Heparin sodium detection method and system based on deep learning

Abstract

The invention provides a heparin sodium detection method and system based on deep learning, which are suitable for the technical field of data processing, and the method comprises the following steps: carrying out feature extraction on heparin sodium detection spectrum signal information and heparin sodium detection electrochemical signal information; obtaining heparin sodium detection spectral signal feature information and heparin sodium detection electrochemical signal feature information; according to the heparin sodium detection spectral signal feature information, the heparin sodium detection electrochemical signal feature information, the heparin sodium detection signal feature optimization iteration times and the heparin sodium detection signal feature optimization amplitude information, generating heparin sodium detection signal optimization feature information; and performing classification processing on the heparin sodium detection signal optimization feature information to obtain heparin sodium detection information. The method is used for fully optimizing the heparin sodium detection signal through a deep learning method so as to improve the accuracy, robustness and reliability of heparin sodium detection.

Description

Technical Field

[0001] The present application belongs to the field of data processing technology, and in particular relates to a heparin sodium detection method and system based on deep learning. Background Art

[0002] Heparin sodium, a clinically important anticoagulant, is widely used in areas such as thrombosis prevention and cardiovascular surgery. The development of its detection technology directly impacts drug quality control and clinical drug safety. Currently, heparin sodium detection technologies primarily include traditional methods such as high-performance liquid chromatography, spectrophotometry, and electrochemical analysis.

[0003] Traditional methods rely on expensive equipment and complex sample preprocessing, and have problems such as long detection cycle, high cost, and low sensitivity. In addition, the instrument equipment is unable to detect low-concentration sodium heparin signals, resulting in limited detection sensitivity. It is difficult to meet the detection needs of trace samples, which limits the improvement of the detection accuracy of sodium heparin. Summary of the Invention

[0004] In view of this, the embodiments of the present application provide a heparin sodium detection method and system based on deep learning, aiming to solve the problems of high detection cost, low detection accuracy, low effectiveness and insufficient stability for low-concentration heparin sodium in the prior art.

[0005] A first aspect of the embodiments of the present application provides a heparin sodium detection method based on deep learning, comprising:

[0006] Obtaining heparin sodium detection spectral signal information and heparin sodium detection electrochemical signal information;

[0007] Performing feature extraction on the heparin sodium detection spectral signal information and the heparin sodium detection electrochemical signal information to obtain heparin sodium detection spectral signal feature information and heparin sodium detection electrochemical signal feature information;

[0008] Generate multiple heparin sodium detection signal optimization feature information according to the heparin sodium detection spectral signal feature information, the heparin sodium detection electrochemical signal feature information, the preset heparin sodium detection signal feature optimization iteration number, and the preset heparin sodium detection signal feature optimization amplitude information;

[0009] Based on the preset heparin sodium detection signal feature classification convergence threshold information, a plurality of heparin sodium detection signal optimization feature information is classified and processed to obtain heparin sodium detection information.

[0010] A second aspect of the embodiments of the present application provides a heparin sodium detection system based on deep learning, comprising:

[0011] A heparin sodium detection signal information acquisition module is used to obtain heparin sodium detection spectral signal information and heparin sodium detection electrochemical signal information;

[0012] a heparin sodium detection signal characteristic information generation module, configured to extract characteristics of the heparin sodium detection spectral signal information and the heparin sodium detection electrochemical signal information to obtain heparin sodium detection spectral signal characteristic information and heparin sodium detection electrochemical signal characteristic information;

[0013] a heparin sodium detection signal optimization characteristic information generation module, configured to generate a plurality of heparin sodium detection signal optimization characteristic information based on the heparin sodium detection spectral signal characteristic information, the heparin sodium detection electrochemical signal characteristic information, a preset heparin sodium detection signal characteristic optimization iteration number, and a preset heparin sodium detection signal characteristic optimization amplitude information;

[0014] The heparin sodium detection information generating module is used to classify and process the plurality of heparin sodium detection signal optimization feature information based on the preset heparin sodium detection signal feature classification convergence threshold information to obtain heparin sodium detection information.

[0015] A third aspect of an embodiment of the present application provides a terminal device, comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, the steps of the deep learning-based heparin sodium detection method described in the first aspect above are implemented.

[0016] A fourth aspect of an embodiment of the present application provides a computer-readable storage medium, comprising: storing a computer program, which, when executed by a processor, implements the steps of the deep learning-based heparin sodium detection method as described in the first aspect above.

[0017] Compared with the prior art, the embodiments of the present application have the following beneficial effects: by acquiring spectral signal information and electrochemical signal information for heparin sodium detection, the present application achieves multimodal complementarity of heparin sodium detection information, effectively overcoming the defects of single signal susceptibility to interference and incomplete detection information. By extracting features from the spectral signal information and electrochemical signal information for heparin sodium detection, subtle features in the spectral and electrochemical signals are accurately captured, significantly improving feature characterization capabilities. By using a preset number of iterations for heparin sodium detection signal feature optimization and a preset amplitude information for heparin sodium detection signal feature optimization, the extracted spectral signal feature information and electrochemical signal feature information for heparin sodium detection are deeply optimized, thereby improving adaptability to the requirements of complex heparin sodium samples and changing heparin sodium detection environments, eliminating interference from the complex conditions of various heparin sodium samples and the dynamically changing heparin sodium detection environment, and improving the accuracy, robustness, and effectiveness of heparin sodium detection information. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0019] Figure 1 This is a schematic diagram of the implementation process of the heparin sodium detection method based on deep learning provided in Example 1 of the present application;

[0020] Figure 2 This is a schematic diagram of the implementation process of the heparin sodium detection method based on deep learning provided in Example 2 of the present application;

[0021] Figure 3 This is a schematic diagram of the implementation process of the heparin sodium detection method based on deep learning provided in Example 3 of the present application;

[0022] Figure 4 This is a schematic diagram of the implementation process of the heparin sodium detection method based on deep learning provided in Example 4 of the present application;

[0023] Figure 5 This is a schematic diagram of the implementation process of the heparin sodium detection method based on deep learning provided in Example 5 of the present application;

[0024] Figure 6 This is a schematic diagram of the implementation process of the heparin sodium detection method based on deep learning provided in Example 6 of the present application;

[0025] Figure 7 This is a schematic diagram of the implementation process of the heparin sodium detection method based on deep learning provided in Example 7 of the present application;

[0026] Figure 8 Schematic diagram of the structure of a heparin sodium detection system based on deep learning provided in an embodiment of the present application;

[0027] Figure 9 It is a schematic diagram of the terminal device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0028] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present application with unnecessary detail.

[0029] In order to illustrate the technical solution described in this application, specific embodiments are provided below.

[0030] Figure 1 The following is a flowchart of the implementation of the deep learning-based heparin sodium detection method provided in Example 1 of the present application, which is detailed as follows:

[0031] Step S101 , obtaining heparin sodium detection spectral signal information and heparin sodium detection electrochemical signal information.

