Hemoglobin test paper and application thereof

Abstract

The invention relates to the technical field of biomedical detection, and discloses hemoglobin detection test paper and application thereof.The hemoglobin detection test paper comprises a paper-based chip, and the paper-based chip is formed by soaking a paper base in a mixed solution composed of TMB, DIP and a buffer system and then drying the paper base; the concentration of DIP in the mixed solution is 25-35 mM, the concentration of TMB is 30-37.5 mM, and the buffer system is an acetic acid-sodium acetate buffer system. Compared with the prior art that when hemoglobin in a sample is detected, operation is complex, and cost is high, the scheme combines the high-precision analysis capacity of machine learning and the convenience of paper-based detection, and the method has the advantages of being easy and convenient to operate, low in cost, high in portability and the like, is suitable for being used by primary medical institutions and families, and has wide application prospects. The early diagnosis rate of local or systemic diseases (such as periodontitis) can be effectively improved, and powerful support is provided for early intervention and treatment of the diseases.

Description

Technical Field

[0001] The present invention relates to the technical field of biomedical detection, and in particular to a hemoglobin detection test strip and application thereof. Background Art

[0002] Oral diseases (such as gingivitis / periodontitis, oral ulcers or trauma, oral cancer or tumors, etc.), nasopharyngeal diseases (such as nosebleeds, nasopharyngeal cancer / laryngeal cancer, etc.), respiratory diseases (such as bronchiectasis / tuberculosis, lung cancer, etc.), upper gastrointestinal diseases (such as gastric ulcers / esophageal varices), and systemic diseases (such as coagulation disorders, vitamin C or K deficiency, the effects of anticoagulants, etc.) all have a certain probability of causing blood in saliva. Similarly, the detection of hemoglobin in urine (i.e., hemoglobinuria or hematuria) usually indicates red blood cell destruction or abnormal leakage caused by urinary system or systemic diseases. Therefore, the detection of hemoglobin in samples is an important means of screening for local and systemic bleeding diseases.

[0003] Currently, the commonly used detection methods mainly include chemical colorimetry, immunochromatography, and spectral analysis, but the existing technologies still have the following limitations: (1) Insufficient sensitivity and specificity: The hemoglobin concentration in the sample is usually low (especially in early-stage diseases), and the chemical colorimetry has low sensitivity, making it difficult to detect microbleeding. Similarly, the immunochromatography detection threshold is high, and it is easy to miss microbleeding (such as early nasopharyngeal carcinoma or mild periodontitis). (2) The operation is complex and relies on professional equipment, and the cost is high: For example, spectral analysis requires precision instruments and professional operation, and the detection cycle is long (centrifugation is required to remove impurities in the sample), making it difficult to achieve rapid detection at the bedside or at home. The reagents of the enzyme-linked immunosorbent assay are expensive, the detection process is cumbersome (requiring incubation, washing, color development, etc.), and the antibodies may cross-react with non-human hemoglobin (such as animal dietary residues). (3) Weak anti-interference ability: Complex components in the sample (such as mucus proteins, bacterial metabolites, and food residues in saliva samples) can easily clog the microfluidic channel of the test paper or interfere with the optical signal, resulting in distorted test results. During immunochromatographic testing, proteins such as lysozyme and IgA in saliva samples may nonspecifically adsorb to antibody markers, leading to false-positive signals. Chemical colorimetric methods (such as the benzidine method) are also susceptible to interference from other oxidases in saliva samples (such as peroxidase residues from plant foods), resulting in a high false-positive rate.

[0004] In summary, there is an urgent need to develop a low-cost, fast, simple and efficient determination method that can not only effectively make up for the shortcomings of existing technologies, but also quickly and efficiently detect the hemoglobin concentration in samples, which is of great significance for assisting the screening of local and systemic bleeding diseases. Summary of the Invention

[0005] The present invention aims to provide a hemoglobin detection test strip and its application to solve the technical problem that the existing technology for detecting hemoglobin in a sample is complex in operation and high in cost, which makes it difficult to promote it widely.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a hemoglobin detection test strip, comprising a paper-based chip, wherein the paper-based chip is formed by soaking the paper base in a mixed solution consisting of TMB, DIP, and a buffer system and then drying it; the DIP concentration in the mixed solution is 25-35 mM, the TMB concentration is 30-37.5 mM, and the buffer system is an acetic acid-sodium acetate buffer system.

[0007] Preferably, as an improvement, the concentration of acetic acid in the buffer system is 0.05-0.15 M; the concentration of sodium acetate is 0.01-0.05 M; the molar ratio of acetic acid to acetate is 2:1-5:1, and the pH value of the buffer system is 3.5-4.5.

[0008] Preferably, as an improvement, it further includes a PVC rubber plate, which includes a top plate and a bottom plate that are fastened together, and the paper-based chip is fixed between the top plate and the bottom plate; the top plate is provided with a strip observation window corresponding to the paper-based chip, the area of which is smaller than that of the paper-based chip, and scale lines are provided on the side of the strip observation window; the sample adding area of the paper-based chip is located at the low value end of the scale line.

[0009] Preferably, as an improvement, the present solution also provides a hemoglobin test strip for use in detecting hemoglobin in saliva.

[0010] Preferably, as an improvement, the application includes the following steps:

[0011] Step 1: Establish a standard curve: Prepare a gradient sample solution containing different standard hemoglobin concentrations using artificial saliva. Add the gradient sample solution to a paper-based chip for reaction, and take photos of the color development of the different gradient sample solutions. Input the color development photos into a computer vision analysis module to obtain the RGB value and color development distance of the color development area. Then, draw linear relationship diagrams and corresponding linear equations for the standard hemoglobin concentration, color intensity, and color development distance.

