Intelligent detector for combined detection of CRP and SAA

Abstract

The invention provides an intelligent detector for combined detection of CRP and SAA. The detector comprises a sample processing module, a detection module, an intelligent analysis module and an interactive display module, the sample processing module is used for collecting, separating and conveying a detection sample; the detection module is used for carrying out synchronous detection on CRP and SAA on a detection sample by adopting a fluorescence immunochromatography mode; the intelligent analysis module is used for acquiring the detection data transmitted by the detection module, and analyzing and identifying the detection data to obtain a detection and diagnosis result; the interactive display module is used for displaying a detection and diagnosis result for a user through interactive equipment; according to the invention, synchronous detection and intelligent analysis of CRP and SAA are realized, infection types and risk assessment results can be efficiently output, and the detection accuracy and clinical application value are improved.

Description

Technical Field

[0001] The present invention relates to the field of biomedical testing and artificial intelligence analysis technology, and in particular to an intelligent detector for combined detection of CRP and SAA. Background Art

[0002] C-reactive protein and serum amyloid A are common biomarkers of inflammation and infection, and are widely used in clinical infection screening, disease monitoring, and treatment effect evaluation. Existing combined CRP and SAA detection methods mostly use colloidal gold, immunochromatography, or enzyme-linked immunosorbent assay, and mainly focus on the quantitative detection of markers. However, in disease management and individual health monitoring, a single numerical measurement is often difficult to fully reflect the changing trend of the patient's infection status, and lacks the ability to comprehensively distinguish infection types and development trends, which limits the intelligent application and clinical guidance value of the test results.

[0003] At the same time, traditional detection schemes mostly rely on manual interpretation or fixed threshold judgments, and fail to fully integrate historical test data, clinical auxiliary information and sample data quality factors. The interpretation of test data has problems such as strong subjectivity and poor timeliness, which makes it difficult to meet the needs of dynamic, accurate and personalized infection assessment in multiple scenarios.

[0004] A review of relevant published technical solutions reveals that the technology with publication number CN204964521U proposes a CRP / SAA quantitative combined detection immunofluorescence chromatography test strip, comprising a box body with a lid, containing a test strip and a blood diluent bottle. The test strip comprises a backing with a nitrocellulose membrane provided on the backing, one end of the nitrocellulose membrane being connected to a fluorescent microsphere-labeled antibody binding pad, which is connected to a sample pad, and the other end of the nitrocellulose membrane being connected to an absorption pad. The fluorescent microsphere-labeled antibody binding pad is coated with a CRP monoclonal antibody and a SAA monoclonal antibody. The nitrocellulose membrane is provided with a CRP detection line coated with a CRP monoclonal antibody, an SAA detection line coated with a SAA monoclonal antibody, and a quality control line coated with a goat anti-mouse IgG polyclonal antibody in parallel. This solution can be widely used by the community or individuals for diagnosis, physical examinations, and other aspects. However, this solution only achieves the combined quantitative detection of CRP and SAA, and its functions are relatively simple, lacking the ability for combined detection and intelligent analysis. Summary of the Invention

[0005] The purpose of the present invention is to address the current deficiencies and propose an intelligent detector for the combined detection of CRP and SAA.

[0006] The present invention adopts the following technical solutions:

[0007] An intelligent detector for the combined detection of CRP and SAA, comprising a sample processing module, a detection module, an intelligent analysis module, and an interactive display module; the sample processing module is used to complete the collection, separation, and transportation of detection samples; the detection module is used to perform simultaneous detection of CRP and SAA on the detection samples using a fluorescent immunochromatographic method; the intelligent analysis module is used to obtain detection data transmitted by the detection module, and perform analysis and identification to obtain detection and diagnosis results; the interactive display module is used to display the detection and diagnosis results to the user via an interactive device.

[0008] The sample processing module includes a sampling unit, a blood separation unit and a microfluidic diversion unit; the sampling unit is used to obtain a blood sample from the test subject; the blood separation unit is used to separate the plasma and blood cells in the blood sample after receiving the blood sample; the microfluidic separation unit is used to guide the separated plasma to the CRP detection channel and the SAA detection channel respectively through a Y-shaped bifurcation channel.

