A Home-Based Detection Method and System for 25-Hydroxyvitamin D3

Abstract

The application discloses a kind of 25-hydroxyvitamin D3 Home detection method and system, it is related to intelligent skin care product preparation technical field, the method includes: collection fingertip blood sample is applied to disposable electrochemical test strip;Voltammetry scanning obtains original volt-ampere curve;Read test strip batch identification, call corresponding concentration prediction submodel consisting of shared feature extractor and batch adaptation head, obtain preliminary concentration prediction value;Extract the background electrochemical characteristics of non-target potential section Identification matrix type and call compensation parameter to correct;Based on volt-ampere curve morphological characteristics evaluation operation quality and generate confidence score, below threshold when intercept output.The application realizes the accurate detection without user calibration, effectively eliminates batch difference and matrix interference, and real-time interception is carried out to non-standard operation, suitable for vitamin D nutritional status complete management under home scene.

Description

Technical Field

[0001] This invention relates to the field of 25-hydroxyvitamin D3 detection, and more particularly to a home detection method and system for 25-hydroxyvitamin D3. Background Technology

[0002] Serum 25-hydroxyvitamin D3 (25(OH)D3) concentration is widely recognized as the gold standard for assessing vitamin D nutritional status in humans.

[0003] Currently, the main methods for clinical detection of 25(OH)D3 include liquid chromatography-tandem mass spectrometry (LC-MS / MS), chemiluminescence immunoassay (CLIA), and enzyme-linked immunosorbent assay (ELISA). All of these methods have the following drawbacks: complex sample processing, requiring venous blood collection and serum separation by centrifugation, making them unsuitable for home use or in primary care settings without centrifugation facilities; long testing cycles, typically taking several hours to days from blood collection to obtaining a report, hindering immediate decision-making; strong equipment dependence, requiring large instruments and specialized laboratory personnel, making them difficult to implement in primary healthcare institutions and home settings; and low monitoring compliance, requiring regular follow-up examinations after vitamin D supplementation, leading to frequent hospital visits and poor patient adherence.

[0004] In recent years, some progress has been made in the detection technology of 25(OH)D3 based on electrochemical sensors, including aptamer sensors, molecularly imprinted sensors, and immunosensors. However, current electrochemical detection methods still have the following problems: (1) Batch-to-batch variation problem: During the production of screen-printed electrodes, factors such as carbon paste viscosity, identification element fixation density, and nanomaterial coating amount may fluctuate between batches, resulting in significant differences in the response signals of different batches of test strips to the same concentration sample, which affects the accuracy of quantitative results; (2) Matrix effect: 25(OH)D3 is a fat-soluble sterol that is highly bound to vitamin D-binding protein. Abnormal samples such as hyperlipidemia, hemolysis, and high bilirubin levels can significantly interfere with the electron transfer process on the electrode surface, causing the test results to deviate from the true value; (3) Lack of effective operation quality control mechanism: In the home setting, the user's operation standardization cannot be guaranteed, and problems such as insufficient blood volume, incomplete reaction, and abnormal temperature are difficult to be automatically identified. The direct output of low-quality data may mislead clinical decision-making. (4) Interpretation of results: Existing equipment only outputs concentration values ​​and does not provide personalized supplementary suggestions based on individual characteristics.

[0005] Therefore, there is an urgent need for a home testing solution for 25(OH)D3 that combines ease of operation, accuracy of results, and intelligent decision-making. Summary of the Invention

[0006] The purpose of this invention is to provide a home testing method and system for 25-hydroxyvitamin D3, which can achieve accurate home testing of 25-hydroxyvitamin D3 without relying on venous blood collection and professional equipment, while eliminating batch differences and matrix interference, ensuring operational quality, and outputting personalized nutritional intervention plans.

[0007] To achieve the above objectives, the present invention provides a home-based method for detecting 25-hydroxyvitamin D3, comprising the following steps: Collect a blood sample from the user's fingertip and drop the blood sample onto the sampling area of ​​a disposable electrochemical test strip; An electrochemical excitation signal is applied to a disposable electrochemical test strip, and the response signal is acquired by voltammetric scanning to generate the original voltammetric curve. The batch identification information stored on the disposable electrochemical test strip is read, and the corresponding concentration prediction sub-model is called from the pre-built concentration prediction model library based on the batch identification information; the concentration prediction sub-model includes: a shared feature extractor and a batch adapter head; The original voltammetric curve is input into the concentration prediction sub-model, and the general electrochemical features are extracted sequentially through the shared feature extractor. The batch adapter head corrects the general electrochemical features according to the inherent offset characteristics of the batch to obtain the preliminary concentration prediction value. Background electrochemical features of non-target potential regions in the original voltammetric curves are extracted as matrix fingerprints and input into a pre-trained matrix effect recognition model to obtain the sample matrix type. Based on the sample matrix type, the corresponding matrix compensation parameter is called to compensate and correct the preliminary concentration prediction value, and the calibrated 25-hydroxyvitamin D3 concentration is obtained. The operational quality evaluation index is calculated based on the morphological characteristics of the original voltammetric curve, and the confidence score of the test result is generated based on the operational quality evaluation index. When the confidence score is lower than the preset threshold, the concentration detection results are intercepted and a corresponding abnormal prompt is output.

[0008] The voltammetric scanning in this invention employs differential pulse voltammetry or square wave voltammetry, with the potential window set to -0.2V to +1.0V, the sampling interval to 1mV to 5mV, and the detection time to 30 to 90s.

[0009] Preferably, the disposable electrochemical test strip is prepared by screen printing and includes, from bottom to top, a PET insulating substrate, a three-electrode layer, a signal amplification layer, an electronic dielectric layer, a recognition element layer, and a blood filtration layer.

[0010] Preferably, the three-electrode layer comprises: a carbon working electrode, an Ag / AgCl reference electrode, and a carbon counter electrode; The signal amplification layer is formed by modification with at least one of reduced graphene oxide, carbon nanotubes, and gold nanoparticles; The recognition element layer is immobilized with an anti-25-hydroxyvitamin D3 monoclonal antibody or a specific nucleic acid aptamer. The filtration layer is a glass fiber membrane used to separate red blood cells from fingertip blood and allow plasma components to permeate into the reaction zone.