[0012] Step 2. Clinical sample testing: Under the same conditions, add the sample to be tested to the paper-based chip for reaction, take a photo to record its color development, input it into the computer vision analysis module, obtain the RGB value and color development distance of the color development area, and calculate the hemoglobin concentration in the sample to be tested based on the linear equation obtained in step 1.

[0013] Preferably, as an improvement, in step 1, the concentration range of hemoglobin in the gradient sample solution is 0 to 2000 μg / mL.

[0014] Preferably, as an improvement, in step 1 and step 2, the reaction time of the gradient sample solution and the sample to be tested with the test paper is 3 to 5 minutes.

[0015] Preferably, as an improvement, in step one, the color-developed photograph is input into a computer vision analysis module, and the analysis steps for obtaining the RGB value and color-developed distance of the color-developed area are as follows: traverse all image files of the specified format and read them as OpenCV image format BGR; convert the image from BGR to HSV and set a color threshold, find the area in the image that meets the color range, and generate a mask; use morphological operation opening to remove small noise points and retain a larger target area; perform contour detection and draw a positioning rectangular box on the original image as the ROI area; for each positioning-marked ROI area, extract the average color value and the color-developed distance of the boxed area.

[0016] Preferably, as an improvement, in step 1, after the computer vision analysis module obtains the RGB value of the color display area, the calculation formula of the color intensity in the paper-based chip is: color intensity = 0.3R + 0.59G + 0.11B.

[0017] Preferably, as an improvement, in step 1, the linear equations of hemoglobin concentration, color intensity, and color development distance are as follows:

[0018] Color Intensity = -40.826C Hb +192.929(1);

[0019] In formula (1), C Hb is the hemoglobin concentration, ranging from 1.6 to 200 μg / mL, R 2 =0.998;

[0020] Color rendering distance = 0.133C Hb +10.782(2);

[0021] In formula (2), C Hb is the hemoglobin concentration, ranging from 3 to 200 μg / mL, R 2 =0.972;

[0022] Color rendering distance = 0.110C Hb +33.752(3);

[0023] In formula (3), C Hb is the hemoglobin concentration, ranging from 200 to 2000 μg / mL, R 2 =0.992.

[0024] The principles of this program are:

[0025] This solution involves soaking TMB and DIP in a paper-based chip. After combining TMB (3,3',5,5'-tetramethylbenzidine) and DIP (diisopropyl hydroperoxide), the two substances react and develop color in the presence of hemoglobin. The color intensity and color development distance of the test paper after development effectively indicate the hemoglobin content in the sample to be tested. The color development principle of TMB (3,3',5,5'-tetramethylbenzidine) and DIP (3,5-diisopropyl hydroperoxide) with hemoglobin is as follows: Figure 1 As shown: The heme group in hemoglobin (Hb) has peroxidase-like activity, catalyzing the decomposition of 3,5-diisopropylbenzene hydroperoxide (DIP) to generate hydroxyl radicals (·OH), which in turn convert the colorless 3,3',5,5'-tetramethylbenzidine (TMB) into oxidized 3,3',5,5'-tetramethylbenzidine (oxTMB), resulting in a blue color. The color depth and distance of the color development are then accurately calculated by a computer, revealing the hemoglobin concentration in the sample being tested.

[0026] The advantages of this solution are:

[0027] 1. Easy operation and low cost: Compared with the existing technology for detecting hemoglobin in samples, which is complex and costly, this solution combines the high-precision analysis capabilities of machine learning with the convenience of paper-based testing. This method has the advantages of simple operation, low cost, and strong portability. It is suitable for primary medical institutions and home use, and can effectively improve the early diagnosis rate of local or systemic diseases (such as periodontitis), providing strong support for early intervention and treatment of diseases.

[0028] 2. Dual signal output: The dual signal output method of colorimetry and flow distance is adopted. The two characteristic information of color intensity and color display distance are extracted through the computer vision analysis module. Compared with the single signal output, it more comprehensively reflects the content of salivary hemoglobin and significantly improves the accuracy and reliability of detection.

[0029] 3. This solution effectively adjusts the pH of the test paper by infiltrating the paper-based chip with an acetic acid-sodium acetate buffer system, providing a suitable environment for the color development reaction, effectively enhancing the reaction intensity and sensitivity, and improving the detection sensitivity. Specifically, the test paper of this solution is easy to use. When using the test paper of this solution to detect the hemoglobin content in saliva, the detection limit is 3μg / mL, which is highly sensitive. In addition, during the detection process, the test paper of this solution can eliminate the interference of other components in saliva and specifically detect hemoglobin, thereby improving the detection specificity and accuracy, thereby helping to quickly assist in confirming whether the patient has related diseases, so as to facilitate timely intervention and treatment. BRIEF DESCRIPTION OF THE DRAWINGS BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1Schematic diagram of the color development principle of TMB and DIP with hemoglobin in the hemoglobin detection test strip according to an embodiment of the present invention.

[0031] Figure 2 This is a pH optimization diagram of the buffer system in the hemoglobin test strip according to an embodiment of the present invention.

[0032] Figure 3 This is a wavelength optimization diagram for the UV-visible absorption spectrum used in the embodiment of the present invention.

[0033] Figure 4 This is a diagram showing the optimized concentration of TMB in the hemoglobin test strip according to an embodiment of the present invention.

[0034] Figure 5 This is a diagram showing the optimized concentration of DIP in the hemoglobin test strip according to an embodiment of the present invention.

[0035] Figure 6 This is a fitting diagram of the CI predicted value and actual value based on the BPNN output in an embodiment of the present invention.

[0036] Figure 7 It is the relative error between the CI predicted value output by BPNN and the actual value in the embodiment of the present invention.

[0037] Figure 8 The figure is a schematic diagram of the process of preparing and using the hemoglobin detection test strip according to an embodiment of the present invention.