[0032] In this embodiment, heparin sodium detection spectral signal information may refer to optical signal data related to the molecular structure or physicochemical properties of heparin sodium, obtained through spectral analysis techniques. This information may include ultraviolet-visible absorption spectral signals, infrared spectral signals, and fluorescence spectral signals. The ultraviolet-visible absorption spectral signal refers to the absorption characteristics of heparin sodium molecules in the ultraviolet-visible light band (typically 200-800 nm). Heparin sodium of different concentrations or purities exhibits specific absorption peaks. The infrared spectral signal refers to the vibrational absorption signal of functional groups in heparin sodium molecules under infrared light irradiation, which can be used to reflect characteristic chemical bond information in the molecular structure. The fluorescence spectral signal may refer to the intensity and peak position of the emission spectrum of heparin sodium or its derivatives when excited at a specific wavelength, which can be used as a detection signal. The electrochemical signal information for heparin sodium detection may refer to electrical signal data related to the electrochemical reaction of heparin sodium, obtained through electrochemical analysis methods. This electrochemical signal information for heparin sodium detection may include potential signals, current signals, and impedance signals. The potential signal refers to changes in electrode potential caused by ion exchange and coordination reactions between sodium heparin and the electrode surface, such as the sudden change in endpoint potential in potentiometric titration. The current signal can refer to the electrolytic current generated when sodium heparin undergoes redox reactions on the electrode surface, such as the current versus voltage curve in voltammetry. The impedance signal can refer to changes in resistance and capacitance caused by the interaction between sodium heparin molecules and the electrode interface in electrochemical impedance spectroscopy (EIS), which can be used to reflect molecular adsorption or reaction kinetics. Spectral signal information for sodium heparin detection can be obtained through ultraviolet-visible spectrophotometry, infrared spectroscopy, and fluorescence spectroscopy. Electrochemical signal information for sodium heparin detection can be obtained through potentiometric titration, voltammetry, and electrochemical impedance spectroscopy.

[0033] Step S102 , performing feature extraction on the heparin sodium detection spectral signal information and the heparin sodium detection electrochemical signal information to obtain heparin sodium detection spectral signal feature information and heparin sodium detection electrochemical signal feature information.

[0034] In this embodiment, for the spectral signal information of sodium heparin detection, the spectral signal can be arranged into a one-dimensional sequence in order of wavelength or wave number, and the sequence is scanned for local features through a sliding window. The data in each window is processed by weighted calculation and activation function to extract local features such as characteristic peak position, peak intensity, peak width, etc.; as the level deepens, the local features extracted from the shallow layer are combined and abstracted to gradually form more representative high-level features, thereby obtaining the characteristic signal information of the spectral signal of sodium heparin detection. For the electrochemical signal information of sodium heparin detection, the importance weights of different time points or potential intervals in the signal can be calculated first. By summarizing and statistically analyzing the global information of the signal, analyzing the correlation between each part of the data and the characteristic reaction of sodium heparin, assigning corresponding weights to different sections of the signal, focusing on the key signal segments that can reflect the characteristics of sodium heparin redox reaction, ion exchange, etc., suppressing noise and irrelevant signals, and thus extracting the characteristic signal information of sodium heparin detection electrochemical signal.

[0035] Step S103 , generating a plurality of heparin sodium detection signal optimization feature information according to the heparin sodium detection spectral signal feature information, the heparin sodium detection electrochemical signal feature information, the preset heparin sodium detection signal feature optimization iteration number, and the preset heparin sodium detection signal feature optimization amplitude information.

[0036] In this embodiment, the preset number of heparin sodium detection signal feature optimization iterations and the preset heparin sodium detection signal feature optimization amplitude information can both be manually set. The preset number of heparin sodium detection signal feature optimization iterations can be used as a loop termination condition, and in each iteration, the current heparin sodium detection spectral signal feature information and heparin sodium detection electrochemical signal feature information are adjusted based on the preset heparin sodium detection signal feature optimization amplitude information. Specifically, an initial feature space may be first constructed based on the characteristic information of the sodium heparin detection spectral signal and the characteristic information of the sodium heparin detection electrochemical signal. The step size is adjusted using the preset sodium heparin detection signal feature optimization amplitude information as a parameter. The weight of the feature information is randomly increased or decreased, or the combination method is rearranged. A search is performed in the feature space to attempt to generate a new sodium heparin detection signal feature combination. Then, based on preset evaluation criteria, such as calculating the inter-class distance and information gain of the new feature combination for the sodium heparin signal and the interference signal, the ability of the new feature combination to distinguish sodium heparin from interference is quantitatively evaluated. If the new feature combination is superior to the original combination in key indicators such as discrimination, the feature information is updated according to the new feature combination; otherwise, the original combination is retained. The iterative update process is terminated by a preset number of heparin sodium detection signal feature optimization iterations. After multiple cycles of iteration, multiple sodium heparin detection signal optimized feature information with better performance in distinguishing sodium heparin characteristics are gradually screened out, thereby effectively improving the accuracy and effectiveness of the feature for sodium heparin detection.

[0037] Step S104 : Based on the preset heparin sodium detection signal feature classification convergence threshold information, the plurality of heparin sodium detection signal optimized feature information are classified and processed to obtain heparin sodium detection information.

[0038] In this embodiment, the similarity or distance measurement between the optimized feature information of multiple heparin sodium detection signals may be calculated first, the degree of correlation between the optimized feature information of the heparin sodium detection signals may be determined, and the feature information with high similarity may be classified into one category. During the classification process, the degree of difference in features within the class and the degree of separation between classes may be continuously monitored. When the intra-class difference is less than the preset convergence threshold information of the heparin sodium detection signal feature classification and the inter-class difference reaches the distinguishable standard, the classification result is considered to have stably converged, and the classification adjustment is stopped. Then, based on the classification result, the concentration, purity and other properties of the heparin sodium in the sample are determined, and the heparin sodium detection information is output, completing the entire detection and analysis process.

[0039] The deep learning-based heparin sodium detection method provided in the embodiments of the present application achieves multimodal complementarity of heparin sodium detection information by acquiring heparin sodium detection spectral signal information and heparin sodium detection electrochemical signal information, effectively overcoming the defects of single signal susceptibility to interference and incomplete detection information. By extracting features from the heparin sodium detection spectral signal information and heparin sodium detection electrochemical signal information, subtle features in the spectral and electrochemical signals are accurately captured, significantly improving feature characterization capabilities. The extracted heparin sodium detection spectral signal feature information and heparin sodium detection electrochemical signal feature information are deeply optimized through a preset number of heparin sodium detection signal feature optimization iterations and preset heparin sodium detection signal feature optimization amplitude information. This improves the adaptability to the requirements of complex heparin sodium samples and changing heparin sodium detection environments, eliminates interference from the complex conditions of various heparin sodium samples and the dynamically changing heparin sodium detection environment, and improves the accuracy, robustness, and effectiveness of heparin sodium detection information.

[0040] Figure 2 The flowchart of the implementation of the heparin sodium detection method based on deep learning provided in the second embodiment of the present application is shown. The difference between the second embodiment and the first embodiment is that the step S102 specifically includes:

[0041] Step S201 , performing time alignment processing and filtering processing on the heparin sodium detection spectral signal information and the heparin sodium detection electrochemical signal information to obtain heparin sodium detection spectral signal denoising information and heparin sodium detection electrochemical signal denoising information.