[0011] Preferably, the shared feature extractor is a one-dimensional convolutional neural network jointly trained by the voltammetric curves of multiple disposable electrochemical test strip standards from different production batches under various matrices; the batch adapter is used to correct the sensitivity shift and baseline drift of the corresponding batch of disposable electrochemical test strips. When adding a new production batch, all network parameters of the shared feature extractor are frozen, and the corresponding batch adapter head is trained only using the factory calibration standard data of the new production batch. No manual calibration is required on the user's end.

[0012] Preferably, the non-target potential segment is the potential segment between the voltammetric scan start potential and +0.3V; The matrix types include normal matrix, high-lipid matrix, hemolytic matrix, and high-bilirubin matrix; The matrix classification model identifies matrix types in the following ways: When the background capacitance value of the background electrochemical characteristics decreases by more than the first threshold relative to the factory reference capacitance value, and the broadening factor of the target peak's full width at half maximum (FWHM) relative to the standard FWHM is greater than the second threshold, it is determined to be a high-lipid matrix. When the background electrochemical characteristics show local maxima or inflection points in the potential range of +0.30V to +0.40V, it is identified as a hemolytic matrix. When the absolute value of the baseline slope of the background electrochemical characteristics in the non-target potential region is greater than the third threshold, it is determined to be a high bilirubin matrix. The first threshold, the second threshold, and the third threshold are determined by the matrix classification model during the training phase based on the standard sample set.

[0013] Preferably, the operational quality evaluation indicators include: blood volume adequacy index, reaction completion index, temperature compliance index, and test strip effectiveness index; The blood volume adequacy index is calculated based on the ratio of the absolute value of the peak current at the target potential response peak in the original voltammetric curve to the root mean square of the current noise in the baseline segment; the ratio is the signal-to-noise ratio. The reaction completion index is calculated based on the symmetry factor of the target potential response peak; the symmetry factor is the ratio of the first half width to the second half width of the response peak. The temperature compliance index is calculated based on the difference between the peak potential value of the target potential response peak and the reference peak potential value calibrated at the factory of the disposable electrochemical test strip. The effectiveness index of the test strip is calculated based on the relative deviation between the capacitance value calculated from the background double-layer charging current of the original voltammetry curve and the reference capacitance value calibrated at the factory for disposable electrochemical test strips. The confidence score is obtained by multiplying each operational quality evaluation index by its corresponding preset weight coefficient and then summing the results; the sum of the preset weight coefficients is 1. This invention uses voltammetric curve morphology to evaluate operational quality indicators such as blood volume, reaction completion, temperature, and test strip effectiveness in real time, intercepting low-quality detections, avoiding false alarms, and ensuring operational quality.

[0014] Preferably, the home testing method for 25-hydroxyvitamin D3 further includes the following steps: Obtain the user's individual characteristic data; the individual characteristic data includes at least one of the following: age, body mass index, geographical location information, and concomitant medication information; the geographical location information includes: latitude data or sunshine duration data of the user's location; The calibrated 25-hydroxyvitamin D3 concentration and individual characteristic data are input into the nutrition assessment engine to generate a personalized nutritional status assessment and supplementation recommendations. The supplementation recommendations include: recommended supplementation dosage and follow-up examination cycle. The nutrition assessment engine includes: a rule engine and a neural network. The rule engine has preset concentration grading standards and basic supplementation dosage and basic follow-up examination cycle corresponding to each concentration grading.

[0015] Preferably, the calibrated 25-hydroxyvitamin D3 concentration and individual characteristic data are input into the nutrition assessment engine to generate personalized nutritional status assessment and supplementation recommendations, including the following: The calibrated 25-hydroxyvitamin D3 concentration is input into the rule engine. Based on the concentration range in which the calibrated 25-hydroxyvitamin D3 concentration falls, the corresponding basic supplementation dose and basic review cycle are output. The rule engine has multiple consecutive and non-overlapping concentration ranges preset, and each concentration range corresponds to a basic supplementation dose and a basic review cycle. Individual characteristic data are input into a neural network to obtain dose adjustment coefficients and period adjustment coefficients; both dose adjustment coefficients and period adjustment coefficients are positive numbers within a preset range. Calculate the recommended supplemental dose by multiplying the baseline supplemental dose by the dose adjustment factor; The recommended review cycle is calculated by multiplying the basic review cycle by the cycle adjustment coefficient.

[0016] This invention also provides a home testing system for 25-hydroxyvitamin D3, used to implement the above-mentioned home testing method for 25-hydroxyvitamin D3, comprising: The sample collection module is used to collect the user's fingertip blood sample and drop the fingertip blood sample onto the sampling area of ​​the disposable electrochemical test strip; The signal scanning module is used to apply an electrochemical excitation signal to a disposable electrochemical test strip, acquire the response signal by voltammetric scanning, and generate the original voltammetric curve. An artificial intelligence processing module is used to read the batch identification information stored on disposable electrochemical test strips, and call the corresponding concentration prediction sub-model from a pre-built concentration prediction model library based on the batch identification information. The concentration prediction sub-model includes a shared feature extractor and a batch adapter. The module is used to input the original voltammetric curve into the concentration prediction sub-model, extract general electrochemical features sequentially through the shared feature extractor, and correct the general electrochemical features according to the inherent offset characteristics of the batch to obtain a preliminary concentration prediction value. It is also used to extract background electrochemical features from the non-target potential region of the original voltammetric curve as a matrix fingerprint, input it into a pre-trained matrix effect recognition model to obtain the sample matrix type, and call the corresponding matrix compensation parameters according to the sample matrix type to compensate and correct the preliminary concentration prediction value to obtain the calibrated 25-hydroxyvitamin D3 concentration. The output module is used to calculate the operation quality evaluation index based on the morphological characteristics of the original voltammetric curve, and generate a confidence score for the detection result based on the operation quality evaluation index; when the confidence score is lower than the preset threshold, the concentration detection result is intercepted and a corresponding abnormal prompt is output.