[0038] Figure 9 This is a standard curve diagram of the hemoglobin content and color intensity in the gradient sample solution when the hemoglobin detection test strip according to the embodiment of the present invention is used.

[0039] Figure 10 This is a standard curve diagram of the hemoglobin content and color development distance in a gradient sample solution when the hemoglobin detection test strip according to an embodiment of the present invention is used.

[0040] Figure 11 Schematic diagram of the results of hemoglobin content and color intensity in a test sample (taking saliva as an example) when the hemoglobin test strip according to an embodiment of the present invention is used.

[0041] Figure 12 Schematic diagram of the results of the hemoglobin content and color development distance in the test sample (taking saliva as an example) when the hemoglobin detection test strip according to an embodiment of the present invention is used.

[0042] Figure 13 This is a schematic diagram of the results of the hemoglobin test strip according to an embodiment of the present invention when the test sample (taking saliva as an example) is used.

[0043] Figure 14 The scatter plot shows the correlation between the color concentration and BOP% in saliva samples.

[0044] Figure 15 The bar graph shows the paired BOP% (red bar, left axis) and color concentration (blue bar, right axis) in 103 saliva samples.

[0045] Figure 16 The scatter plot shows the correlation between the color concentration and BOP% in saliva samples.

[0046] Figure 17 The bar graph shows the paired BOP% (red bar, left axis) and color concentration (blue bar, right axis) in 103 saliva samples.

[0047] Figure 18 is the ROC curve before treatment.

[0048] Figure 19 This is the ROC curve 1 month after treatment. DETAILED DESCRIPTION

[0049] The present invention will be further described in detail below with reference to the examples, but the embodiments of the present invention are not limited thereto. Unless otherwise specified, the technical means used in the following examples and experimental examples are conventional means well known to those skilled in the art, and the materials, reagents, etc. used are all commercially available.

[0050] Program Overview

[0051] This solution provides a method for preparing a paper-based chip assisted by machine learning. The paper-based chip is prepared by soaking a paper substrate in a mixed solution consisting of TMB, DIP, and a buffer system and then drying it. The method comprises the following steps:

[0052] Step 1: Optimize the buffer system: prepare a gradient pH solution of acetic acid-sodium acetate buffer system, add salivary hemoglobin to react, and measure the absorbance value to obtain the optimized buffer system pH;

[0053] As a reference, the following is included:

[0054] 1.1 Set up 10 acetic acid-sodium acetate (HAc / NaAc) buffer gradients from pH = 2 to pH = 6.5, each gradient corresponding to one experimental group, with 2 replicates per group;

[0055] 1.2 Add 25 μL of buffer solutions of different pH values, 20 μL of DIP solution, and 30 μL of TMB solution to each centrifuge tube, mix well, and then add 25 μL of human standard hemoglobin solution diluted to 100 μg / mL with artificial saliva;

[0056] 1.3 Place the centrifuge tube in a 37°C water bath for 15 minutes. Measure the absorbance of the reaction solution at 652 nm. Compare the absorbance at different pH values to determine the pH with the highest absorbance and find the optimal pH.

[0057] As a reference, the concentration of acetic acid in the buffer system optimized in this scheme is 0.05-0.15 M; the concentration of sodium acetate is 0.01-0.05 M; the molar ratio of acetic acid to acetate is 2:1-5:1, and the pH value of the buffer system is 3.5-4.5.

[0058] Step 2: Optimize TMB and DIP concentrations: Prepare gradient TMB solutions and gradient DIP solutions, adjust the solution pH to the buffer system pH optimized in step 1, add salivary hemoglobin to react, and measure the absorbance to obtain the optimized TMB and DIP concentrations;

[0059] As a reference, the following is included:

[0060] 2.1 Prepare 10 sets of TMB solutions with different concentration gradients, ranging from 5mM to 40mM, and 8 sets of DIP solutions with different concentration gradients, ranging from 5mM to 50mM;

[0061] 2.2 Under the determined optimal pH buffer system, TMB and DIP solutions of varying concentrations were mixed with a standard human hemoglobin solution diluted to 100 μg / mL with artificial saliva and reacted at 37°C for 15 min.

[0062] 2.3 Measure the absorbance of each reaction system at 652 nm. Analyze the relationship between the absorbance and the concentrations of TMB and DIP to determine the optimal concentrations of TMB and DIP under the optimal pH conditions.

[0063] As a reference, the concentration of DIP in the mixed solution optimized in this protocol is 25-35 mM, and the concentration of TMB is 30-37.5 mM.

[0064] Step 3: Obtain a paper-based chip: Draw a test paper chip pattern according to the detection requirements, cut the test paper to obtain a paper base, and fix the mixed solution optimized in steps 1 and 2 onto the paper base to obtain a paper-based chip.

[0065] As a reference, the test paper was cut using a laser engraving machine. The parameters of the laser engraving machine were set to a speed of 500 to 800 mm / s, a frequency of 10 to 30 kHz, and a power of 40 to 70%.

[0066] As a reference, the immersion method is used to fix the optimized mixed solution on the paper base. The steps are as follows: soak the paper base in a mixed solution consisting of TMB solution, DIP solution and buffer system for 1 to 2 minutes, take it out and spread it flat on a glass plate, wrap it with non-woven fabric, put it in an oven at 50 to 60°C and dry it for 5 to 10 minutes, and stick it on a PVC board to obtain a paper-based chip.

[0067] Step 4: Build a machine learning model: Build a computer vision analysis module to perform image analysis on the test strips’ test results; build a back-propagation neural network model to train and predict the data set; and build a genetic algorithm model to iteratively optimize the system to obtain the optimized paper-based chip.