[0009] The detection module includes a CRP detection channel and an SAA detection channel, which are respectively connected to the output end of the microfluidic shunt unit; the CRP detection channel and the SAA detection channel are both provided with an antibody reaction area and a fluorescence signal acquisition unit, wherein a fluorescent-labeled antibody that specifically binds to the target biomarker is fixed in the antibody reaction area and is used to generate a specific immune reaction with the plasma sample shunted in; the fluorescence signal acquisition unit is used to excite, receive and convert the fluorescent signal in the antibody reaction area into detection data to achieve quantitative analysis of CRP and SAA concentrations, and the detection data includes CRP concentration and SAA concentration.

[0010] Furthermore, the intelligent analysis module includes a feature extraction unit and a model prediction output unit; the feature extraction unit is used to extract key features required for analysis and identification from the detection data; the model prediction output unit is used to complete the identification of the current infection type and the predictive evaluation of the infection risk level through a pre-trained infection recognition model.

[0011] Furthermore, the key features extracted by the feature extraction unit include the original concentration values, ratios and logarithmic conversion values of CRP and SAA.

[0012] Furthermore, the establishment process of the infection identification model is as follows:

[0013] S11: Obtain a sample data set from a third-party hospital resource database, where each sample data in the sample data set includes test data, infection type label, sample data sampling time, and diagnosis description information;

[0014] S12: Preprocess each sample data in the sample data set. The preprocessing process includes:

[0015] S121: Key feature extraction: extracting key features of the test data from the training data in the same manner as that of the feature extraction unit;

[0016] S122: Infection type label conversion: converting the infection type label in the sample data into a structured description label;

[0017] S123: Infection risk level labeling: Based on the diagnostic description information and test data in the sample data, the severity of the infection is comprehensively judged and graded into numerical labels as the infection risk level corresponding to the sample data;

[0018] S124: Data quality assessment: Based on the sampling time of each sample data, the completeness of the diagnostic description information, and the reliability of the sample data source, the reliability of the sample data is comprehensively assessed, and a quality parameter corresponding to the sample data is generated, with a value range of (0,1);

[0019] S13: Based on the preprocessed sample dataset, an infection recognition model is constructed and supervised learning training is performed. The infection recognition model adopts a multi-task learning structure, which is used for infection type classification and infection risk level prediction respectively. The loss function of the infection recognition model during training is expressed as follows:

[0020]

[0021] Among them, L is the loss function of the infection identification model, q i is the learning weight of the i-th sample data, N is the total number of sample data, C is the total number of infection types, is the true label of the i-th sample data on the c-th infection type, that is, the infection type description label in the sample data; is the infection type description label of the i-th sample data in the c-th infection type predicted by the model; λ is the task weight coefficient, which is used to control the relative importance of classification loss and risk level loss and is set through pre-experimental settings; is the true label of the infection risk level of the i-th sample data, that is, the infection risk level in the sample data; is the infection risk level prediction value of the i-th sample data predicted by the model; for q i satisfy:

[0022] q i =1-exp(-γ·m i );

[0023] Among them, m iis the quality parameter corresponding to the i-th sample data; γ is the adjustment coefficient, which is used to control the sensitivity of sample quality to weight and is set through pre-experimental settings.

[0024] The beneficial effects achieved by the present invention are:

[0025] The present invention realizes the simultaneous detection of CRP and SAA, the joint identification of infection type and risk level, and the dynamic prediction of infection trend, and has the advantages of intelligent detection, precise evaluation and continuous monitoring. In the training process of the infection recognition model, the learning weight is set based on the data quality, so that high-quality sample data occupies a higher proportion in the training, thereby improving the model's sensitivity to effective features and learning ability, and enhancing the system's stability and predictive reliability in complex clinical environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The present invention can be further understood from the following description in conjunction with the accompanying drawings. The components in the figures are not necessarily drawn to scale, but rather the emphasis is placed on illustrating the principles of the embodiments. In different views, the same reference numerals designate corresponding parts.

[0027] Figure 1 It is a schematic diagram of the overall module of the present invention.

[0028] Figure 2 A schematic diagram of the process for establishing the infection identification model of the present invention.

[0029] Figure 3 This is a schematic diagram of the sample data preprocessing operation flow during the infection recognition model training process of the present invention.