[0017] Preferably, the artificial intelligence processing module includes: an information recognition module, a preliminary prediction module, a matrix type determination module, and a compensation and correction module; The information identification module is used to read the batch identification information stored on the disposable electrochemical test strip, and according to the batch identification information, call the corresponding concentration prediction sub-model from the pre-built concentration prediction model library; the concentration prediction sub-model includes: a shared feature extractor and a batch adapter head; The preliminary prediction module is used to input the original voltammetric curve into the concentration prediction sub-model, and then extract general electrochemical features through the shared feature extractor. The batch adapter head corrects the general electrochemical features according to the inherent offset characteristics of the batch to obtain the preliminary concentration prediction value. The matrix type determination module is used to extract the background electrochemical features of the non-target potential region in the original voltammetric curve as a matrix fingerprint, which is then input into a pre-trained matrix effect recognition model to obtain the sample matrix type. The compensation and correction module is used to call the corresponding matrix compensation parameters according to the sample matrix type to compensate and correct the preliminary concentration prediction value, so as to obtain the calibrated 25-hydroxyvitamin D3 concentration.

[0018] In summary, the home detection method and system for 25-hydroxyvitamin D3 provided by this invention has the following advantages compared with traditional technologies: By using a concentration prediction sub-model architecture with a shared feature extractor and batch adapter head, combined with automatic batch identification, the consistency of detection results for disposable electrochemical test strips from different production batches is achieved, with the coefficient of variation controlled within 8%, eliminating the need for manual calibration by the user and removing batch-to-batch differences; By automatically identifying the sample matrix type through the background electrochemical characteristics of non-target potential regions, and dynamically calling the compensation model for correction, the detection deviation of abnormal samples is reduced from ±25% in traditional methods to within ±10%, and this compensation matrix effect improves the accuracy of the results; This invention integrates detection, evaluation, intervention, and retesting, allowing users to obtain complete health management services such as concentration values, personalized supplementation plans, and follow-up reminders without going to the hospital. It enables home-based testing management and significantly reduces the cost per test compared to LC-MS / MS and chemiluminescence methods. This makes it suitable for large-scale grassroots screening and high-frequency home monitoring, thus lowering testing costs.

[0019] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0020] Figure 1 This is a flowchart of the home testing method for 25-hydroxyvitamin D3 in this invention; Figure 2 This is a schematic diagram of the structure of the disposable electrochemical test strip in this invention; Figure 3 This is a flowchart of the AI ​​three-layer dynamic calibration process in this invention; Figure 4 This is a comparison diagram of the characteristics of current-voltage curves for different matrix types in this invention; Figure 5 These are schematic diagrams of personalized nutrition assessment in Embodiments 1 and 2 of the present invention; Figure 6 This is a schematic diagram of the modules of the home detection system for 25-hydroxyvitamin D3 in this invention.

[0021] Figure label: 1. PET insulating substrate; 2. Three-electrode layer; 3. Signal amplification layer; 4. Electron mediator layer; 5. Identification element layer; 6. Blood filtration layer; 7. Sampling area; 8. NFC chip. Detailed Implementation

[0022] The technical method of the present invention will be further described below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of this application.

[0023] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the scope of this application and its application or use.

[0024] Techniques, systems, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, they should be considered part of the instruction manual.

[0025] In all the examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0026] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0027] This invention provides a home-based method for detecting 25-hydroxyvitamin D3, such as... Figure 1 As shown, it includes the following steps: S1. Collect a blood sample from the user's fingertip and add the blood sample to the sampling area of ​​a disposable electrochemical test strip.

[0028] Among them, the disposable electrochemical test strips are prepared using a screen printing process, such as... Figure 2 As shown, from bottom to top, it includes: PET insulating substrate 1, three-electrode layer 2, signal amplification layer 3, electronic mediator layer 4, identification element layer 5, blood filtration layer 6, sampling area 7, and NFC chip 8.

[0029] The three electrode layers consist of a carbon working electrode, an Ag / AgCl reference electrode, and a carbon counter electrode.

[0030] The signal amplification layer is formed by modification with at least one of reduced graphene oxide, carbon nanotubes and gold nanoparticles to enhance electron transfer efficiency.

[0031] Electron mediator layer coated with [Fe(CN)6] 3- / 4- Redox pairs serve as probes for indirect electrochemical detection.

[0032] The recognition element layer is immobilized with anti-25-hydroxyvitamin D3 monoclonal antibody or specific nucleic acid aptamer (such as truncated aptamer VDBA14-35) to specifically capture 25(OH)D3 in the sample.

[0033] The filtration layer is a glass fiber membrane used to separate red blood cells from fingertip blood and allow plasma components to permeate into the reaction zone.

[0034] NFC chip: Embedded at the end of disposable electrochemical test strips, storing batch number, factory calibration parameters for that batch (standard curve slope, intercept, background double layer capacitance reference value), and expiration date.

[0035] The detection principle of disposable electrochemical test strips is that after 25(OH)D3 combines with the recognition element, it forms an insulating complex on the electrode surface, hindering the reaction of [Fe(CN)6]. 3- / 4- Electron transfer leads to a decrease in redox peak current or an increase in interfacial impedance. The current decrease is quantitatively related to the concentration of 25(OH)D3.

[0036] In an exemplary embodiment of the present invention, the preparation steps of the disposable electrochemical test strip are as follows: Three-electrode fabrication: A silver paste conductive layer, a carbon paste working electrode and a counter electrode, and an Ag / AgCl paste reference electrode were sequentially printed on a 0.2 mm thick PET substrate using screen printing technology. After drying and curing, a three-electrode system was formed, in which the effective working electrode has a diameter of 3 mm.

[0037] Signal amplification layer modification: A 1 mg / mL reduced graphene oxide (rGO) dispersion was drop-coated onto the surface of the working electrode and dried at room temperature. Then, a gold nanoparticle (AuNPs) dispersion was drop-coated to form an rGO / AuNPs composite modification layer, which increases the specific surface area of ​​the electrode and accelerates electron transfer.