[0068] As an improvement, the following is included:

[0069] S1. Establish a computer vision analysis module, including using Python to establish a computer vision analysis module for image analysis of the test strip test results. The steps are as follows:

[0070] S101. Use Python to build a computer vision analysis module. The module script traverses all image files of a specified format from the file and reads them into the OpenCV image format BGR;

[0071] S102. Perform color space conversion and color detection, converting the image from BGR to HSV, setting a color threshold, finding the area within the image that matches the color range, and generating a mask. Use morphological operations to remove small noise points while retaining the larger target area. Then perform contour detection and location marker recognition, drawing a red rectangle on the original image to highlight the ROI area.

[0072] S103. Perform feature extraction and result storage. For each ROI area of the positioning mark, extract the average color value HSV, HLS and BGR; store the data in a dictionary, organize all the results into a table, save the processed image to the specified output folder, and save the feature value table as an Excel file for subsequent analysis and visualization.

[0073] S2. Establish a BP neural network model, including using MATLAB R2023b to establish a back propagation neural network model, and train and predict the data set. The steps are as follows:

[0074] S201. Build a backpropagation neural network model using MATLAB R2023b. In MATLAB R2023b, extract data using xlsread. Randomly partition the data using randperm, with 80% of the data used as a training set and the remaining 20% as a test set. Normalize the training data using mapminmax to a range of [-1, 1].

[0075] S202. Construct a BPNN with input and output layer nodes set based on the number of data columns and 8 to 10 hidden layers. Use newff, tansig for the input and hidden layers, purelin for the hidden and output layers, and trainlm for the training function. Parameters are configured as follows: 1500 to 2000 training cycles, a learning rate of 0.01 to 0.05, and a target error of 1e. -5 ~1e -4 , momentum factor 0.01, minimum gradient 1e -6 ~1e -5 , the maximum number of failures is 6 to 10;

[0076] S203. Train the network with normalized training data and use sim to predict the normalized input of the test set; after the prediction results are denormalized, calculate MAE, MSE, RMSE, MAPE and R 2 Performance indicators, and draw comparison charts and error charts; BPNN model training is completed, and the network, input and output normalization parameters are saved to provide support for subsequent predictions.

[0077] S3. Establish a genetic algorithm model, including using MATLAB R2023b to establish a genetic algorithm model and perform iterative optimization. The steps are as follows:

[0078] S301. Use MATLAB R2023b to establish a genetic algorithm (GA) and perform iterative optimization. Use the trained BPNN model as the GA's fitness function. In the GA, each individual represents a set of parameters or input variables. The BPNN model predicts the output values corresponding to these parameters, thereby calculating its fitness.

[0079] S302. Initialize GA parameters, including the number of generations, population size, crossover probability, and mutation probability; set the encoding length and value range of the variables, initialize the population, calculate the fitness value of each individual, find the individual with the best fitness in the initial population and its fitness value, and record the average fitness and the best fitness;

[0080] S303. Repeat the process in S302. Through multiple generations of evolution, the individuals in the population gradually approach the optimal solution, thereby outputting the optimal paper-based parameters and obtaining the optimized system of the paper-based chip, including the composition and dosage of the reagents in the mixed solution used to soak the paper-based chip.

[0081] This solution also provides a hemoglobin test strip, comprising a paper-based chip prepared by the above method, including a paper base and a mixed solution system fixed to the paper base. In other embodiments, the commercially available hemoglobin test strip product further includes a PVC sheet, the PVC sheet comprising a top plate and a bottom plate that snap together, with the paper-based chip fixed between the top and bottom plates; the top plate is provided with a strip-shaped observation window corresponding to the paper-based chip, the area of which is smaller than that of the paper-based chip, and scale lines are provided on the side of the strip-shaped observation window. When assembling the paper-based chip product, the paper-based chip is first attached to the PVC bottom plate, and then the top plate is snapped together to form the device.

[0082] This solution also provides an application of a hemoglobin test strip for detecting hemoglobin in saliva. The hemoglobin test strip is prepared using the above method, and the application includes the following steps:

[0083] Step (1), establishment of a standard curve: dilute human standard hemoglobin with artificial saliva to prepare a gradient sample solution containing different standard hemoglobin concentrations.

[0084] For reference, artificial saliva consists of the following raw materials by weight: NaCl 0.4 g / L, KCl 0.4 g / L, CaCl2·2H2O 0.795 g / L, KH2PO4 0.340 g / L, Na2HPO4·12H2O 0.336 g / L, and deionized water as the solvent, adjusted to a pH of 6.8–7.0. The hemoglobin concentration in the gradient sample solution ranges from 0 to 2000 μg / mL.

[0085] The gradient sample solution was then added to the paper-based chip and reacted for 3 to 5 minutes. The color development photos of the different gradient sample solutions were recorded. The color development photos were input into the computer vision analysis module to obtain the RGB value and color development distance of the color development area. The linear relationship diagrams between the standard hemoglobin concentration, color intensity, and color development distance were then drawn, along with the corresponding linear equations.

[0086] Among them, the color development photos are input into the computer vision analysis module to obtain the RGB value and color development distance of the color development area. The analysis steps are as follows: the color development photos of different gradient sample solutions are placed in the same folder in a specified format as an image dataset, all image files in the specified format of the image dataset are traversed, and read into the OpenCV image format BGR; the image is converted from BGR to HSV and the color threshold is set, the area in the image that meets the color range is found, and a mask is generated; the morphological operation opening operation is used to remove small noise points and retain the larger target area; contour detection is performed and a positioning rectangular box is drawn on the original image as the ROI area; for each positioning marked ROI area, the average color value and the color development distance of the boxed area are extracted.

[0087] After the computer vision analysis module obtains the RGB value of the color display area, the calculation formula of the color intensity ColorIntensity in the paper-based chip is: Color Intensity = 0.3R + 0.59G + 0.11B.

