of the body fluid image, such as Figure 5 In the example, if the horizontal axis is locked to 250, the corresponding vertical axis 15000 is the indicator concentration value.

[0103] Based on the obtained index concentration values ​​corresponding to each body fluid image, the patient characteristic information and the historical test data of the patient at different test times are obtained. Among them, the patient characteristic information includes but is not limited to the patient's height, weight and age. The historical test data includes the index concentration value at each test, the arithmetic relationship between the various index concentration values ​​and the analysis results. The arithmetic relationship between the index concentration values ​​includes the proportional relationship between the various index concentration values, which can be specifically the proportional relationship between one index concentration value and multiple index concentration values, or the proportional relationship between multiple index concentration values ​​and all index concentration values, or other proportional relationships, which are not limited here.

[0104] Then, the evaluation model is used to process the patient's characteristic information, the patient's historical test data at different test times, the concentration values ​​of each indicator and the arithmetic relationship between the indicator concentration values ​​to determine the analysis results. Among them, the evaluation model is a trained neural network model. When training this model, the actual clinical data is mainly used as sample data for training, that is, the test data of each patient at each test, and the disease type obtained by experts based on the test data analysis are used as samples to train the neural network.

[0105] Based on the trained evaluation model, the patient's characteristic information, the patient's historical test data at different test times, the concentration values ​​of each indicator and the arithmetic relationship between the indicator concentration values ​​are input into the evaluation model to obtain the patient's analysis results.

[0106] It should be noted that the analysis results include the disease type corresponding to the patient and the predicted probability of the disease type, where the calculation process of the predicted probability is:

[0107] Assume that the existing features in the evaluation model are ,Label , the specific example sample is 20,000 cases, there are 15 features, of which 6 are body fluids to be tested, 9 are characteristic information of patients, and there are 6 types of diseases set by labels, such as nephritis, myocarditis, muscular dystrophy, diabetes, kidney stones, and no abnormality. The data set in the evaluation model can be expressed as:

[0108] ,

[0109] in, ,y i ∈{nephritis, myocarditis, muscular dystrophy, diabetes, kidney stones, no abnormality}, the table structure in the dataset is 20000*16.

[0110] The evaluation model calculates the predicted probability of any sample (disease type) as:

[0111] , ,

[0112] f k is a base learner. The evaluation model consists of multiple base learners. K is the number of base learners. K represents the set of base learners in the evaluation model. The objective function of the evaluation model is:

[0113] ,

[0114] is the predicted probability of the sample by the first t-1 learners, is the predicted probability of the current learner for the sample, n is the number of samples, l is the loss function, which is based on y i and The loss function established is is the regularization term of the t-th learner.

[0115] The Taylor quadratic expansion of the objective function is as follows:

[0116] ,

[0117] in, , .

[0118] Therefore, after obtaining the analysis results, the disease type and predicted probability in the analysis results are analyzed. If the predicted probability is lower than the preset probability threshold, it means that the analysis results obtained by the current evaluation model are inaccurate, and the evaluation model needs to be optimized.

[0119] The specific optimization process is as follows: first, retrieve the disease atlas library, which includes multiple disease types and standard images of body fluids corresponding to each disease type. The standard image of body fluids is a standard image of body fluids used to characterize the disease when suffering from a certain type of disease. Usually, each disease type corresponds to multiple standard images of body fluids, and the multiple standard images of body fluids contain the states of different types of body fluids. Then, calculate the similarity of each standard image of body fluids and the body fluid image corresponding to each detection hole 6 respectively, specifically, first select a disease type and multiple standard images of body fluids corresponding to the disease type from the disease atlas library, and then match the standard images of body fluids and body fluid images of the same type of body fluids and calculate the similarity. For disease types with multiple standard images of body fluids, the average similarity of multiple groups of images is usually calculated as the similarity between the patient's disease and the disease type, and each group of images includes a standard image of body fluids and a body fluid image.

[0120] The above process is repeated until the similarity between the patient's disease in this test data and each disease in the disease atlas library can be determined.

[0121] Finally, the disease type with the highest similarity is selected as the analysis result of the patient's test, and the disease type is imported into the evaluation model to achieve the purpose of dynamically supplementing the richness of the disease types in the evaluation model. For example, the disease types set by the above labels are expanded from 6 to 10, that is, y i ϵ{nephritis, myocarditis, muscular dystrophy, diabetes, kidney stones, no abnormality, new disease 1, new disease 2, new disease 3, new disease 4}, new disease 1, new disease 2, new disease 3, new disease 4 are different types of diseases.