[0042] In this embodiment, the time alignment process can be implemented using a sliding window cross-correlation algorithm. This can be done by sorting the electrochemical signal and the spectral signal by timestamp, sliding the window with a minimum sampling interval as the step size, calculating the Pearson correlation coefficient of the signal within the window, taking the window offset corresponding to the maximum correlation coefficient as the time calibration parameter, and resampling or interpolating the electrochemical signal to achieve time synchronization of the heparin sodium detection spectral signal information and the heparin sodium detection electrochemical signal information. The filtering process can be performed using a Savitzky-Golay filter combined with a wavelet transform to filter the time-aligned heparin sodium detection spectral signal information and the heparin sodium detection electrochemical signal information to obtain filtered heparin sodium detection spectral signal denoising information and heparin sodium detection electrochemical signal denoising information.

[0043] Step S202 , performing convolution calculation on the denoised heparin sodium detection spectral signal information and the denoised heparin sodium detection electrochemical signal information according to a preset heparin sodium detection signal feature extraction matrix to obtain a plurality of sub-feature information of the heparin sodium detection spectral signal and a plurality of sub-feature information of the heparin sodium detection electrochemical signal.

[0044] In this embodiment, the preset heparin sodium detection signal feature extraction matrix can be manually set. The denoised heparin sodium detection spectral signal information can be arranged into a two-dimensional matrix by wavelength order. The heparin sodium detection signal feature extraction matrix is ​​slid across the spectral matrix to extract local features such as absorption intensity changes and peak profiles within specific wavelength ranges through matrix dot multiplication and accumulation operations, thereby generating sub-feature information for the heparin sodium detection spectral signal. Alternatively, the potential-current information from the denoised heparin sodium detection electrochemical signal information can be constructed into a matrix format. Through matrix convolution operations on the potential-current information constructed by the heparin sodium detection signal feature extraction matrix, electrochemical sub-features such as the current response slope at the moment of potential mutation and the peak spacing of the cyclic voltammetry curve can be captured, thereby obtaining multiple sub-feature information for the heparin sodium detection electrochemical signal.

[0045] Step S203: performing interactive enhancement processing on the plurality of heparin sodium detection spectral signal sub-feature information and the plurality of heparin sodium detection electrochemical signal sub-feature information according to a plurality of preset heparin sodium detection signal sub-feature enhancement matrices to obtain heparin sodium detection spectral signal feature information and heparin sodium detection electrochemical signal feature information.

[0046] In this embodiment, the plurality of preset heparin sodium detection signal sub-feature enhancement matrices can be manually set. Each heparin sodium detection signal sub-feature enhancement matrix corresponds to the detection requirements of a specific physical and chemical property of the heparin sodium molecule, such as highlighting the spectral absorption characteristics associated with sulfate groups or enhancing the electrochemical current signal corresponding to redox reactions. This can be achieved by performing a dot product operation on the heparin sodium detection signal sub-feature enhancement matrix and the heparin sodium detection spectral signal sub-feature information and the heparin sodium detection electrochemical signal sub-feature information, and performing a nonlinear transformation on the result in combination with an activation function. Based on the intrinsic correlation between the heparin sodium molecule in the spectral and electrochemical fields, a higher weight is assigned to signal fragments in the two types of sub-feature information that reflect the same molecular structure or reaction process. For example, when a specific wavelength absorption peak in the spectral signal and a current peak in a certain potential range in the electrochemical signal both indicate a functional group of heparin sodium, the heparin sodium detection signal sub-feature enhancement matrix is ​​used to increase the weight of this part of the sub-features, suppress background interference signals, and ensure that the heparin sodium detection spectral signal feature information and the heparin sodium detection electrochemical signal feature information, during the fusion process, retain the unique advantages of each modality while strengthening the complementary characteristics, thereby more accurately characterizing the essential characteristics of the heparin sodium molecule.

[0047] The deep learning-based heparin sodium detection method provided in the embodiments of the present application completely eliminates the deviation and data noise interference of multimodal signals in the time dimension through time alignment processing and filtering processing. Through a preset heparin sodium detection signal feature extraction matrix and a preset heparin sodium detection signal sub-feature enhancement matrix, progressive processing of heparin sodium detection signal features from basic extraction to deep enhancement is achieved, greatly improving the accuracy of feature extraction and significantly enhancing the feature's ability to characterize heparin sodium characteristics in complex samples. At the same time, the complementary value of spectral detection signals and electrochemical detection signals is deeply explored, effectively addressing the challenges of complex sample composition and changing detection environment in heparin sodium detection scenarios, and greatly improving the accuracy, reliability and anti-interference performance of detection results.

[0048] Figure 3 The flowchart of the implementation of the heparin sodium detection method based on deep learning provided in Example 3 of the present application is shown. The difference between it and the above-mentioned Example 2 is that:

[0049] The plurality of preset heparin sodium detection signal sub-feature enhancement matrices include a preset first heparin sodium detection signal sub-feature enhancement matrix, a preset second heparin sodium detection signal sub-feature enhancement matrix, and a preset third heparin sodium detection signal sub-feature enhancement matrix;

[0050] The step S203 specifically includes:

[0051] Step S301 , performing feature interactive fusion processing on the plurality of heparin sodium detection spectral signal sub-feature information and the plurality of heparin sodium detection electrochemical signal sub-feature information to obtain heparin sodium detection signal sub-feature fusion information.

[0052] In this embodiment, feature interaction fusion processing can be performed using a method that combines tensor concatenation with fully connected layer mapping. Multiple heparin sodium detection spectral signal sub-feature information and multiple heparin sodium detection electrochemical signal sub-feature information can be expanded according to the channel dimension. A high-dimensional feature tensor containing both spectral and electrochemical features is formed through tensor concatenation. This tensor is then linearly transformed and nonlinearly activated using a fully connected layer to achieve information cross-transfer between the spectral and electrochemical sub-features, capturing the correlation between features of different modalities. This results in fused heparin sodium detection signal sub-feature information that combines the strengths of both signal types.

[0053] Step S302 : obtaining first heparin sodium detection signal sub-feature enhancement information according to the heparin sodium detection signal sub-feature fusion information and a preset first heparin sodium detection signal sub-feature enhancement matrix.

[0054] In this embodiment, the preset first heparin sodium detection signal sub-feature enhancement matrix, the preset second heparin sodium detection signal sub-feature enhancement matrix, and the preset third heparin sodium detection signal sub-feature enhancement matrix can all be manually set. The preset first heparin sodium detection signal sub-feature enhancement matrix can be used to focus on the detection of physical and chemical properties related to sulfate groups in heparin sodium molecules. A matrix multiplication operation can be performed on the heparin sodium detection signal sub-feature fusion information and the first enhancement matrix. Signal components such as spectral absorption intensity changes and electrochemical current responses related to sulfate group characteristics can be amplified using weight parameters of the first heparin sodium detection signal sub-feature enhancement matrix, while irrelevant background features are suppressed. Nonlinear transformation is then performed through an activation function to highlight signal features closely related to the sulfate group structure, ultimately obtaining enhanced first heparin sodium detection signal sub-feature enhancement information that enhances the specificity of sulfate group detection.

[0055] Step S303 : obtaining second heparin sodium detection signal sub-feature enhancement information according to the heparin sodium detection signal sub-feature fusion information and a preset second heparin sodium detection signal sub-feature enhancement matrix.