[0088] In addition, the linear equations of hemoglobin concentration, color intensity, and color development distance obtained after fitting are as follows:

[0089] Color Intensity = -40.826C Hb +192.929(1);

[0090] In formula (1), C Hb is the hemoglobin concentration, ranging from 1.6 to 200 μg / mL, R 2 =0.998;

[0091] Color rendering distance = 0.133C Hb +10.782(2);

[0092] In formula (2), C Hb is the hemoglobin concentration, ranging from 3 to 200 μg / mL, R 2 =0.972;

[0093] Color rendering distance = 0.110C Hb +33.752(3);

[0094] In formula (3), C Hb is the hemoglobin concentration, ranging from 200 to 2000 μg / mL, R 2 =0.992.

[0095] Step (2), clinical sample testing: Under the same conditions, add the sample to be tested to the paper-based chip and react for 3-5 minutes. Take a photo to record the color development and input it into the computer vision analysis module to obtain the RGB value and color development distance of the color development area. According to the linear equation obtained in step 1, calculate the hemoglobin concentration in the sample to be tested.

[0096] This protocol specifically uses saliva samples as an example to illustrate the hemoglobin test strips, preparation methods, and applications of this protocol.

[0097] Example 1: Preliminary preparation of paper-based chips

[0098] Step 1: Optimize the buffer system: prepare a gradient pH solution of acetic acid-sodium acetate buffer system, add salivary hemoglobin to react, and measure the absorbance value to obtain the optimized buffer system pH;

[0099] like Figure 2 To optimize the pH of the buffer system, an acetic acid-sodium acetate (HAc / NaAc) buffer system was used to set a total of 10 pH gradients from 2 to 6.5 (i.e., the pH values of the 10 gradient sample solutions were 2, 2.5, 3, 3.5, 4, 4.5, 5, 5.5, 6, and 6.5, respectively). Reactions were performed under different pH systems. Ten experimental groups were set up according to the pH value, with two parallel groups in each group. Each group was first added with 25 μL of acetic acid-sodium acetate (HAc / NaAc) buffer of different pH values, followed by 20 μL of DIP solution, 30 μL of TMB solution, and 25 μL of salivary hemoglobin solution. After mixing, the mixture was placed in a constant temperature water bath at 37°C for 15 minutes, and the absorbance of the reaction solution at 652 nm was then measured (e.g., Figure 2 By comparing the absorbance values at different pH values, the pH value with the highest absorbance is determined, thereby finding the optimal pH condition.

[0100] like Figure 3 As shown, by comparing the UV-visible absorption spectra of the reagent combination of this scheme, the absorbance value of the reaction solution was detected at 652 nm.

[0101] Step 2: Optimize TMB and DIP concentrations: Prepare gradient TMB solutions and gradient DIP solutions, adjust the solution pH to the buffer system pH optimized in step 1, add salivary hemoglobin to react, and measure the absorbance to obtain the optimized TMB and DIP concentrations;

[0102] like Figure 4 and Figure 5To optimize the concentrations of TMB and DIP, 8 groups of TMB solutions (5-40 mM) with different concentration gradients were prepared (i.e., the TMB concentrations in the 8 gradient sample solutions were 5 mM, 10 mM, 15 mM, 20 mM, 25 mM, 30 mM, 35 mM, and 40 mM, respectively), and 10 groups of DIP solutions (5-50 mM) with different concentration gradients were prepared (i.e., the DIP concentrations in the 10 gradient sample solutions were 5 mM, 10 mM, 15 mM, 20 mM, 25 mM, 30 mM, 35 mM, and 40 mM, respectively). 0mM, 25mM, 30mM, 35mM, 40mM, 45mM, and 50mM). In the optimal pH buffer system determined in step 2, mix the TMB and DIP gradient sample solutions of varying concentrations with a human standard hemoglobin solution diluted to 100μg / mL with artificial saliva. React according to the established reaction conditions (i.e., in a 37°C constant temperature water bath for 15 minutes). Measure the absorbance of each reaction system at 652nm. Analyze the relationship between absorbance and TMB and DIP concentrations to determine the optimal TMB and DIP concentrations under the optimal pH conditions.

[0103] Step 3: Obtaining a paper-based chip: Drawing a test paper chip pattern according to the test requirements, cutting the test paper to obtain a paper base, and fixing the mixed solution obtained by optimizing in steps 1 and 2 onto the paper base to obtain a paper-based chip;

[0104] First, AutoCAD software was used to draw the graphics of various test paper chips (with varying widths and lengths). A laser engraver was then used to cut the various paper chips. The laser engraver's parameters were set to a speed of 500 to 800 mm / s, a frequency of 10 to 30 kHz, and a power of 40 to 70%. For reference, the laser engraver's parameters in this embodiment were specifically set to a speed of 750 mm / s, a frequency of 10 kHz, and a power of 50%.

[0105] The immersion method is used to fix the optimized system in steps one and two onto a paper-based chip. The specific steps are as follows: soak the paper-based chip in a mixed solution consisting of TMB solution, DIP solution and buffer system for 1 minute (the immersion time can be selected from 1 to 2 minutes), take it out and spread it on a glass plate, wrap it with non-woven fabric, and put it in an oven to dry at 60°C for 5 minutes (the drying condition can be selected from 50 to 60°C for 5 to 10 minutes), and finally stick it on a PVC sheet to form a test paper product that is easy to sell on the market. Among them, the sample addition area of the paper-based chip is at the low end of the scale line.

[0106] Example 2: Machine Learning Cycle Model Establishment and Parameter Optimization of Paper-Based Chip

[0107] This approach builds a paper-based chip based on a machine learning cyclic model. Using paper parameters (width, length, reaction time, and drop volume) as input, the neural network dataset is constructed using color development distance and color intensity obtained through a computer vision algorithm. The model is constructed based on the backpropagation properties of the BP neural network, and training and validation procedures are performed on the paper-based dataset. During model development, the data samples are randomly divided into three parts, with a ratio of 80%, 10%, and 10%, serving as the training, validation, and test sets, respectively. The model training process is implemented using MATLAB software.