[0122] The correct sample labels remain unchanged, only the sample labels with low accuracy are recalibrated, and their relevant feature indicators are collected, and the evaluation model is trained again to make the evaluation model more perfect. For specific explanation, the above example is still used, that is, there are 20,000 samples in the evaluation model, and the features are increased to 18 items, of which 9 are body fluids and 9 are patient feature information. After the disease type is added to the data set, the evaluation model is trained again. The data set at this time is:

[0123]

[0124] in, ,y i ϵ{nephritis, myocarditis, muscular dystrophy, diabetes, kidney stones, no abnormality, new disease 1, new disease 2, new disease 3, new disease 4}. Therefore, by supplementing the data set in the evaluation model, the effect of optimizing the evaluation model is achieved, so that the evaluation model can output more accurate analysis results in the subsequent processing flow.

[0125] Similarly, if the disease atlas library does not contain the indicators and labels in the evaluation model, the disease atlas library can be newly created and updated with nodes and relationships.

[0126] Furthermore, the analysis result obtained by the server 8 can be sent to the mobile terminal 9 .

[0127] In the above process, the acquired detection data and the index concentration value of each body fluid to be detected are stored in the server 8. The server 8 can be deployed locally or in the cloud.

[0128] The mobile terminal 9 is connected to the server 8 for binding with the detection device 7, setting the patient's characteristic information, performing the detection delay, and viewing the detection results. In some specific embodiments, the mobile terminal 9 can be a mobile phone, a tablet, a WeChat applet, or an APP software.

[0129] In an embodiment of the present application, the mobile terminal 9 and the server 8 can communicate via wired or wireless means, such as a network cable, or wireless means such as wifi, Bluetooth, 4G or 5G network, etc.

[0130] The following will further introduce the system by introducing how to use the intelligent body component metabolism detection system.

[0131] First, the patient needs to register an account using the mobile terminal 9, and then bind the mobile terminal 9 to the detection device 7 via a network cable or Bluetooth / wifi, and enter the patient's characteristic information at the same time.

[0132] Then, the information on the detection card 5 is scanned, and the body fluid to be tested is dropped into the detection hole 6 of the detection card 5 .

[0133] Then, the test card 5 is placed in the incubator, and the upper cover 4 is closed to perform constant temperature incubation for a specified time. After the constant temperature incubation is completed, the detection device 7 can automatically collect the detection data and upload it to the server 8.

[0134] Finally, the server 8 analyzes the detection data to generate analysis results, and returns the analysis results to the mobile terminal 9 so that the patient can view them through the mobile terminal 9.

[0135] When the patient is tested again, the patient's characteristic information can be directly entered for testing.

[0136] The implementation principle of an intelligent body component metabolism detection system in an embodiment of the present application is as follows: the detection device 7 can incubate the body fluid to be tested at a constant temperature to obtain the detection data, and the server 8 can analyze the patient's test data, historical test data and characteristic information, and obtain the analysis results, so that the patient can view the analysis results through the mobile terminal 9. Since the historical test data and characteristic information are combined when analyzing the test data, the obtained analysis results are more comprehensive and accurate. The body component metabolism detection system in the present application can realize qualitative, semi-quantitative and quantitative detection and monitoring of the body fluid to be tested.

[0137] Because the system has complete body composition metabolism detection indicators and corresponding reagents, consumables, equipment, algorithms and patient terminal software, the test results have clear clinical indications, the equipment is small and portable, easy to operate, not restricted by site, and patients can see the test results and health evaluation status anytime and anywhere. It is suitable for both hospital and home testing scenarios.

[0138] The above are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Any feature disclosed in this specification (including the abstract and drawings), unless otherwise stated, can be replaced by other equivalent or alternative features with similar purposes. That is, unless otherwise stated, each feature is only an example of a series of equivalent or similar features.

Claims

1. An intelligent body component metabolism detection system, characterized by: It comprises a mobile terminal (9), a detection device (7) and a server (8); The mobile terminal (9) is used to input and store characteristic information of the patient; The detection device (7) is bound to the mobile terminal (9) and is used to obtain detection data after incubating the body fluid to be detected at a constant temperature for a specified time and upload the detection data to the server (8), wherein the body fluid to be detected includes a colorimetric reagent and a patient's body fluid; The server (8) is configured as follows: Acquiring the detection data; Processing the detection data based on a preset analysis model to obtain an indicator concentration value corresponding to the body fluid to be tested; Based on a preset evaluation model, the index concentration value corresponding to the body fluid to be tested, the historical test data of the same patient at different test times, and the patient's characteristic information are processed to obtain an analysis result; Sending the analysis result to the mobile terminal (9); The mobile terminal (9) is used to obtain and display the analysis result; The historical detection data includes: the index concentration value obtained in each detection, and the arithmetic relationship between the index concentration values; The arithmetic relationship between the index concentration values ​​includes: a proportional relationship between the respective index concentration values; The analysis results include: the disease type corresponding to the patient and the predicted probability of the disease type; The detection device (7) comprises a detection card (5) and an incubator; The detection card (5) is provided with a plurality of detection holes (6) for accommodating body fluid to be detected; The incubator is provided with a positioning groove (2), an imaging hole (3) and a heating module, the imaging hole (3) is located at the bottom of the positioning groove (2), a CMOS camera is provided at the bottom of the imaging hole (3) and the imaging sight line of the CMOS camera is arranged toward the opening end of the imaging hole (3), and the detection card (5) is snap-connected with the positioning groove (2); The heating module is arranged close to the positioning groove (2), and is used to perform constant temperature incubation on the body fluid to be tested for a specified time. After the constant temperature incubation for the specified time, the body fluid to be tested is imaged by the CMOS camera to obtain detection data.