[0038] Recognition element immobilization: Take 0.5 mg / mL of anti-25(OH)D3 monoclonal antibody (or 1 μM of DNA aptamer VDBA14-35) and drop it onto the modified working electrode surface. Incubate at 4 °C for 12 h to immobilize the antibody / aptamer on the electrode surface through physical adsorption and covalent bonding.

[0039] Electron mediator layer coating: A coating containing 5 mM [Fe(CN)6] 3- / 4- Chitosan solution was drop-coated onto the recognition element layer to form an electronic mediator and an anti-fouling protective layer, and then air-dried at room temperature.

[0040] Blood filtration layer assembly: Cut the glass fiber blood filtration membrane into 5mm×5mm pieces and attach it to the sampling end of the test strip to completely cover the reaction area for filtering red blood cells from fingertip blood.

[0041] NFC chip writing: Write the test strip batch number (e.g., “VD202509-A”), the measured response values ​​of 20 standard concentration points of the batch, the background capacitance reference value, and the expiration date into the NFC chip embedded at the end of the test strip.

[0042] S2. Apply an electrochemical excitation signal to the disposable electrochemical test strip, collect the response signal by voltammetric scanning, and generate the original voltammetric curve.

[0043] In an exemplary embodiment of this invention, a handheld reader is used to apply an electrochemical excitation signal to a disposable electrochemical test strip, and a voltammetric scan is used to acquire the response signal, generating a raw voltammetric curve, thus obtaining the raw voltammetric curve. The handheld reader integrates an electrochemical workstation module, an NFC reading module, a Bluetooth communication module, and a power management module. During detection, the handheld reader automatically identifies the batch information of the test strip via NFC, and performs scanning using differential pulse voltammetry or square wave voltammetry. The potential window is set to -0.2V to +1.0V (relative to the Ag / AgCl reference electrode), the sampling interval is 2mV, a single curve has approximately 600 data points, and the detection time is 30 to 90 seconds.

[0044] The AI ​​three-layer dynamic calibration process of this invention, such as Figure 3 As shown, the specific content includes the following steps S3 to S7.

[0045] S3. Read the batch identification information stored on the disposable electrochemical test strip, and call the corresponding concentration prediction sub-model from the pre-built concentration prediction model library based on the batch identification information. The concentration prediction sub-model includes: a shared feature extractor and a batch adapter head.

[0046] The shared feature extractor is a one-dimensional convolutional neural network (1D-CNN) jointly trained from the voltammetric curves of multiple disposable electrochemical test strip standards from different production batches under various matrices. The batch adapter is used to correct the sensitivity shift and baseline drift of the corresponding batch of disposable electrochemical test strips, achieving accurate matching of "one model per batch".

[0047] When adding a new production batch, all network parameters of the shared feature extractor are frozen, and the corresponding batch adapter head is trained only using the factory calibration standard data (50~100 standard concentration points) of the new production batch, without requiring manual calibration by the user.

[0048] S4. Input the original voltammetric curve into the concentration prediction sub-model, and extract the general electrochemical features sequentially through the shared feature extractor. The batch adapter head corrects the general electrochemical features according to the inherent offset characteristics of the batch to obtain the preliminary concentration prediction value.

[0049] S5. Extract the background electrochemical features of the non-target potential region in the original voltammetric curve as the matrix fingerprint, input it into the pre-trained matrix effect recognition model, and obtain the sample matrix type.

[0050] The non-target potential segment is the potential segment between the starting potential of the voltammetric scan and +0.3V.

[0051] The matrix types include normal matrix, high-lipid matrix, hemolyzed matrix, and high-bilirubin matrix. A comparison of the voltammetric curves for each matrix is ​​shown below. Figure 4 As shown, the matrix classification model identifies matrix types in the following ways: When the background capacitance value of the background electrochemical characteristics decreases by more than the first threshold relative to the factory reference capacitance value, and the broadening factor of the target peak's full width at half maximum (FWHM) relative to the standard FWHM is greater than the second threshold, it is determined to be a high-lipid matrix.

[0052] When the background electrochemical characteristics show local maxima or inflection points in the potential range of +0.30V to +0.40V, it is identified as a hemolytic matrix.

[0053] When the absolute value of the baseline slope of the background electrochemical characteristics in the non-target potential region is greater than the third threshold, it is determined to be a high bilirubin matrix.

[0054] The first threshold, the second threshold, and the third threshold are determined by the matrix classification model during the training phase based on the standard sample set.

[0055] S6. Based on the sample matrix type, call the corresponding matrix compensation parameter to compensate and correct the preliminary concentration prediction value (gain correction or baseline subtraction) to obtain the calibrated 25-hydroxyvitamin D3 concentration.

[0056] S7. Calculate the operation quality evaluation index based on the morphological characteristics of the original voltammetric curve, and generate the confidence score of the test result based on the operation quality evaluation index.

[0057] The operational quality evaluation indicators include: blood volume adequacy, reaction completion, temperature compliance, and test strip effectiveness.

[0058] The blood volume adequacy index is calculated based on the ratio (signal-to-noise ratio) of the absolute value of the peak current at the target potential response peak in the original volt-ampere curve to the root mean square of the current noise in the baseline section.

[0059] The response completion index is calculated based on the symmetry factor of the target potential response peak. The symmetry factor is the ratio of the first half width to the second half width of the response peak.

[0060] The temperature compliance index is calculated based on the difference between the peak potential value of the target potential response peak and the reference peak potential value calibrated at the factory for disposable electrochemical test strips.

[0061] The validity index of the test strip is calculated based on the relative deviation between the capacitance value calculated from the background double-layer charging current of the original volt-ampere curve and the reference capacitance value calibrated at the factory for disposable electrochemical test strips.

[0062] The operational quality evaluation indicators are based on the morphological basis of the original voltammetric curve and the qualification standards, as shown in Table 1.

[0063] Table 1. Morphological basis and qualification standards of operation quality evaluation indicators based on the original voltammetric curve.