[0108] First, the machine learning cycle model is established, which includes the following steps:

[0109] S1. Establishment of Computer Vision Analysis Module

[0110] S101. Use Python to establish a computer vision (CV) analysis module for image analysis of the test strip: the script traverses all image files of the specified format from the file and reads them into the OpenCV image format (BGR).

[0111] S102. Perform color space conversion and color detection, converting the image from BGR to HSV to make it easier to detect a specific color range. Set a color threshold, find the area in the image that matches the color range, and generate a mask. Use morphological operations (opening operations) to remove small noise points and retain the larger target area. Then perform contour detection and position marker recognition. Use cv2.findContours to find the contours in the mask and create a rectangular border for the marker. Draw a red rectangular box on the original image.

[0112] S103. Perform feature extraction and result storage. For each ROI region marked by the location, extract the average color value (HSV, HLS, and BGR). Store the data in a dictionary and organize all the results into a table using pandas. Save the processed image to a designated output folder and save the feature value table as an Excel file for subsequent analysis and visualization.

[0113] S2. BP neural network model establishment

[0114] S201. Build a Back Propagation Neutral Network (BPNN) model using MATLAB R2023b to train and predict the dataset: In MATLAB R2023b, extract the data using xlsread. Randomly partition the data using randperm, with 80% of the data set used as the training set and the remaining 20% used as the test set. Normalize the training data using mapminmax to a range of [-1, 1].

[0115] S202. Construct a BPNN with input and output layer nodes set based on the number of data columns and 8 to 10 hidden layers. Use newff, tansig for the input and hidden layers, purelin for the hidden and output layers, and trainlm for the training function. Parameters are configured as follows: 1500 to 2000 training cycles, a learning rate of 0.01 to 0.05, and a target error of 1e. -5 ~1e -4 , momentum factor 0.01, minimum gradient 1e -6 ~1e -5 , the maximum number of failures is 6 to 10.

[0116] S203. Train the network with normalized training data and use sim to predict the normalized input of the test set. After the prediction results are denormalized, calculate MAE, MSE, RMSE, MAPE and R 2 Performance indicators are generated, and comparison and error graphs are plotted. BPNN model training is completed, and the network and input and output normalization parameters are saved to provide support for subsequent predictions.

[0117] S3. Genetic Algorithm Model Establishment: A genetic algorithm (GA) was established using MATLAB R2023b for iterative optimization. The trained BPNN model was used as the GA's fitness function. In the GA, each individual represents a set of parameters or input variables. The BPNN model predicts the output values corresponding to these parameters, thereby calculating its fitness. GA parameters were initialized, including the number of evolutionary generations (100-200 generations), population size (40-100 individuals), crossover probability (0.4-0.7), and mutation probability (0.02-0.1). The encoding length and value range of the variables were set, the population was initialized, and the fitness value of each individual was calculated. The individual with the best fitness in the initial population and its fitness value were found. The average fitness and the best fitness were recorded. This process was repeated. Through multiple generations of evolution, the individuals in the population gradually approached the optimal solution, thus outputting the optimal paper-based parameters and obtaining the optimized system for the paper-based chip, including the composition and dosage of the mixed solution used to immerse the paper-based chip.

[0118] Figure 6This is a fitting diagram of the CI predicted value and the actual value based on the BPNN output. The fitting curves of the predicted value and the true value have a high degree of overlap, indicating that the established model has good accuracy.

[0119] Figure 7 is the error percentage between the CI predicted value and the actual value based on the BPNN output. It can be seen that the error between the predicted value and the true value is small, which further illustrates that the model has good accuracy and high credibility.

[0120] Example 3: cyclic application of machine learning models

[0121] like Figure 8 The following is a schematic diagram of the principle and process for preparing and applying a machine learning-assisted hemoglobin test strip. A salivary hemoglobin concentration gradient and a drop volume gradient are set, and the test strip is added to a paper-based chip for reaction. A camera (such as a mobile phone) is used to record the color development of each test strip at different reaction times. The original test strip image is used as input, and a computer vision analysis module is used to extract features and calculate color intensity and color development distance. The formula for calculating color intensity in the paper-based data is: Color Intensity = 0.3R + 0.59G + 0.11B.

[0122] The extracted data is fed into a BP neural network, which trains and validates the paper-based dataset, outputting predicted values for the paper-based parameters. The trained BP neural network is then used as the fitness function of the genetic algorithm for iterative optimization, and the output accuracy is calculated. If the accuracy is low, the newly added data is fed back to the computer vision analysis module, and optimization continues until high detection accuracy is achieved.

[0123] Example 4: Standard curve establishment

[0124] Human standard hemoglobin was diluted with artificial saliva and different concentration gradients were set. Salivary hemoglobin solutions of different concentrations (0-2000 μg / mL) were added to the S5-optimized paper base and reacted for 3-5 minutes. Then, a smart camera device (such as a mobile phone) was used to take photos and record the color development photos at different concentrations. The photos were input into the computer vision analysis module to obtain the RGB value and color development distance of the color development area, and a linear relationship diagram of the target substance hemoglobin and the color intensity and color development distance and the corresponding linear equation were drawn.

[0125] like Figure 9 , is the standard curve of the color development distance of this detection method, and the linear equation is: Distance = 0.133C Hb +10.782(C Hb :3~200μg / mL,R 2 =0.972), Distance = 0.110CHb +33.752(C Hb :200~2000μg / mL,R 2 =0.992), the linear relationship is good, and the detection limit is 3μg / mL. As can be seen from the figure, the color development distance value of the detection area (such as Figure 9 The middle vertical axis (Colorimetric Distance) increases with the increase of salivary hemoglobin concentration.