[0056] In this embodiment, the redox reaction characteristics of the heparin sodium molecule can be targeted using a preset second heparin sodium detection signal sub-feature enhancement matrix. This can be accomplished by performing a dot product operation on the second heparin sodium detection signal sub-feature enhancement matrix and the heparin sodium detection signal sub-feature fusion information. The second heparin sodium detection signal sub-feature enhancement matrix assigns higher weights to current variation characteristics within corresponding potential intervals and spectral absorption variation characteristics at specific wavelengths based on the characteristic manifestations of the redox reaction in the spectrum and electrochemical signals, thereby weakening other interfering features. After processing with an activation function, the second heparin sodium detection signal sub-feature enhancement information highlighting the redox reaction characteristics is generated.

[0057] Step S304 : obtaining third heparin sodium detection signal sub-feature enhancement information according to the heparin sodium detection signal sub-feature fusion information and a preset third heparin sodium detection signal sub-feature enhancement matrix.

[0058] In this embodiment, a preset third heparin sodium detection signal sub-feature enhancement matrix can be used to focus on characterizing the overall structural integrity of the heparin sodium molecule. This can be accomplished by performing a weighted calculation on the heparin sodium detection signal sub-feature fusion information and the third heparin sodium detection signal sub-feature enhancement matrix. The third heparin sodium detection signal sub-feature enhancement matrix, using pre-trained parameters, comprehensively considers the overall distribution of functional group vibrational absorption in the spectrum and the kinetic characteristics of the interaction between the molecule and the electrode in electrochemistry. It enhances signal features reflecting the intact molecular structure and suppresses abnormal features caused by sample impurities or detection noise, thereby obtaining the third heparin sodium detection signal sub-feature enhancement information reflecting the structural characteristics of the heparin sodium molecule.

[0059] Step S305 : Obtaining heparin sodium detection signal sub-feature interaction enhancement information based on the first heparin sodium detection signal sub-feature enhancement information and the second heparin sodium detection signal sub-feature enhancement information.

[0060] In this embodiment, a bidirectional interactive operation may be performed on the first heparin sodium detection signal sub-feature enhancement information and the second heparin sodium detection signal sub-feature enhancement information. The first heparin sodium detection signal sub-feature enhancement information and the second heparin sodium detection signal sub-feature enhancement information may be element-wise added together to fuse the sulfate group feature and the redox reaction feature to obtain heparin sodium detection signal sub-feature interactive enhancement information that integrates these two key characteristics.

[0061] Step S306 , obtaining heparin sodium detection signal feature information according to the heparin sodium detection signal sub-feature interaction enhancement information and the third heparin sodium detection signal sub-feature enhancement information.

[0062] In this embodiment, the heparin sodium detection signal sub-feature interaction reinforcement information and the third heparin sodium detection signal sub-feature reinforcement information can be deeply fused. The heparin sodium detection signal sub-feature interaction reinforcement information and the third heparin sodium detection signal sub-feature reinforcement information can be first nonlinearly mapped using a multilayer perceptron. The multilayer network structure in the multilayer perceptron can be used to mine high-order correlations between features. A residual connection network can be combined to avoid the vanishing gradient problem caused by increased network depth. The heparin sodium detection signal sub-feature interaction reinforcement information and the third heparin sodium detection signal sub-feature reinforcement information, which have been nonlinearly mapped using the residual connection network, can then be multiplied together, and the multiplication result can be used as the heparin sodium detection signal feature information.

[0063] Step S307 , separating and processing the characteristic information of the heparin sodium detection signal to obtain characteristic information of the heparin sodium detection spectral signal and characteristic information of the heparin sodium detection electrochemical signal.

[0064] In this embodiment, two branch convolutional neural networks can be used to perform convolution operations on the characteristic information of the sodium heparin detection signal. One branch network performs feature screening on the wavelength distribution characteristics of the spectral signal characteristics, and the other branch network focuses on the potential-current change characteristics of the electrochemical signal characteristics for feature extraction. Each branch network gradually filters out the features of non-corresponding modes through the convolution layer and the pooling layer, and outputs the characteristic information of the sodium heparin detection spectral signal and the characteristic information of the sodium heparin detection electrochemical signal, respectively.

[0065] The deep learning-based heparin sodium detection method provided in the embodiments of the present application progressively enhances the sub-feature information of the spectral signal and the sub-feature information of the electrochemical signal of heparin sodium detection through multiple preset sub-feature enhancement matrices of the heparin sodium detection signal. It adopts a feature interactive fusion and separation mechanism to achieve deep synergy between spectral and electrochemical signals while retaining the independence of each modal feature, effectively balancing the complementarity and specificity of multimodal information, thereby enhancing the anti-interference ability of complex heparin sodium sample detection, significantly improving the specificity and sensitivity of heparin sodium detection, and providing more reliable technical support for heparin sodium quality control.

[0066] Figure 4 The flowchart of the implementation of the heparin sodium detection method based on deep learning provided in the fourth embodiment of the present application is shown. The difference between the fourth embodiment and the first embodiment is that the step S103 specifically includes:

[0067] Step S401 : performing fusion coding processing on the characteristic information of the heparin sodium detection spectral signal and the characteristic information of the heparin sodium detection electrochemical signal to generate a plurality of individual characteristic information of the heparin sodium detection signal.

[0068] In this embodiment, the characteristic information of the spectral signal of heparin sodium detection and the characteristic information of the electrochemical signal of heparin sodium detection can be first expanded into a feature vector sequence by dimension. By designing a multi-level feature interaction module, the features of different modalities can be associated and calculated at multiple abstract levels. For example, in the first level of interaction, the correlation weights of the spectral absorption peak position features and the electrochemical redox potential features are calculated to generate preliminary cross-features; in subsequent levels, the preliminary cross-features are subjected to nonlinear transformation and quadratic combination to gradually extract more abstract cross-modal feature associations. Each individual feature information contains a multi-level feature expression from the bottom signal to the high-level semantics. In this way, the complementary information between the two signal modalities is fully mined.

[0069] Step S402 : Based on the preset heparin sodium detection signal feature optimization boundary, a plurality of initial heparin sodium detection signal feature individuals are generated according to the plurality of heparin sodium detection signal feature individual information.

[0070] In this embodiment, the preset optimization boundaries for sodium heparin detection signal features can be manually set to define the physically feasible domain of the feature parameters. For example, the spectral absorption intensity feature is constrained to the interval [0, 1], and the electrochemical current response feature is constrained to the linear range of the detection device. A boundary-constrained mapping algorithm is used to map the individual features of multiple sodium heparin detection signal features to the optimization boundaries. Eigenvalues ​​exceeding the upper limit are proportionally compressed to the boundary value; eigenvalues ​​below the lower limit are adjusted to the feasible domain through interpolation. A feature smoothing mechanism is also introduced to apply gradient corrections to eigenvalues ​​near the boundaries to avoid optimization oscillations caused by sudden changes in the boundaries, thereby generating initial individual features that conform to physical laws.

[0071] Step S403 , calculating and obtaining a plurality of heparin sodium detection signal characteristic fitnesses based on the plurality of initial heparin sodium detection signal characteristic individuals.