[0064] The confidence score is obtained by summing the results of multiplying each operational quality evaluation indicator by its corresponding preset weight coefficient. The sum of the preset weight coefficients is 1. The expression for the confidence score is: ; in, H For confidence scoring, α is the weighting coefficient for blood volume adequacy, a is the blood volume adequacy index, β is the weighting coefficient for response completion, b is the response completion index, γ is the weighting coefficient for temperature compliance, and c is the temperature compliance index. denoted as the test strip effectiveness weighting coefficient, and d as the test strip effectiveness index.

[0065] In an exemplary embodiment of the present invention, the training data for the shared feature extractor, batch adapter training, and matrix effect recognition model are as follows: (1) Training of shared feature extractor: Collect 50 disposable electrochemical test strips from different production batches. For each batch, 10 concentration gradients of 25(OH)D3 standard (0, 5, 10, 20, 30, 50, 75, 100, 150, 200 ng / mL) were tested under 5 matrix types (normal, high lipid, hemolysis, high protein, high bilirubin) to obtain a total of 2500 voltammetric curves as training set.

[0066] A 1D-CNN architecture was used to construct a shared feature extractor. The network structure was as follows: Input layer: [batch, 1, 600]; Convolutional layer 1: 64 channels, kernel size 11, ReLU activation; Max pooling layer 1: pooling size 2; Convolutional layer 2: 128 channels, kernel size 7, ReLU activation; Max pooling layer 2: pooling size 2; Convolutional layer 3: 256 channels, kernel size 5, ReLU activation; Global average pooling; Output: 128-dimensional feature vector. The Adam optimizer was used for training for 30 epochs with an initial learning rate of 0.001.

[0067] (2) Batch adapter training: For each production batch, all parameters of the shared feature extractor are frozen, and the adapter is trained only for that batch. The batch adapter adopts a 2-layer fully connected structure: 128→64→1, and outputs the concentration prediction value. Each batch is trained for 10 epochs using 50~100 factory calibration data points, and then stored in the model library after completion.

[0068] (3) Training of matrix effect recognition model: Extract 250 current data points in the range of -0.2V to +0.3V from all training samples, label the corresponding matrix type, train a 1D-CNN classifier with a structure of 3 convolutional layers, global pooling and softmax output layer to achieve 5 classifications (normal, high lipid, hemolysis, high protein, high bilirubin).

[0069] (4) Setting of operational quality assessment thresholds: Based on the statistical distribution of the training set, the qualified thresholds for four quality indicators are set: the blood volume adequacy threshold is a signal-to-noise ratio (SNR) of 10; the reaction completion threshold is a target peak symmetry factor of 0.9~1.2; and the temperature compliance threshold is a peak potential shift |ΔE. p |<20mV; The freshness threshold for the test strip is a background capacitance deviation of <30%.

[0070] S8. When the confidence score is lower than the preset threshold, intercept the concentration detection result and output the corresponding abnormal prompt. For example, when the confidence score is lower than the preset threshold (e.g., 70%), refuse to output the concentration result and prompt the user with the specific reason (e.g., insufficient blood volume, please resample).

[0071] In an exemplary embodiment of the present invention, the home testing method for 25-hydroxyvitamin D3 further includes the following steps: S9. Obtain the user's individual characteristic data. This individual characteristic data includes at least one of the following: age, body mass index, geographic location information, and concomitant medication information. Geographic location information includes: latitude data or sunshine duration data of the user's location.

[0072] Age: Recommended dosages differ for children, adults, and the elderly.

[0073] Body Mass Index (BMI): Obese individuals (BMI>30) have an increased volume of distribution of 25(OH)D3, requiring 2 to 3 times the supplemental dose.

[0074] Geographic location information: High latitude regions with insufficient sunlight in winter; it is recommended to increase the supplemental dosage.

[0075] Combined medication information: Antiepileptic drugs (such as carbamazepine, phenytoin, phenobarbital) can accelerate the metabolism of 25(OH)D3, requiring increased supplemental doses.

[0076] S10. Input the calibrated 25-hydroxyvitamin D3 concentration and individual characteristic data into the nutrition assessment engine to generate a personalized nutritional status assessment and supplementation recommendations. These recommendations include: recommended supplementation dosage and follow-up examination frequency.

[0077] The nutrition assessment engine includes a rule engine and a neural network. The rule engine has preset concentration grading standards and corresponding basic supplement dosages and basic follow-up examination cycles for each concentration grading.

[0078] Specifically, step S10 includes the following steps: S101. Input the calibrated 25-hydroxyvitamin D3 concentration into the rule engine. Based on the concentration range of the calibrated 25-hydroxyvitamin D3 concentration, output the corresponding basic supplementation dose and basic review cycle. The rule engine has multiple preset continuous and non-overlapping concentration ranges, each corresponding to a basic supplementation dose and a basic review cycle.

[0079] S102. Input the individual characteristic data into the neural network to obtain the dose adjustment coefficient and the period adjustment coefficient. Both the dose adjustment coefficient and the period adjustment coefficient are positive numbers within a preset range.

[0080] S103. Multiply the basic supplementary dose by the dose adjustment factor to calculate the recommended supplementary dose; multiply the basic follow-up examination cycle by the cycle adjustment factor to calculate the recommended follow-up examination cycle.

[0081] The concentration of 25(OH)D3 and the comparison with nutritional status are shown in Table 2.

[0082] Table 2. Comparison of 25(OH)D3 concentration and nutritional status

[0083] In an exemplary embodiment of the present invention, a user's 25(OH)D3 level is 18 ng / mL, indicating a 25(OH)D3 deficiency. Considering the user's BMI of 32 and current carbamazepine use, it is recommended to supplement with 6000 IU of 25(OH)D3 daily, with a follow-up examination after 8 weeks. Another user's 25(OH)D3 level is 45 ng / mL, indicating sufficient 25(OH)D3 and an effective current supplementation regimen. It is recommended to maintain a daily supplementation of 2000 IU, with a follow-up examination after 6 months.