[0030] Figure 4 Schematic diagram of the microfluidic diversion unit and detection module structure of the present invention.

[0031] Figure 5 Schematic diagram of the control strategy of the microfluidic diversion unit of the present invention.

[0032] Meaning of the numbers in the figure: 1-CRP detection channel, 2-SAA detection channel, 3-CRP microfluidic valve, 4-SAA microfluidic valve. DETAILED DESCRIPTION

[0033] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below in conjunction with its embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention; for those skilled in the art, after reviewing the following detailed description, other systems, methods and / or features of the present embodiment will become apparent; it is intended that all such additional systems, methods, features and advantages are included in this specification; included within the scope of the present invention and protected by the appended claims; additional features of the disclosed embodiments are described in the following detailed description, and these features will be apparent from the following detailed description.

[0034] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if there are terms such as "upper", "lower", "left", "right", etc. indicating directions or positional relationships, they are based on the directions or positional relationships shown in the drawings. This is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or component referred to must have a specific direction, be constructed and operate in a specific direction. Therefore, the terms describing the positional relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting this patent. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0035] Example 1:

[0036] like Figure 1 As shown, this embodiment provides an intelligent detector for the joint detection of CRP and SAA, the detector comprising a sample processing module, a detection module, an intelligent analysis module and an interactive display module; the sample processing module is used to complete the collection, separation and transportation of the test sample; the detection module is used to perform simultaneous detection of CRP and SAA on the test sample using a fluorescent immunochromatographic method; the intelligent analysis module is used to obtain the test data transmitted by the detection module, and perform analysis and identification to obtain the test diagnosis results; the interactive display module is used to display the test diagnosis results to the user through an interactive device;

[0037] The sample processing module includes a sampling unit, a blood separation unit, and a microfluidic diversion unit; the sampling unit is used to obtain a blood sample from a test subject; the blood separation unit is used to separate plasma and blood cells in the blood sample after receiving the blood sample; the microfluidic separation unit is used to guide the separated plasma to the CRP detection channel and the SAA detection channel respectively through a Y-shaped bifurcated channel;

[0038] The detection module includes a CRP detection channel and a SAA detection channel, and the CRP detection channel and the SAA detection channel are respectively connected to the output end of the microfluidic shunt unit; the CRP detection channel and the SAA detection channel are both provided with an antibody reaction area and a fluorescence signal acquisition unit, and the antibody reaction area is fixed with a fluorescent-labeled antibody that specifically binds to the target biomarker and is used to generate a specific immune reaction with the plasma sample shunted in; the fluorescence signal acquisition unit is used to excite, receive and convert the fluorescence signal in the antibody reaction area into detection data to achieve quantitative analysis of CRP and SAA concentrations, and the detection data includes CRP concentration and SAA concentration; further, the intelligent analysis module includes a feature extraction unit and a model prediction output unit; the feature extraction unit is used to extract key features required for analysis and identification from the detection data; the model prediction output unit is used to complete the identification of the current infection type and the predictive evaluation of the infection risk level through a pre-trained infection recognition model;

[0039] Furthermore, the key features extracted by the feature extraction unit include the original concentration values, ratios, and logarithmic transformation values of CRP and SAA;

[0040] Further, such as Figure 2 、 Figure 3 As shown, the establishment process of the infection identification model is as follows:

[0041] S1: Obtain a sample data set from a third-party hospital resource database, where each sample data in the sample data set includes test data, infection type label, sample data sampling time, and diagnostic description information;

[0042] S2: Preprocess each sample data in the sample data set. The preprocessing process includes:

[0043] S21: Key feature extraction: Extract key features of the test data from the training data in the same way as the feature extraction unit;

[0044] S22: Infection type label conversion: convert the infection type label in the sample data into a structured description label;

[0045] S23: Infection risk level labeling: A clinically qualified physician will comprehensively judge the severity of the infection based on the diagnostic description information and test data in the sample data, and grade and label the infection risk level corresponding to the sample data with a numerical label;

[0046] S24: Data quality assessment: Based on the sampling time of each sample data, the completeness of the diagnostic description information and the reliability of the sample data source, the reliability of the sample data is comprehensively assessed, and the quality parameter corresponding to the sample data is generated, with a value range of (0,1);