[0072] In this embodiment, a comprehensive fitness evaluation function for the heparin sodium detection signal feature can be designed to calculate the fitness of the heparin sodium detection signal feature. This comprehensive fitness evaluation function can include three core indicators: inter-class separation, which measures the ability of a feature individual to distinguish between heparin sodium samples and interference samples, evaluated by calculating the distance between the two classes of samples in the feature space; information gain, which evaluates the feature individual's ability to predict target attributes such as heparin sodium concentration and purity; and robustness, which calculates the change in feature classification accuracy before and after the disturbance by adding noise simulating the detection environment disturbance to the feature individual. These three indicators can be weighted and fused to obtain a final fitness value. A higher fitness value indicates that the feature individual has a stronger ability to identify heparin sodium in complex environments. This is like giving each feature individual a multi-dimensional ability test, with those with higher scores having a greater advantage in subsequent optimization.

[0073] Step S404 : using the initial heparin sodium detection signal characteristic individual corresponding to the maximum value of the fitness of the plurality of heparin sodium detection signal characteristics as a reference heparin sodium detection signal characteristic individual.

[0074] In this embodiment, the fitness of all heparin sodium detection signal features may be compared, and the initial heparin sodium detection signal feature individual corresponding to the maximum value may be selected as the reference heparin sodium detection signal feature individual. In the subsequent iterative optimization process, the reference heparin sodium detection signal feature individual will serve as a guiding direction to ensure that the optimization direction is iterated in the direction of improving the accuracy of heparin sodium detection.

[0075] Step S405 , generating a plurality of target heparin sodium detection signal feature individuals based on the plurality of initial heparin sodium detection signal feature individuals, the reference heparin sodium detection signal feature individuals, the preset heparin sodium detection signal feature optimization iteration number, and the preset heparin sodium detection signal feature optimization amplitude information.

[0076] In this embodiment, an iterative optimization method is used to generate target heparin sodium detection signal signature individuals. A preset number of iterations for optimizing the heparin sodium detection signal signature can be used as a termination condition. Each iteration includes the following operations: first, the parameters of the initial signature individuals are randomly adjusted, with the adjustment amplitude controlled by preset optimization amplitude information; then, the adjusted individuals are combined with reference signature individuals to generate new test individuals; finally, the fitness of the test individuals is compared with that of the original individuals, and individuals with higher fitness are retained for the next generation. During the iterative process, the parameter adjustment strategy is dynamically adjusted based on the optimization progress. In the early stages, different signature combinations are extensively explored, and in the later stages, high-quality signature combinations are refined and optimized. Through this continuous iterative evolutionary mechanism, multiple target signature individuals with enhanced ability to distinguish heparin sodium characteristics are gradually generated. Through continuous variation, combination, and screening, the signature individuals gradually become more adapted to the needs of heparin sodium detection.

[0077] Step S406: generating a plurality of heparin sodium detection signal optimization feature information based on the plurality of target heparin sodium detection signal feature individuals.

[0078] In this embodiment, the target sodium heparin detection signal feature individuals can be classified and analyzed and feature screening can be performed. The similarity between all target individuals can be calculated first, and similar feature individuals can be clustered into one category. The center of each category represents a group of feature combinations with similar recognition patterns. The importance of each category of features is then evaluated. The contribution of each feature dimension to the sodium heparin detection result is calculated through an algorithm, and feature dimensions with high contribution are screened out. The screened features are then standardized. Combined with professional knowledge of sodium heparin detection, the abstract feature vectors are converted into optimized feature information with clear physical meaning, such as "sulfate content feature" and "molecular chain length feature", so that the feature information is easier to understand and apply in actual sodium heparin detection analysis.

[0079] The deep learning-based heparin sodium detection method provided in the embodiments of the present application achieves complementary advantages of spectral and electrochemical signals through deep fusion of multi-dimensional features, effectively extracts features that better reflect the essential characteristics of heparin sodium, ensures the physical rationality and detection reliability of the heparin sodium detection signal features through preset heparin sodium detection signal feature optimization boundaries, and adopts an iterative optimization strategy combined with the guidance of individual reference heparin sodium detection signal features to improve the efficiency and accuracy of heparin sodium detection signal feature optimization, thereby enhancing the precision and robustness of heparin sodium detection.

[0080] Figure 5 The flowchart of the implementation of the heparin sodium detection method based on deep learning provided in the fifth embodiment of the present application is shown. The difference between the fifth embodiment and the fourth embodiment is that the step S403 specifically includes:

[0081] Step S501 : performing dimensionality reduction processing on a plurality of individual features of the initial heparin sodium detection signal to obtain a plurality of dimensionality reduction information of the initial heparin sodium detection signal features.

[0082] In this embodiment, dimensionality reduction can be performed using a multi-layer feature mapping approach. Initial heparin sodium detection signal feature individuals in a high-dimensional feature space can be projected into a low-dimensional space by constructing a nonlinear transformation function. Each initial feature individual can be treated as a high-dimensional vector, and a nonlinear mapping function can be used to capture the complex relationships between features. For example, initial heparin sodium detection signal feature individuals containing multidimensional features such as spectral absorption peak position, peak intensity, and electrochemical current response can be mapped into a low-dimensional space through a nonlinear transformation, while retaining key information such as the correlation between absorption peaks at specific wavelengths and corresponding potential and current peaks. This avoids the loss of feature relationships caused by traditional linear dimensionality reduction, thereby obtaining reduced dimensionality information of the initial heparin sodium detection signal features that retains core discriminatory power.

[0083] Step S502 : extracting the dimensionality reduction information of the plurality of initial heparin sodium detection signal characteristics to obtain a plurality of dimensionality reduction information of heparin sodium detection spectrum signal characteristics and dimensionality reduction information of heparin sodium detection electrochemical signal characteristics.

[0084] In this embodiment, information extraction can be achieved by using feature modal decoupling. Two independent feature screening modules are designed for the initial heparin sodium detection signal feature dimensionality reduction information after dimensionality reduction: one module focuses on identifying wavelength-related feature components, such as the low-dimensional coordinates corresponding to the ultraviolet absorption peak, and the other module focuses on the potential-current related feature components, such as the low-dimensional coordinates corresponding to the peak spacing of the voltammetric curve. By calculating the correlation coefficient between each low-dimensional feature component and the original spectral / electrochemical signal, components with correlations above a threshold are screened out to form the heparin sodium detection spectral signal feature dimensionality reduction information and the heparin sodium detection electrochemical signal feature dimensionality reduction information, respectively. For example, if a low-dimensional component is highly correlated with the intensity change of the 230nm ultraviolet absorption peak, it is classified as spectral dimensionality reduction information; if it is correlated with the current response at a potential of 0.5V, it is classified as electrochemical dimensionality reduction information, thereby achieving decoupling and targeted retention of multimodal features.

[0085] Step S503 , calculating and obtaining a plurality of heparin sodium detection signal attenuation information based on the plurality of heparin sodium detection spectral signal feature dimensionality reduction information, the heparin sodium detection electrochemical signal feature dimensionality reduction information, and a preset heparin sodium detection signal attenuation calculation function.