[0126] like Figure 10 , is the standard curve of the color intensity of this detection method, and the linear equation is: ColorIntensity=-40.826C Hb +192.929(C Hb :1.6-200μg / mL,R 2 =0.998), the linear relationship is good, and the detection limit is 1.6μg / mL. As can be seen from the figure, the color intensity of the detection area (such as Figure 10 The middle vertical axis (Color Intensity) decreases as the salivary hemoglobin concentration increases, which means that the color of the test paper becomes more obvious as the salivary hemoglobin concentration increases.

[0127] This shows that the salivary hemoglobin detection method provided by the present invention has a low detection limit and can meet the actual detection requirements for early diagnosis of periodontitis.

[0128] Example 5: Specificity test of the detection method of the present invention

[0129] To test the specificity of the salivary hemoglobin test strip provided by the present invention for the early diagnosis of periodontitis, several common oral salivary substances, lysozyme, α-amylase, L-glutamic acid, and glucose, were selected and processed according to the operating procedures of Example 2 to verify the specificity of the method; the concentrations of lysozyme, α-amylase, L-glutamic acid, and glucose were 1 mg / mL, and the Hb concentration was 100 μg / mL.

[0130] like Figure 11 、 Figure 12 and Figure 13 , is a schematic diagram of the specific experimental results of the salivary hemoglobin test strip provided by the present invention for the early diagnosis of periodontitis. Figure 11It can be seen that even if the concentration of lysozyme, α-amylase, glutamic acid, and glucose is 10 times higher than that of salivary hemoglobin (Hb), it still cannot cause the color intensity (such as Figure 11 There is no obvious change in the vertical coordinate Color Intensity) and no color rendering distance (such as Figure 12 This further demonstrates that the salivary hemoglobin detection method provided by the present invention has good selectivity.

[0131] Experimental Example 1: Clinical actual sample detection research

[0132] Clinical sample testing: According to the linear equation obtained in Example 5, under the same conditions, a clinical sample (saliva) was added to the optimized paper substrate for reaction. After 3 to 5 minutes, a color development photo was taken using an intelligent camera device (such as a mobile phone). The photo was input into a computer vision analysis module to obtain the RGB value and color development distance of the color development area, and the hemoglobin content in the clinical sample (saliva) was calculated.

[0133] As shown in the table below, the colorimetric synergistic flow distance method of this embodiment can accurately measure the hemoglobin concentration in actual samples for early diagnosis of periodontitis. The results are shown in Table 1.

[0134] Table 1 Actual clinical sample testing

[0135]

[0136] The salivary hemoglobin test strip provided by the present invention is used for the early diagnosis method of periodontitis and has the advantages of simple operation, high selectivity, high sensitivity, and easy portability.

[0137] Experimental Example 2: Relationship between the color concentration of the test paper and the degree of gingival inflammation

[0138] BOP is defined as follows: Bleeding on probing (BOP) is one of the most reliable and widely used clinical parameters for reflecting gingival tissue health. It is defined as marginal gingival bleeding induced by standardized pressure applied to the gingival sulcus or the side of the periodontal pocket in the absence of a visible periodontal pocket. This phenomenon reflects micro-injury or inflammation-related capillary dilation and increased permeability within the gingival epithelium.

[0139] BOP is highly sensitive to inflammation and is widely used to assess periodontal health, monitor disease activity, and follow up on treatment response. In the 2018 European and American Consensus on Periodontology, BOP was explicitly included in the diagnostic criteria for gingival health: clinically healthy gingiva is defined as BOP-positive sites within intact or stable periodontal tissues with a probing depth of ≤3mm and a proportion of less than 10%. Furthermore, studies have shown that persistent BOP is a risk factor for future attachment loss of periodontal tissues, while BOP-negative sites generally have a lower risk of disease progression.

[0140] Therefore, BOP can not only be used for early detection of periodontal inflammation but also serve as a key indicator for evaluating treatment effectiveness and disease control during follow-up. Its simplicity, non-invasiveness, and high reproducibility make it of great value in both basic research and clinical practice.

[0141] First, this study explored the relationship between color intensity and periodontal bleeding percentage (BOP%). Figure 14 and Figure 15 shown.

[0142] Figure 14 The scatter plot in the middle shows the correlation between the color concentration and BOP% in saliva samples. The results show a significant negative correlation (Pearson's r = -0.8279). Each blue dot represents an independent sample, the red regression line is the linear fit result, and the shaded area represents the 95% confidence interval. This trend indicates that the higher the BOP%, the lower the color concentration of the test strip, suggesting that color changes can be used to reflect the degree of gingival inflammation.

[0143] Figure 15 The middle bar graph shows the results of 103 saliva samples (such as Figure 11 The BOP% (red column, left axis, Bleeding on porbing score) and color intensity (blue column, right axis, Color Intensity) are paired and displayed on the horizontal axis (Saliva sample number). In most samples, higher BOP% corresponds to lower color intensity. The purple horizontal line represents the reference threshold for color intensity, approximately 160. The samples on the right are concentrated in the area with lower BOP% and higher color intensity. This trend is consistent with the negative correlation results in the scatter plot, further supporting the feasibility of color intensity as an indicator for assessing the severity of gingival inflammation.

[0144] In summary, the color concentration of the test paper can effectively evaluate the degree of gingival inflammation.

[0145] Experimental Example 3: Relationship between the color development distance of the test paper and the degree of gingival inflammation

[0146] This study explored the relationship between color intensity and periodontal bleeding percentage (BOP%). Figure 16 and Figure 17 shown.