[0047] S3: Based on the preprocessed sample dataset, an infection recognition model is constructed and supervised learning training is performed. The infection recognition model adopts a multi-task learning structure, which is used for infection type classification and infection risk level prediction respectively. The loss function of the infection recognition model during training is expressed as follows:

[0048]

[0049] Among them, L is the loss function of the infection identification model, q i is the learning weight of the i-th sample data, N is the total number of sample data, C is the total number of infection types, is the true label of the i-th sample data on the c-th infection type, that is, the infection type description label in the sample data; is the infection type description label of the i-th sample data in the c-th infection type predicted by the model; λ is the task weight coefficient, which is used to control the relative importance of classification loss and risk level loss and is set through pre-experimental settings; is the true label of the infection risk level of the i-th sample data, that is, the infection risk level in the sample data; is the infection risk level prediction value of the i-th sample data predicted by the model; for q i satisfy:

[0050] q i =1-exp(-γ·m i );

[0051] Among them, m i is the quality parameter corresponding to the i-th sample data; γ is the adjustment coefficient, which is used to control the sensitivity of sample quality to weight and is set through pre-experimental settings;

[0052] Furthermore, the model prediction output unit inputs the key features extracted by the feature extraction unit into the trained infection recognition model, thereby completing the identification of the infection type of the test subject and the prediction and evaluation of the infection risk level;

[0053] Furthermore, in step S122, the infection type label is specifically infection diagnosis description information, such as bacterial infection or viral infection, and the structured description label description examples include 0-non-infection, 1-viral infection, 2-bacterial infection, and 3-mixed infection;

[0054] Furthermore, in step S123, a numerical code may be used to describe the severity of the infection as the infection risk level, such as using numbers 0 to 10 as the infection risk level;

[0055] Furthermore, in step S124, a weighted comprehensive evaluation may be performed based on the sampling time of the sample data, the integrity of the diagnostic description information, and the reliability of the source to generate a quality parameter. A method for calculating the quality parameter is as follows:

[0056]

[0057] Among them, ω1, ω2 and ω3 are pre-set weight coefficients, θ is a pre-set time attenuation adjustment coefficient; t now is the current model training time; t sample,i is the sampling time of the i-th sample data; U i is the number of diagnostic description keyword matches in the i-th sample data, U all The total number of keywords describing the pre-set diagnosis; D i The mapping coefficient is pre-set based on the sample data to obtain the institution level. For example, if the institution is a tertiary hospital, the mapping coefficient is set to 1, and if it is a secondary hospital, the mapping coefficient is set to 0.8;

[0058] This solution sets a model prediction output unit to input the key features extracted by the feature extraction unit into the pre-trained infection recognition model, which can achieve simultaneous prediction of the infection type and infection risk level of the test subject, effectively improving recognition efficiency and diagnostic accuracy; in the training process of the infection recognition model, by setting learning weights based on data quality, high-quality sample data occupies a higher proportion in training, thereby improving the model's sensitivity and learning ability to effective features, and enhancing the system's stability and prediction reliability in complex clinical environments.

[0059] Example 2:

[0060] This embodiment should be understood to include at least all the features of any of the aforementioned embodiments and be further improved thereon;

[0061] This embodiment provides an intelligent detector for the combined detection of CRP and SAA, comprising a sample processing module, a detection module, an intelligent analysis module, and an interactive display module; the sample processing module is used to complete the collection, separation, and transportation of the test sample; the detection module is used to perform simultaneous detection of CRP and SAA on the test sample using a fluorescent immunochromatographic method; the intelligent analysis module is used to obtain the test data transmitted by the detection module, and perform analysis and identification to obtain the test and diagnosis results; the interactive display module is used to display the test and diagnosis results to the user through an interactive device;

[0062] The sample processing module includes a sampling unit, a blood separation unit, and a microfluidic diversion unit; the sampling unit is used to obtain a blood sample from a test subject; the blood separation unit is used to separate plasma and blood cells in the blood sample after receiving the blood sample; the microfluidic separation unit is used to guide the separated plasma to the CRP detection channel and the SAA detection channel respectively through a Y-shaped bifurcated channel;