[0086] In this embodiment, the preset heparin sodium detection signal attenuation calculation function can be manually set and can be used to quantify the degree of information loss of the feature after dimensionality reduction. The spectral and electrochemical dimensionality reduction information can be used as input to calculate the attenuation value through the following steps: (1) For the spectral dimensionality reduction information, the root mean square error (RMSE) between the original spectral feature and the feature after dimensionality reduction is calculated to reflect the information loss of the spectral feature during dimensionality reduction; similarly, the RMSE is calculated for the electrochemical dimensionality reduction information; (2) by calculating the change in mutual information between the spectral-electrochemical feature pair before and after dimensionality reduction, such as the mutual information between a wavelength absorption peak and the corresponding potential current peak, the degree of retention of cross-modal correlation information is evaluated; (3) the intra-modal attenuation and the inter-modal correlation attenuation are weighted and summed according to a preset weight (such as 0.6:0.4) to obtain the heparin sodium detection signal attenuation information. It can be understood that the lower the heparin sodium detection signal attenuation information, the more effective information and cross-modal correlation in the original signal are retained by the feature after dimensionality reduction.

[0087] Step S504 , calculating a plurality of heparin sodium detection signal feature fitnesses based on the plurality of heparin sodium detection spectral signal feature dimensionality reduction information, the heparin sodium detection electrochemical signal feature dimensionality reduction information, and the preset heparin sodium detection signal feature fitness weight information.

[0088] In this embodiment, the preset heparin sodium detection signal characteristic fitness weight information can be manually set and can include a spectral feature weight matrix and an electrochemical feature weight matrix, which respectively correspond to the importance coefficients of each feature component in the spectral / electrochemical dimensionality reduction information. The spectral dimensionality reduction information can be first multiplied by the corresponding weight coefficient for each feature component and then accumulated to obtain the spectral modal fitness; similarly, the electrochemical modal fitness is calculated, and the weight coefficient is pre-set based on the molecular characteristics of heparin sodium, for example, spectral absorption features related to sulfate groups have a higher weight; then, the covariance of the spectral and electrochemical dimensionality reduction information is calculated to reflect the collaborative discrimination ability of the two types of features in the dimensionality reduction space. A higher covariance indicates stronger cross-modal complementarity; then, the intra-modal fitness (spectral × 0.5 + electrochemical × 0.5) and the inter-modal collaborative fitness (× 0.5) are weighted and summed to obtain the heparin sodium detection signal characteristic fitness. It can be understood that the higher the fitness of the heparin sodium detection signal feature, the better the individual initial heparin sodium detection signal feature performs in retaining single modality validity and cross-modality synergy, and is more suitable for heparin sodium detection.

[0089] The deep learning-based heparin sodium detection method provided in the embodiments of the present application reduces feature redundancy during the dimensionality reduction process, retains key information about the molecular structure and electrochemical reaction of heparin sodium, significantly improves the optimization efficiency of heparin sodium detection signal features and the generalization ability of the detection model, is suitable for accurately distinguishing heparin sodium from interfering substances in complex samples, effectively improves the accuracy of heparin sodium detection, and enhances the efficiency and reliability of heparin sodium quality control.

[0090] Figure 6 The flowchart of the implementation of the heparin sodium detection method based on deep learning provided in Example 6 of the present application is shown. The difference between the method and the above-mentioned Example 4 is that the step S405 specifically includes:

[0091] Step S601: performing iterative optimization processing on the plurality of initial heparin sodium detection signal feature individuals according to the reference heparin sodium detection signal feature individual and the preset heparin sodium detection signal feature optimization amplitude information, to obtain a plurality of intermediate heparin sodium detection signal feature individuals and the number of iterative optimization times of the heparin sodium detection signal feature individual.

[0092] In this embodiment, the preset heparin sodium detection signal feature optimization amplitude information can be manually set. Using the reference heparin sodium detection signal feature individual as the optimization guide, the following operations can be performed on each initial heparin sodium detection signal feature individual: According to the preset heparin sodium detection signal feature optimization amplitude information, the parameters of the initial heparin sodium detection signal feature individual are randomly perturbed, such as by a ±10% weight adjustment. The perturbation amplitude dynamically decays with the iterative process, ensuring global exploration in the early stages and local fine-tuning in the later stages. The perturbed initial heparin sodium detection signal feature individual is then weightedly fused with the reference heparin sodium detection signal feature individual dimension by dimension. For example, the spectral feature is the weight of the initial individual plus the weight of the reference heparin sodium detection signal feature individual, and the electrochemical feature is the weight of the initial individual plus the weight of the reference heparin sodium detection signal feature individual, to obtain multiple intermediate heparin sodium detection signal feature individuals. After each round of optimization to generate an intermediate heparin sodium detection signal feature individual, the number of iterative optimizations of the heparin sodium detection signal feature individual is automatically incremented by 1 to record the current optimization progress.

[0093] Step S602 , determining whether the number of iterative optimizations of the individual heparin sodium detection signal characteristics is equal to the preset number of iterative optimizations of the heparin sodium detection signal characteristics; if so, proceeding to step S603 ; if not, proceeding to step S604 .

[0094] In this embodiment, a preset number of heparin sodium detection signal feature optimization iterations can be used as a fixed threshold. When the number of iterative optimization iterations reaches this threshold, the feature space search is considered to have covered the primary optimization direction, and iterations are stopped. If not, subsequent optimization steps are continued to ensure sufficient feature optimization. The preset number of heparin sodium detection signal feature optimization iterations can be set to 100.

[0095] Step S603: using the plurality of intermediate heparin sodium detection signal characteristic individuals as a plurality of target heparin sodium detection signal characteristic individuals.

[0096] In this embodiment, the intermediate heparin sodium detection signal characteristic individuals can be used as multiple target heparin sodium detection signal characteristic individuals, which not only retains the basic distinguishing ability of the initial heparin sodium detection signal characteristic individuals, but also integrates the optimal feature combination pattern of the reference heparin sodium detection signal characteristic individuals, thereby more accurately characterizing the spectral and electrochemical properties of the heparin sodium molecule.

[0097] Step S604 , calculating and obtaining a plurality of iterative optimization fitnesses of heparin sodium detection signal characteristics based on the plurality of individual intermediate heparin sodium detection signal characteristics.

[0098] In this embodiment, the inter-class separation (such as Mahalanobis distance), information gain rate (such as Gini index), and robustness index (difference in classification accuracy before and after noise perturbation) of the intermediate feature individual can be calculated; then, the fitness improvement rate of the current intermediate feature individual relative to the individual in the previous round of iteration is calculated, for example, improvement rate = (current fitness - previous round fitness) / previous round fitness × 100%; then, the basic evaluation result and the evolutionary efficiency are weightedly fused to obtain the iterative optimization fitness of the heparin sodium detection signal feature. A high fitness value indicates that the feature individual has achieved effective evolution during the iterative process.

[0099] Step S605: Use the plurality of intermediate heparin sodium detection signal characteristic individuals as initial heparin sodium detection signal characteristic individuals, use the intermediate heparin sodium detection signal characteristic individual corresponding to the maximum value of the iterative optimization fitness of the plurality of heparin sodium detection signal characteristics as the reference heparin sodium detection signal characteristic individual, and return to step S601.

[0100] In this embodiment, the currently generated intermediate feature individual is used as the initial input for the next round of iteration to ensure the continuity and cumulativeness of the optimization process; the intermediate feature individual corresponding to the maximum value of the iterative optimization fitness is selected as the new reference standard to guide subsequent iterations to search for higher fitness areas; thereby, each round of iteration is carried out based on the current optimal feature pattern, accelerating the convergence to the global optimal feature combination.