[0147] Figure 16 The scatter plot in the middle shows the correlation between the color concentration and BOP% in saliva samples. The results show a significant negative correlation (Pearson's r = -0.8279). Each blue dot represents an independent sample, the red regression line is the linear fit result, and the shaded area represents the 95% confidence interval. This trend indicates that the higher the BOP%, the lower the color concentration of the test strip, suggesting that color changes can be used to reflect the degree of gingival inflammation.

[0148] Figure 17 The middle bar graph shows the results of 103 saliva samples (such as Figure 11 The BOP% (red column, left axis, Bleeding on porbing score) and color intensity (blue column, right axis, Color Intensity) are paired and displayed on the horizontal axis (Saliva sample number). In most samples, higher BOP% corresponds to lower color intensity. The purple horizontal line represents the reference threshold for color intensity, approximately 160. The samples on the right are concentrated in the area with lower BOP% and higher color intensity. This trend is consistent with the negative correlation results in the scatter plot, further supporting the feasibility of color intensity as an indicator for assessing the severity of gingival inflammation.

[0149] In summary, the color development distance of the test paper can effectively evaluate the degree of gingival inflammation.

[0150] Experimental Example 4: Diagnostic stability of the test strip at different inflammation levels

[0151] In order to evaluate the diagnostic ability of the test paper on gingival inflammation at different clinical stages, the inventors drew the blood samples before treatment ( Figure 18 ) and 1 month after treatment ( Figure 19 )’s ROC curve.

[0152] The results showed that the color parameters showed good discriminative performance both in the initial screening and diagnosis stage and in the efficacy follow-up stage, especially the AUC of the color distance in both stages was close to or reached 1.0, suggesting that it has stable predictive value at different levels of inflammatory activity.

[0153] The above is only an embodiment of the present invention, and the common knowledge such as the specific technical solutions and / or characteristics in the solution are not described in detail here. It should be pointed out that for those skilled in the art, without departing from the technical solution of the present invention, several variations and improvements can be made, which should also be regarded as the scope of protection of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the description can be used to interpret the content of the claims.

Claims

1. A hemoglobin test strip, characterized in that: The invention comprises a paper-based chip, wherein the paper-based chip is formed by soaking the paper in a mixed solution consisting of TMB, DIP and a buffer system and then drying the paper; the concentration of DIP in the mixed solution is 25-35 mM, the concentration of TMB is 30-37.5 mM, and the buffer system is an acetic acid-sodium acetate buffer system.

2. A hemoglobin test strip according to claim 1, characterized in that: The concentration of acetic acid in the buffer system is 0.05-0.15 M; the concentration of sodium acetate is 0.01-0.05 M; the molar ratio of acetic acid to acetate is 2:1-5:1, and the pH value of the buffer system is 3.5-4.

5.

3. A hemoglobin test strip according to claim 2, characterized in that: It also includes a PVC rubber plate, which includes a top plate and a bottom plate that are fastened together, and the paper-based chip is fixed between the top plate and the bottom plate; the top plate is provided with a strip observation window with an area smaller than the paper-based chip corresponding to the paper-based chip, and scale lines are provided on the side of the strip observation window; the sample addition area of the paper-based chip is located at the low value end of the scale line.

4. Use of the hemoglobin detection test strip according to any one of claims 1 to 3 in detecting hemoglobin in saliva.

5. The use according to claim 4, characterized in that: The steps include: Step 1: Establish a standard curve: Prepare a gradient sample solution containing different standard hemoglobin concentrations using artificial saliva. Add the gradient sample solution to a paper-based chip for reaction, and take photos of the color development of the different gradient sample solutions. Input the color development photos into a computer vision analysis module to obtain the RGB value and color development distance of the color development area. Then, draw linear relationship diagrams and corresponding linear equations for the standard hemoglobin concentration, color intensity, and color development distance. Step 2. Clinical sample testing: Under the same conditions, add the sample to be tested to the paper-based chip for reaction, take a photo to record its color development, input it into the computer vision analysis module, obtain the RGB value and color development distance of the color development area, and calculate the hemoglobin concentration in the sample to be tested based on the linear equation obtained in step 1.

6. The use according to claim 5, characterized in that: In step 1, the concentration of hemoglobin in the gradient sample solution ranges from 0 to 2000 μg / mL.

7. The use according to claim 5, characterized in that: In step 1 and step 2, the reaction time of the gradient sample solution and the sample to be tested with the test paper is 3 to 5 minutes.

8. The use according to claim 5, characterized in that: In step 1, the color-developed photo is input into the computer vision analysis module to obtain the RGB value and color distance of the color-developed area. The analysis steps are as follows: traverse all image files in the specified format and read them into the OpenCV image format BGR; convert the image from BGR to HSV and set the color threshold, find the area in the image that meets the color range, and generate a mask; use morphological operation opening to remove small noise points and retain the larger target area; perform contour detection and draw a positioning rectangular box on the original image as the ROI area; for each positioning marked ROI area, extract the average color value and the color distance of the boxed area.

9. The use according to claim 8, characterized in that: In step 1, after the computer vision analysis module obtains the RGB value of the color display area, the calculation formula of the color intensity in the paper-based chip is: color intensity = 0.3R + 0.59G + 0.11B.

10. The use according to claim 5, characterized in that: In step 1, the linear equations for hemoglobin concentration, color intensity, and color development distance are as follows: Color Intensity = -40.826C Hb +192.929(1); In formula (1), C Hb is the hemoglobin concentration, ranging from 1.6 to 200 μg / mL, R 2 =0.998; Color rendering distance = 0.133C Hb +10.782(2); In formula (2), C Hb is the hemoglobin concentration, ranging from 3 to 200 μg / mL, R 2 =0.972; Color rendering distance = 0.110C Hb +33.752(3); In formula (3), C Hb is the hemoglobin concentration, ranging from 200 to 2000 μg / mL, R 2 =0.992.

Images