[0063] The detection module includes a CRP detection channel and an SAA detection channel, which are respectively connected to the output end of the microfluidic shunt unit; the CRP detection channel and the SAA detection channel are both provided with an antibody reaction area and a fluorescence signal acquisition unit, wherein a fluorescent-labeled antibody that specifically binds to a target biomarker is fixed in the antibody reaction area and is used to generate a specific immune reaction with the plasma sample shunted in; the fluorescence signal acquisition unit is used to excite, receive and convert the fluorescence signal in the antibody reaction area into detection data to achieve quantitative analysis of CRP and SAA concentrations, and the detection data includes CRP concentration and SAA concentration;

[0064] Further, such as Figure 4 As shown, the microfluidic diversion unit further includes a CRP microfluidic valve and an SAA microfluidic valve, which are respectively arranged at the entrances of the CRP detection channel and the SAA detection channel to control the introduction or closing of plasma in the two detection channels;

[0065] Furthermore, the microfluidic diversion unit differentially adjusts the opening and closing of the CRP microfluidic valve and the SAA microfluidic valve based on the changes in the CRP and SAA fluorescence signal intensities during the detection process, so as to dynamically adjust the plasma volume introduced into the CRP detection channel and the SAA detection channel, thereby achieving adaptive balance control of the detection reaction intensity;

[0066] Further, such as Figure 5 As shown, the specific control strategy of the microfluidic diversion unit is as follows:

[0067] S21: At the start of the test, a preset reaction volume of plasma is introduced into the CRP detection channel and the SAA detection channel respectively by controlling the opening and closing of the CRP microfluidic valve and the SAA microfluidic valve;

[0068] S22: After a preset reaction volume of plasma is introduced, the fluorescence signal intensities of CRP and SAA are collected at a fixed frequency within their respective preset reaction times;

[0069] S23: Obtain the fluorescence signal intensity change rates of the last three CRP and SAA samples and the fluorescence signal intensity of the last sample, and determine whether plasma volume compensation is required in the CRP detection channel and the SAA detection channel according to the following formula:

[0070]

[0071] Wherein, R1 is the CRP compensation judgment value, R2 is the SAA compensation judgment value, and when a compensation judgment value is less than its corresponding preset compensation threshold, the plasma volume of the detection channel corresponding to the compensation judgment value is compensated; I last,1 is the CRP fluorescence signal intensity collected last time, I last,2 is the intensity of the SAA fluorescence signal collected for the last time; Δs,1 is the average of the change rates of the CRP fluorescence signal intensity collected for the last three times; Δs,2 is the average of the change rates of the SAA fluorescence signal intensity collected for the last three times;

[0072] S24: When a certain detection channel needs to perform plasma volume compensation, the compensated plasma volume of the corresponding detection channel is calculated:

[0073] B1=K1·[1+ln(1+θ1·R1)];

[0074] B2=K2·[1+ln(1+θ2·R2)];

[0075] Wherein, B1 is the amount of plasma compensated for in the CRP detection channel, B2 is the amount of plasma compensated for in the SAA detection channel, K1 is the basic compensation coefficient for CRP, K2 is the basic compensation coefficient for SAA, K1 and K2 are pre-set based on the size of the corresponding detection area; θ1 is the CRP compensation sensitivity adjustment coefficient, θ2 is the SAA compensation sensitivity adjustment coefficient, θ1 and θ2 are pre-set through experiments;

[0076] S25: Repeat steps S22 to S24 until the CRP compensation judgment value and the SAA compensation judgment value are both less than their corresponding preset compensation thresholds; this scheme realizes the joint detection of CRP and SAA, and during the detection process, the respective compensation judgment values of CRP and SAA are combined to achieve accurate judgment of the adequacy of the detection reaction, and the plasma volume compensation is achieved by controlling the opening and closing of the CRP microfluidic valve and the SAA microfluidic valve, so that the compensation behavior has adaptive and quantitative characteristics, which can not only avoid detection deviations caused by insufficient reaction, but also prevent signal interference caused by excessive compensation, significantly improving the intelligence level, sensitivity control capability and result stability of the detection process.

[0077] The contents disclosed above are only preferred feasible embodiments of the present invention and do not limit the scope of protection of the present invention. Therefore, all equivalent technical changes made using the contents of the present invention description and drawings are included in the scope of protection of the present invention. In addition, the elements therein can be updated as technology develops.