[0101] The deep learning-based heparin sodium detection method provided in the embodiments of the present application balances the optimization accuracy of different modal features, strengthens feature complementarity through cross-modal weighted fusion, dynamically adjusts the optimization direction, and avoids falling into local optimality. This can accurately capture the characteristics of heparin sodium molecules, effectively improve the sensitivity and anti-interference ability of heparin sodium detection, and enhance the efficiency and effectiveness of heparin sodium quality control.

[0102] Figure 7 The flowchart of the implementation of the heparin sodium detection method based on deep learning provided in Example 7 of the present application is shown. The difference between the method and the above-mentioned Example 1 is that step S104 specifically includes:

[0103] Step S701 : randomly extracting a plurality of heparin sodium detection signal optimization feature information according to preset heparin sodium detection signal optimization feature extraction quantity information to obtain a plurality of central heparin sodium detection signal optimization feature information.

[0104] In this embodiment, the preset number of heparin sodium detection signal optimization feature extraction information can be manually set, and can be set to 5 or 10, to determine the number of initial classification centers. A corresponding number of features can be extracted from all heparin sodium detection signal optimization feature information using a random sampling algorithm as a center sample. For example, 5 of 100 heparin sodium detection signal optimization feature information can be randomly selected as the initial centers. The sampling process adopts a sampling strategy without replacement to ensure that each center feature has a certain degree of random distribution in the feature space, thereby avoiding classification bias caused by excessive concentration of initial centers.

[0105] Step S702 , calculating the logical distances between the plurality of heparin sodium detection signal optimization feature information and the plurality of central heparin sodium detection signal optimization feature information to obtain a plurality of heparin sodium detection signal optimization feature distance information.

[0106] In this embodiment, the logical distance may be calculated as the Euclidean distance, and the Euclidean distance between each heparin sodium detection signal optimization feature information and the central heparin sodium detection signal optimization feature information is calculated as the heparin sodium detection signal optimization feature distance information.

[0107] Step S703: obtaining a plurality of heparin sodium detection signal optimization feature class information based on the plurality of heparin sodium detection signal optimization feature distance information and the central heparin sodium detection signal optimization feature information; the heparin sodium detection signal optimization feature class information corresponds one-to-one to the central heparin sodium detection signal optimization feature information; and one heparin sodium detection signal optimization feature class information includes a plurality of heparin sodium detection signal optimization feature information.

[0108] In this embodiment, clustering can be performed using the nearest neighbor assignment principle: each optimized feature is assigned to the cluster to which the central feature with the smallest distance belongs. For example, if feature A has the smallest distance from center 3, feature A is assigned to the cluster corresponding to center 3. Through this assignment process, all optimized feature information is divided into clusters equal to the number of centers. Each cluster contains a number of feature information, forming optimized feature cluster information for heparin sodium detection signals. This achieves a preliminary partitioning of the feature space, grouping optimized feature distance information for heparin sodium detection signals with similar spectral-electrochemical feature combinations into one cluster.

[0109] Step S704 , calculating the average value of each of the heparin sodium detection signal optimization feature information to obtain a plurality of heparin sodium detection signal optimization feature average value information.

[0110] In this embodiment, mean calculation is performed on each feature class: for all optimized feature information within the class, the average value is calculated according to the feature dimension. For example, a class contains 10 feature information, and each feature information contains 5 dimensions such as ultraviolet absorption peak position and infrared vibration intensity. Then, the mean value of the class in each dimension is calculated to generate a new feature vector as the average feature information of the class. The average value information of the optimized feature of the sodium heparin detection signal can be used to represent the overall feature distribution center of the corresponding class, which can eliminate random fluctuations in the features within the class and highlight common features.

[0111] Step S705 , calculating the average value information of the plurality of heparin sodium detection signal optimization feature information and the convergence distance value of the central heparin sodium detection signal optimization feature information to obtain the heparin sodium detection signal optimization feature classification convergence value information.

[0112] In this embodiment, the convergence distance value calculates the Euclidean distance between the average feature information of each class and the original center feature information, and sums the convergence distances of all classes to obtain the convergence value information of the optimized feature classification of the heparin sodium detection signal, which is used to reflect the update amplitude of the classification center. The smaller the convergence value information of the optimized feature classification of the heparin sodium detection signal, the more stable the classification result.

[0113] Step S706 , determining whether the heparin sodium detection signal optimization feature classification convergence value information is less than the preset heparin sodium detection signal feature classification convergence threshold information; if so, proceeding to step S707 ; if not, proceeding to step S708 .

[0114] In this embodiment, the preset heparin sodium detection signal feature classification convergence threshold information can be manually set. When the heparin sodium detection signal optimization feature classification convergence value information is less than the preset heparin sodium detection signal feature classification convergence threshold information, it indicates that after the current iteration, the change range of the classification center is sufficiently small and the classification result is stable; if the heparin sodium detection signal optimization feature classification convergence value information is not less than the preset heparin sodium detection signal feature classification convergence threshold information, it is necessary to continue iteratively optimizing the classification center to ensure the accuracy of the classification.

[0115] Step S707: Optimize feature information according to the plurality of heparin sodium detection signals to obtain heparin sodium detection information.

[0116] In this embodiment, after classification converges, each feature class corresponds to a specific attribute of sodium heparin: the average spectral absorption intensity of the feature information within the class and the electrochemical current value correspond to the standard concentration curve to determine the sample concentration range; the degree of dispersion of the features within the class reflects the sample purity, and the lower the dispersion, the higher the purity; if the distance between a certain class feature and the standard sodium heparin feature exceeds a preset threshold, it is judged as an abnormal sample; then, the attribute information corresponding to each class is integrated to output sodium heparin detection information, such as "the sample concentration is 2.5 IU / mL, the purity is 98.3%, and there is no abnormal interference."

[0117] Step S708 , taking the average value of the optimized characteristic information of the plurality of heparin sodium detection signals as the optimized characteristic information of the central heparin sodium detection signal, and returning to step S702 .

[0118] In this embodiment, the currently calculated class average feature information is used as the new center feature information to replace the original center feature, and then the distances between all optimized features and the new center are recalculated and the categories are redistributed. By continuously updating the center position, the classification results gradually approach the true distribution of the feature space, avoiding classification deviations caused by the initial random center until the convergence distance meets the threshold requirement.

[0119] The deep learning-based heparin sodium detection method provided in the embodiments of the present application utilizes a strategy of iteratively updating classification centers to dynamically adapt to the distribution characteristics of the feature space. In complex heparin sodium sample detection, it can accurately infer key indicators such as heparin sodium concentration and purity based on the average attributes of the feature class. At the same time, it ensures the stability of the classification results through convergence judgment. It exhibits greater robustness when there are large differences between heparin sodium sample batches or when the detection environment changes, thereby improving the accuracy and stability of heparin sodium detection.

[0120] Corresponding to the method of the above embodiment, Figure 8 The structural block diagram of the deep learning-based heparin sodium detection system provided in an embodiment of the present application is shown. For the sake of convenience, only the parts related to the embodiment of the present application are shown. Figure 8 The exemplary deep learning-based heparin sodium detection system can be the execution subject of the deep learning-based heparin sodium detection method provided in the aforementioned embodiment 1.