[0084] The specific embodiments provided by this invention are as follows, with a 35-year-old female user as an example for testing and explanation: Example 1 The female user had a BMI of 31, lived in a high-latitude region, and was suspected of having vitamin D deficiency due to fatigue and bone pain. The following home tests were conducted: Sampling: After cleaning her fingertip, the female user collected approximately 20 μL of blood from her fingertip using a lancet and added it to the sampling area of ​​a disposable electrochemical test strip. After the blood was filtered through a blood filtration membrane, the plasma permeated into the reaction layer.

[0085] Testing: Insert the disposable electrochemical test strip with a drop of blood from the fingertip into the handheld reader. The handheld reader reads the batch identification information "VD202509-A" via NFC and calls the corresponding concentration prediction sub-model from the pre-built concentration prediction model library. Start the DPV scan (-0.2V to +1.0V, 2mV increments), and obtain the raw voltammetric curve after 60 seconds.

[0086] AI calibration: Batch calibration: The shared feature extractor extracts a 128-dimensional general electrochemical feature vector from the original voltammetric curve. The batch adapter “VD202509-A” is corrected according to the inherent offset characteristics of the batch and outputs a preliminary concentration prediction value.

[0087] Matrix identification: Extract matrix fingerprints from non-target potential regions (-0.2V to +0.3V), the matrix classification model outputs "normal matrix", and the standard sub-model is called.

[0088] Quality Assessment: Four operational quality evaluation indicators were calculated: SNR was 15 (pass), symmetry factor was 1.05 (pass), peak potential offset was 8mV (pass), and capacitance deviation was 12% (pass). All four indicators were satisfactory. The confidence score was 95% (high confidence).

[0089] Concentration output: The output calibrated 25(OH)D3 concentration is 16 ng / mL.

[0090] Personalized assessment: We obtained the user's individual characteristics data. The female user is 35 years old, has a BMI of 31, lives in a high-latitude region, and has no concurrent medications.

[0091] The calibrated 25-hydroxyvitamin D3 concentration is input into the rule engine, which has multiple preset continuous and non-overlapping concentration ranges (<20 ng / mL, 20~30 ng / mL, ≥30 ng / mL). Each concentration range corresponds to a basal supplement dose and a basal follow-up cycle. Based on the concentration range (<20 ng / mL) in which the calibrated 25-hydroxyvitamin D3 concentration (16 ng / mL) falls, the rule engine outputs the corresponding basal supplement dose (2000 IU / day) and basal follow-up cycle (8 weeks).

[0092] Individual characteristic data (age 35, BMI 31, high-latitude region, no concurrent medication) are input into the neural network to obtain the dose adjustment coefficient. Both the dose adjustment coefficient and the periodic adjustment coefficient are positive numbers within a preset range. The preferred value range for the dose adjustment coefficient is 0.5 to 3.0, and the preferred value range for the periodic adjustment coefficient is 0.5 to 2.0. In this embodiment, the dose adjustment coefficient is 3.0 and the periodic adjustment coefficient is 1.0.

[0093] Multiply the baseline supplemental dose (2000 IU / day) by the dose adjustment factor (3.0) to calculate the recommended supplemental dose as 6000 IU / day. Multiply the baseline follow-up interval (8 weeks) by the interval adjustment factor (1.0) to calculate the recommended follow-up interval as 8 weeks.

[0094] Output suggestions, such as Figure 5 As shown, "Your 25-hydroxyvitamin D3 level is 16 ng / mL, which indicates a deficiency. Considering your high body mass index and the fact that you are in a high-latitude region during winter, it is recommended that you supplement with 6000 IU of 25-hydroxyvitamin D3 daily (or 50000 IU weekly), while also increasing your outdoor sun exposure. Please have a follow-up examination after 8 weeks."

[0095] Data synchronization: Test results, assessment suggestions, and follow-up reminders are automatically synchronized to the cloud-based health management platform, which users can authorize their family doctor or online doctor to view.

[0096] Example 2 The detection of abnormal matrix samples: This example describes the processing procedure for a high-lipid matrix sample.

[0097] During the user's follow-up examination, the plasma was milky due to a high-fat diet the previous day. After obtaining the original voltammetric curve according to the steps in Example 1, the AI ​​calibration process is as follows: Matrix identification: The background capacitance in the background region (-0.2V~+0.3V) was identified as 22% lower than the baseline, with a peak broadening factor of 1.35. The matrix classification model outputs "high-lipid matrix".

[0098] Matrix compensation: The high-fat compensation sub-model is invoked to increase the prediction gain coefficient by 18%.

[0099] Quality assessment: Except for peak broadening, all other indicators are normal (SNR is 14, symmetry factor is 1.10, peak potential offset is 5mV, capacitance deviation is 8%), confidence score is 82% (medium confidence), which is higher than the preset threshold of 70%, and output is allowed.

[0100] Concentration output: The original predicted value without compensation was 21 ng / mL, which was corrected to 26 ng / mL after high-fat compensation.

[0101] Personalized assessment: A concentration of 26 ng / mL falls within the insufficient range (20-30 ng / mL). The rule engine outputs a baseline supplemental dose of 1000 IU / day and a baseline follow-up interval of 12 weeks. The neural network outputs a dose adjustment factor of 2.0 and a cycle adjustment factor of 1.0 based on individual characteristics (BMI of 31, high latitude region). The recommended supplemental dose is 2000 IU / day, and the recommended follow-up interval is 12 weeks.

[0102] Output suggestions, such as Figure 5 As shown, "Your vitamin D level has improved (from 16 to 26 ng / mL), but it is still below the target. It is recommended to continue to maintain 2000 IU daily and have a follow-up test in 3 months. The sample for this test was in a high-fat state, and the result has been automatically corrected. It is recommended to retest on an empty stomach to confirm."

[0103] Example 3 In the scenario of operation quality interception, this embodiment is an interception mechanism for low-quality detection.

[0104] The user had insufficient blood volume (only about 5 μL) when taking the sample, and the test was performed according to the steps in Example 1.