2. The intelligent body composition metabolism detection system according to claim 1, characterized in that: The detection device (7) is provided with multiple groups, and each of the detection devices (7) in the multiple groups of the detection devices (7) is used to incubate multiple types of body fluids to be tested at a constant temperature, and the concentration of the color developing reagent corresponding to the same type of body fluids to be tested in different detection devices (7) is different.

3. The intelligent body composition metabolism detection system according to claim 2, characterized in that: The server (8) is configured as follows: Acquire multiple groups of body fluid images, each group of body fluid images includes multiple body fluid images of the same type but with different concentrations of color developing reagents; Determining the horizontal coordinate of the preset coordinate system according to the concentration of the color developing reagent of the plurality of body fluid images in each group of body fluid images; Determine the ordinate in the preset coordinate system according to the color depth of the color developing reagent of the plurality of body fluid images in each group of body fluid images; Combining the abscissa and ordinate determined from the same body fluid image into a standard coordinate; A plurality of the standard coordinates corresponding to each group of body fluid images are connected in series to obtain a standard curve formula, and each group of body fluid images corresponds to a standard curve formula.

4. The intelligent body composition metabolism detection system according to claim 3, characterized in that: The server (8) is also configured to: The vertical coordinate in the preset coordinate system is determined according to the color depth of the color developing reagent of the multiple body fluid images in each group of body fluid images, wherein the vertical coordinate of any body fluid image in each group of body fluid images is determined as follows: Preprocessing the body fluid image; Determining whether the color depth of the preprocessed body fluid image meets a preset uniformity; If yes, determining the corresponding ordinate according to the gray value of the body fluid image; If not, a computational model is used to predict the vertical coordinate of the body fluid image.

5. The intelligent body composition metabolism detection system according to claim 4, characterized in that: The server (8) is also configured to: The using a computational model to predict the ordinate of the body fluid image comprises: Dividing the body fluid image into a plurality of sub-regions according to color depth; Dividing the sub-region into different chromaticity levels according to the color depth of the sub-region, where different chromaticity levels correspond to different grayscale values; Calculating a plurality of classification probabilities according to the division results of the sub-regions of the body fluid image, wherein one of the classification probabilities corresponds to one of the chromaticity levels, and the classification probability is used to characterize the probability that the body fluid image belongs to the corresponding chromaticity level; Select the classification probability with the largest classification probability as the target probability; The corresponding ordinate is obtained by matching the gray value corresponding to the target probability.

6. The intelligent body composition metabolism detection system according to claim 5, characterized in that: The server (8) is also configured to: The calculating and obtaining a plurality of classification probabilities according to the division result of the sub-regions of the body fluid image includes: Calculating the area of ​​the sub-region falling within each of the chroma levels; The ratio of each area to the total area of ​​the body fluid image is calculated, and the ratio is used as the classification probability that the body fluid image belongs to the chromaticity level.

7. The intelligent body composition metabolism detection system according to claim 4, characterized in that: The server (8) is also configured to: Determining the ordinate in the preset coordinate system according to the color depth of the color developing reagent of the multiple body fluid images in each group of body fluid images includes: using a neural network model to predict the ordinate of the body fluid image in the preset coordinate system.

8. The intelligent body composition metabolism detection system according to claim 4, characterized in that: The server (8) is also configured to: The determining whether the color depth of the preprocessed body fluid image meets a preset uniformity level includes: Calculating grayscale value differences of regions with different color depths in the body fluid image; When the maximum value of the gray value difference is less than a first preset value, determining that the color depth of the body fluid image meets a preset uniformity; or Counting the number of regions with different color depths in the body fluid image; When the number is less than a second preset value, it is determined that the color depth of the body fluid image meets a preset uniformity.

9. The intelligent body composition metabolism detection system according to claim 3, characterized in that: The server (8) is also configured to: After obtaining the standard curve formula, it includes: According to the type of patient body fluid in the body fluid to be tested, the type and concentration of the color developing reagent, using a preset analysis model to match a standard curve formula corresponding to each of the body fluid images from a plurality of standard curve formulas; The index concentration value corresponding to the body fluid image is obtained according to the standard curve formula.

Images