Claims

1. An intelligent detector for combined detection of CRP and SAA, characterized in that: The detector includes a sample processing module, a detection module, an intelligent analysis module, and an interactive display module; the sample processing module is used to complete the collection, separation, and transportation of the test sample; the detection module is used to use fluorescent immunochromatography to perform simultaneous detection of CRP and SAA on the test sample; the intelligent analysis module is used to obtain the test data transmitted by the detection module, and perform analysis and identification to obtain the test diagnosis results; the interactive display module is used to display the test diagnosis results to the user through the interactive device; The sample processing module includes a sampling unit, a blood separation unit and a microfluidic diversion unit; the sampling unit is used to obtain a blood sample from the test subject; The blood separation unit is used to separate the plasma and blood cells in the blood sample after receiving the blood sample; the microfluidic separation unit is used to guide the separated plasma to the CRP detection channel and the SAA detection channel respectively through the Y-shaped bifurcated channel; The detection module includes a CRP detection channel and an SAA detection channel, which are respectively connected to the output end of the microfluidic shunt unit; the CRP detection channel and the SAA detection channel are both provided with an antibody reaction area and a fluorescence signal acquisition unit, wherein a fluorescent-labeled antibody that specifically binds to the target biomarker is fixed in the antibody reaction area and is used to generate a specific immune reaction with the plasma sample shunted in; the fluorescence signal acquisition unit is used to excite, receive and convert the fluorescent signal in the antibody reaction area into detection data to achieve quantitative analysis of CRP and SAA concentrations, and the detection data includes CRP concentration and SAA concentration.

2. The intelligent detector for combined detection of CRP and SAA according to claim 1, characterized in that: The intelligent analysis module includes a feature extraction unit and a model prediction output unit; the feature extraction unit is used to extract key features required for analysis and identification from the detection data; the model prediction output unit is used to complete the identification of the current infection type and the predictive evaluation of the infection risk level through a pre-trained infection recognition model.

3. The intelligent detector for combined detection of CRP and SAA according to claim 2, characterized in that: The key features extracted by the feature extraction unit include the original concentration values, ratios and logarithmic conversion values of CRP and SAA.

4. The intelligent detector for combined detection of CRP and SAA according to claim 3, characterized in that: The establishment process of the infection identification model is as follows: S11: Obtain a sample data set from a third-party hospital resource database, where each sample data in the sample data set includes test data, infection type label, sample data sampling time, and diagnosis description information; S12: Preprocess each sample data in the sample data set. The preprocessing process includes: S121: Key feature extraction: extracting key features of the test data from the training data in the same manner as that of the feature extraction unit; S122: Infection type label conversion: converting the infection type label in the sample data into a structured description label; S123: Infection risk level labeling: Based on the diagnostic description information and test data in the sample data, the severity of the infection is comprehensively judged and graded into numerical labels as the infection risk level corresponding to the sample data; S124: Data quality assessment: Based on the sampling time of each sample data, the completeness of the diagnostic description information, and the reliability of the sample data source, the reliability of the sample data is comprehensively assessed, and a quality parameter corresponding to the sample data is generated, with a value range of (0,1); S13: Based on the preprocessed sample dataset, an infection recognition model is constructed and supervised learning training is performed. The infection recognition model adopts a multi-task learning structure, which is used for infection type classification and infection risk level prediction respectively. The loss function of the infection recognition model during training is expressed as follows: Among them, L is the loss function of the infection identification model, q i is the learning weight of the i-th sample data, N is the total number of sample data, C is the total number of infection types, is the true label of the i-th sample data on the c-th infection type, that is, the infection type description label in the sample data; is the infection type description label of the i-th sample data in the c-th infection type predicted by the model; λ is the task weight coefficient, which is used to control the relative importance of classification loss and risk level loss and is set through pre-experimental settings; is the true label of the infection risk level of the i-th sample data, that is, the infection risk level in the sample data; is the infection risk level prediction value of the i-th sample data predicted by the model; for q i satisfy: what i =1-exp(-γ·m i ); Among them, m i is the quality parameter corresponding to the i-th sample data; γ is the adjustment coefficient, which is used to control the sensitivity of sample quality to weight and is set through pre-experimental settings.

Images