[0105] Quality Assessment: Four operational quality evaluation indicators were calculated: SNR was 6 (unacceptable, threshold 10), symmetry factor confinement was 0.85 (unacceptable, threshold 0.9~1.2), peak potential offset was 25mV (unacceptable, threshold <20mV), and capacitance deviation was 35% (unacceptable, threshold <30%). All four indicators failed to meet the standards.

[0106] Confidence score: The four indicators were multiplied by their respective weights (0.25, 0.20, 0.30, 0.25) and then summed. The score was 38%, which is lower than the preset threshold of 70%.

[0107] The system refused to output the concentration result and displayed an error message to the user: "Detection quality is substandard. Possible cause: insufficient blood volume (low signal-to-noise ratio). Please resample and ensure sufficient blood volume before testing again."

[0108] This invention provides a home testing system for 25-hydroxyvitamin D3, used to implement the aforementioned home testing method for 25-hydroxyvitamin D3, such as... Figure 6 As shown, it includes: The sample collection module is used to collect the user's fingertip blood sample and drop the fingertip blood sample onto the sampling area of ​​a disposable electrochemical test strip.

[0109] The signal scanning module is used to apply an electrochemical excitation signal to a disposable electrochemical test strip, acquire the response signal by voltammetric scanning, and generate the original voltammetric curve.

[0110] The artificial intelligence processing module specifically includes: an information recognition module, a preliminary prediction module, a matrix type determination module, and a compensation and correction module.

[0111] The information identification module reads the batch identification information stored on the disposable electrochemical test strip and, based on this information, calls the corresponding concentration prediction sub-model from a pre-built concentration prediction model library. The concentration prediction sub-model includes a shared feature extractor and a batch adapter head.

[0112] The preliminary prediction module is used to input the original voltammetric curve into the concentration prediction sub-model, and then extract general electrochemical features through the shared feature extractor. The batch adapter head corrects the general electrochemical features according to the inherent offset characteristics of the batch to obtain the preliminary concentration prediction value.

[0113] The matrix type determination module is used to extract the background electrochemical features of the non-target potential region in the original voltammetric curve as a matrix fingerprint, which is then input into the pre-trained matrix effect recognition model to obtain the sample matrix type.

[0114] The compensation and correction module is used to call the corresponding matrix compensation parameters according to the sample matrix type to compensate and correct the preliminary concentration prediction value, so as to obtain the calibrated 25-hydroxyvitamin D3 concentration.

[0115] The output module is used to calculate the operation quality evaluation index based on the morphological characteristics of the original voltammetric curve, and generate a confidence score for the detection result based on the operation quality evaluation index; when the confidence score is lower than the preset threshold, the concentration detection result is intercepted and a corresponding abnormal prompt is output.

[0116] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A home-based method for detecting 25-hydroxyvitamin D3, characterized in that, Includes the following steps: Collect a blood sample from the user's fingertip and drop the blood sample onto the sampling area of ​​a disposable electrochemical test strip; An electrochemical excitation signal is applied to a disposable electrochemical test strip, and the response signal is acquired by voltammetric scanning to generate the original voltammetric curve. The batch identification information stored on the disposable electrochemical test strip is read, and the corresponding concentration prediction sub-model is called from the pre-built concentration prediction model library based on the batch identification information; the concentration prediction sub-model includes: a shared feature extractor and a batch adapter head; The original voltammetric curve is input into the concentration prediction sub-model, and the general electrochemical features are extracted sequentially through the shared feature extractor. The batch adapter head corrects the general electrochemical features according to the inherent offset characteristics of the batch to obtain the preliminary concentration prediction value. Background electrochemical features of non-target potential regions in the original voltammetric curves are extracted as matrix fingerprints and input into a pre-trained matrix effect recognition model to obtain the sample matrix type. Based on the sample matrix type, the corresponding matrix compensation parameter is called to compensate and correct the preliminary concentration prediction value, and the calibrated 25-hydroxyvitamin D3 concentration is obtained. The operational quality evaluation index is calculated based on the morphological characteristics of the original voltammetric curve, and the confidence score of the test result is generated based on the operational quality evaluation index. When the confidence score is lower than the preset threshold, the concentration detection results are intercepted and a corresponding abnormal prompt is output.

2. The home testing method for 25-hydroxyvitamin D3 according to claim 1, characterized in that, The disposable electrochemical test strip is prepared by screen printing and includes, from bottom to top: a PET insulating substrate, a three-electrode layer, a signal amplification layer, an electronic dielectric layer, a recognition element layer, and a blood filtration layer.

3. The home testing method for 25-hydroxyvitamin D3 according to claim 2, characterized in that, The three-electrode layer includes: a carbon working electrode, an Ag / AgCl reference electrode, and a carbon counter electrode; The signal amplification layer is formed by modification with at least one of reduced graphene oxide, carbon nanotubes, and gold nanoparticles; The recognition element layer is immobilized with an anti-25-hydroxyvitamin D3 monoclonal antibody or a specific nucleic acid aptamer. The filtration layer is a glass fiber membrane used to separate red blood cells from fingertip blood and allow plasma components to permeate into the reaction zone.

4. The home testing method for 25-hydroxyvitamin D3 according to claim 1, characterized in that, The shared feature extractor is a one-dimensional convolutional neural network jointly trained by the voltammetric curves of multiple disposable electrochemical test strip standards from different production batches under various matrices; the batch adapter is used to correct the sensitivity shift and baseline drift of the corresponding batch of disposable electrochemical test strips. When adding a new production batch, all network parameters of the shared feature extractor are frozen, and the corresponding batch adapter head is trained only using the factory calibration standard data of the new production batch.

5. The home testing method for 25-hydroxyvitamin D3 according to claim 1, characterized in that, The non-target potential range is the potential range between the voltammetric scan start potential and +0.3V; The matrix types include normal matrix, high-lipid matrix, hemolytic matrix, and high-bilirubin matrix; The matrix classification model identifies matrix types in the following ways: When the background capacitance value of the background electrochemical characteristics decreases by more than the first threshold relative to the factory reference capacitance value, and the broadening factor of the target peak's full width at half maximum (FWHM) relative to the standard FWHM is greater than the second threshold, it is determined to be a high-lipid matrix. When the background electrochemical characteristics show local maxima or inflection points in the potential range of +0.30V to +0.40V, it is identified as a hemolytic matrix. When the absolute value of the baseline slope of the background electrochemical characteristics in the non-target potential region is greater than the third threshold, it is determined to be a high bilirubin matrix. The first threshold, the second threshold, and the third threshold are determined by the matrix classification model during the training phase based on the standard sample set.

6. The home testing method for 25-hydroxyvitamin D3 according to claim 1, characterized in that, The operational quality evaluation indicators include: blood volume adequacy, reaction completion, temperature compliance, and test strip effectiveness. The blood volume adequacy index is calculated based on the ratio of the absolute value of the peak current at the target potential response peak in the original voltammetric curve to the root mean square of the current noise in the baseline segment; the ratio is the signal-to-noise ratio. The reaction completion index is calculated based on the symmetry factor of the target potential response peak; the symmetry factor is the ratio of the first half width to the second half width of the response peak. The temperature compliance index is calculated based on the difference between the peak potential value of the target potential response peak and the reference peak potential value calibrated at the factory of the disposable electrochemical test strip. The effectiveness index of the test strip is calculated based on the relative deviation between the capacitance value calculated from the background double-layer charging current of the original voltammetry curve and the reference capacitance value calibrated at the factory for disposable electrochemical test strips. The confidence score is obtained by multiplying each operational quality evaluation index by its corresponding preset weight coefficient and then summing the results; the sum of the preset weight coefficients is 1.

7. The home testing method for 25-hydroxyvitamin D3 according to claim 1, characterized in that, The home testing method for 25-hydroxyvitamin D3 also includes the following steps: Obtain the user's individual characteristic data; the individual characteristic data includes at least one of the following: age, body mass index, geographical location information, and concomitant medication information; the geographical location information includes: latitude data or sunshine duration data of the user's location; The calibrated 25-hydroxyvitamin D3 concentration and individual characteristic data are input into the nutrition assessment engine to generate a personalized nutritional status assessment and supplementation recommendations. The supplementation recommendations include: recommended supplementation dosage and follow-up examination cycle. The nutrition assessment engine includes: a rule engine and a neural network. The rule engine has preset concentration grading standards and basic supplementation dosage and basic follow-up examination cycle corresponding to each concentration grading.

8. The home testing method for 25-hydroxyvitamin D3 according to claim 7, characterized in that, The calibrated 25-hydroxyvitamin D3 concentration and individual characteristic data are input into the nutrition assessment engine to generate personalized nutritional status assessments and supplementation recommendations, including: The calibrated 25-hydroxyvitamin D3 concentration is input into the rule engine. Based on the concentration range in which the calibrated 25-hydroxyvitamin D3 concentration falls, the corresponding basic supplementation dose and basic review cycle are output. The rule engine has multiple consecutive and non-overlapping concentration ranges preset, and each concentration range corresponds to a basic supplementation dose and a basic review cycle. Individual characteristic data are input into a neural network to obtain dose adjustment coefficients and period adjustment coefficients; both dose adjustment coefficients and period adjustment coefficients are positive numbers within a preset range. Calculate the recommended supplemental dose by multiplying the baseline supplemental dose by the dose adjustment factor; The recommended review cycle is calculated by multiplying the basic review cycle by the cycle adjustment coefficient.

9. A home detection system for 25-hydroxyvitamin D3, characterized in that, A home testing method for 25-hydroxyvitamin D3 as described in any one of claims 1-8, comprising: The sample collection module is used to collect the user's fingertip blood sample and drop the fingertip blood sample onto the sampling area of ​​the disposable electrochemical test strip; The signal scanning module is used to apply an electrochemical excitation signal to a disposable electrochemical test strip, acquire the response signal by voltammetric scanning, and generate the original voltammetric curve. An artificial intelligence processing module is used to read the batch identification information stored on disposable electrochemical test strips, and call the corresponding concentration prediction sub-model from a pre-built concentration prediction model library based on the batch identification information. The concentration prediction sub-model includes a shared feature extractor and a batch adapter. The module is used to input the original voltammetric curve into the concentration prediction sub-model, extract general electrochemical features sequentially through the shared feature extractor, and correct the general electrochemical features according to the inherent offset characteristics of the batch to obtain a preliminary concentration prediction value. It is also used to extract background electrochemical features from the non-target potential region of the original voltammetric curve as a matrix fingerprint, input it into a pre-trained matrix effect recognition model to obtain the sample matrix type, and call the corresponding matrix compensation parameters according to the sample matrix type to compensate and correct the preliminary concentration prediction value to obtain the calibrated 25-hydroxyvitamin D3 concentration. The output module is used to calculate the operation quality evaluation index based on the morphological characteristics of the original voltammetric curve, and generate a confidence score for the detection result based on the operation quality evaluation index; when the confidence score is lower than the preset threshold, the concentration detection result is intercepted and a corresponding abnormal prompt is output.

10. The home detection system for 25-hydroxyvitamin D3 according to claim 9, characterized in that, The artificial intelligence processing module includes: an information recognition module, a preliminary prediction module, a matrix type determination module, and a compensation and correction module; The information identification module is used to read the batch identification information stored on the disposable electrochemical test strip, and according to the batch identification information, call the corresponding concentration prediction sub-model from the pre-built concentration prediction model library; the concentration prediction sub-model includes: a shared feature extractor and a batch adapter head; The preliminary prediction module is used to input the original voltammetric curve into the concentration prediction sub-model, and then extract general electrochemical features through the shared feature extractor. The batch adapter head corrects the general electrochemical features according to the inherent offset characteristics of the batch to obtain the preliminary concentration prediction value. The matrix type determination module is used to extract the background electrochemical features of the non-target potential region in the original voltammetric curve as a matrix fingerprint, which is then input into a pre-trained matrix effect recognition model to obtain the sample matrix type. The compensation and correction module is used to call the corresponding matrix compensation parameters according to the sample matrix type to compensate and correct the preliminary concentration prediction value, so as to obtain the calibrated 25-hydroxyvitamin D3 concentration.

Images