[0121] Reference Figure 8 , the deep learning-based heparin sodium detection system includes:

[0122] Heparin sodium detection signal information acquisition module 810, used to acquire heparin sodium detection spectral signal information and heparin sodium detection electrochemical signal information;

[0123] A heparin sodium detection signal characteristic information generating module 820 is configured to extract characteristics of the heparin sodium detection spectral signal information and the heparin sodium detection electrochemical signal information to obtain heparin sodium detection spectral signal characteristic information and heparin sodium detection electrochemical signal characteristic information;

[0124] a heparin sodium detection signal optimization characteristic information generating module 830, configured to generate a plurality of heparin sodium detection signal optimization characteristic information based on the heparin sodium detection spectral signal characteristic information, the heparin sodium detection electrochemical signal characteristic information, a preset heparin sodium detection signal characteristic optimization iteration number, and a preset heparin sodium detection signal characteristic optimization amplitude information;

[0125] The heparin sodium detection information generating module 840 is configured to classify and process the plurality of heparin sodium detection signal optimized feature information based on a preset heparin sodium detection signal feature classification convergence threshold information to obtain heparin sodium detection information.

[0126] The process of each module realizing its own function in the heparin sodium detection system based on deep learning provided in the embodiment of the present application can be specifically referred to the aforementioned Figure 1 The description of the first embodiment is omitted here.

[0127] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0128] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0129] It will also be understood that the term "and / or" used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0130] As used in this specification and the appended claims, the term "if" can be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0131] In addition, in the description of the present specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish descriptions and should not be understood as indicating or implying relative importance. It should also be understood that although the terms "first", "second", etc. are used in the text to describe various elements in some embodiments of the present application, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first table can be named a second table, and similarly, a second table can be named a first table without departing from the scope of the various described embodiments. Both the first table and the second table are tables, but they are not the same table.

[0132] References to "one embodiment" or "some embodiments" in this specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in one or more embodiments of the present application. Thus, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," and "in other embodiments" appearing in various places in this specification do not necessarily refer to the same embodiment, but rather mean "one or more but not all embodiments," unless otherwise specifically emphasized. The terms "including," "comprising," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0133] The deep learning-based heparin sodium detection method provided in the embodiments of the present application can be applied to terminal devices such as mobile phones, tablet computers, wearable devices, vehicle-mounted devices, augmented reality (AR) / virtual reality (VR) devices, laptop computers, ultra-mobile personal computers (UMPCs), netbooks, and personal digital assistants (PDAs). The embodiments of the present application do not impose any restrictions on the specific type of terminal device.

[0134] For example, the terminal device can be a station (STAION, ST) in a WLAN, a cellular phone, a cordless phone, a Session Initiation Protocol (SIP) phone, a Wireless Local Loop (WLL) station, a Personal Digital Assistant (PDA) device, a handheld device with wireless communication function, a computing device or other processing device connected to a wireless modem, a vehicle-mounted device, a vehicle networking terminal, a computer, a laptop computer, a handheld communication device, a handheld computing device, a satellite wireless device, a wireless modem card, a TV set-top box (STB), customer premise equipment (CPE) and / or other devices for communicating on a wireless system and a next-generation communication system, such as a mobile terminal in a 5G network or a mobile terminal in a future evolved Public Land Mobile Network (PLMN) network.

[0135] As an example and not a limitation, when the terminal device is a wearable device, the wearable device can also be a general term for wearable devices that are intelligently designed and developed using wearable technology for daily wear, such as glasses, gloves, watches, clothing and shoes. A wearable device is a portable device that is worn directly on the body or integrated into the user's clothes or accessories. Wearable devices are not only hardware devices, but also achieve powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable smart devices include those that are full-featured, large in size, and can achieve complete or partial functions without relying on smartphones, such as smart watches or smart glasses, as well as those that only focus on a certain type of application function and need to be used in conjunction with other devices such as smartphones, such as various smart bracelets and smart jewelry for vital sign monitoring.

[0136] Figure 9 This is a schematic diagram of the structure of a terminal device provided by an embodiment of the present application. Figure 9 As shown, the terminal device 9 of this embodiment includes: at least one processor 90 ( Figure 9 Only one is shown), a memory 91, wherein the memory 91 stores a computer program 92 that can be run on the processor 90. When the processor 90 executes the computer program 92, the steps in the above-mentioned embodiments of the heparin sodium detection method based on deep learning are implemented, for example Figure 1 Alternatively, when the processor 90 executes the computer program 92, the functions of the modules / units in the above-mentioned system embodiments are realized, for example, Figure 8Functions of modules 810 to 840 are shown.

[0137] The terminal device 9 can be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The terminal device can include, but is not limited to, a processor 90 and a memory 91. It can be understood by those skilled in the art that Figure 9 It is only an example of the terminal device 9 and does not constitute a limitation on the terminal device 9. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the terminal device may also include an input and sending device, a network access device, a bus, etc.

[0138] The processor 90 may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0139] In some embodiments, the memory 91 may be an internal storage unit of the terminal device 9, such as a hard drive or memory of the terminal device 9. The memory 91 may also be an external storage device of the terminal device 9, such as a plug-in hard drive, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash memory card, etc. Furthermore, the memory 91 may include both an internal storage unit of the terminal device 9 and an external storage device. The memory 91 is used to store an operating system, application programs, a boot loader, data, and other programs, such as the program code of the computer program. The memory 91 may also be used to temporarily store data that has been sent or is about to be sent.

[0140] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0141] An embodiment of the present application also provides a terminal device, which includes at least one memory, at least one processor, and a computer program stored in the at least one memory and executable on the at least one processor. When the processor executes the computer program, the terminal device implements the steps of any of the above-mentioned method embodiments.

[0142] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned various method embodiments can be implemented.

[0143] An embodiment of the present application provides a computer program product. When the computer program product is run on a terminal device, the terminal device can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0144] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, it can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can include: any entity or system that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium.

[0145] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0146] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0147] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0148] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A heparin sodium detection method based on deep learning, characterized in that: include: Obtaining heparin sodium detection spectral signal information and heparin sodium detection electrochemical signal information; Performing feature extraction on the heparin sodium detection spectral signal information and the heparin sodium detection electrochemical signal information to obtain heparin sodium detection spectral signal feature information and heparin sodium detection electrochemical signal feature information; Generate multiple heparin sodium detection signal optimization feature information according to the heparin sodium detection spectral signal feature information, the heparin sodium detection electrochemical signal feature information, the preset heparin sodium detection signal feature optimization iteration number, and the preset heparin sodium detection signal feature optimization amplitude information; Based on the preset heparin sodium detection signal feature classification convergence threshold information, a plurality of heparin sodium detection signal optimization feature information is classified and processed to obtain heparin sodium detection information.

2. The heparin sodium detection method based on deep learning according to claim 1, characterized in that: The step of extracting features from the heparin sodium detection spectral signal information and the heparin sodium detection electrochemical signal information to obtain the heparin sodium detection spectral signal feature information and the heparin sodium detection electrochemical signal feature information specifically includes: performing time alignment processing and filtering processing on the heparin sodium detection spectral signal information and the heparin sodium detection electrochemical signal information to obtain heparin sodium detection spectral signal denoising information and heparin sodium detection electrochemical signal denoising information; performing convolution calculation on the denoised information of the heparin sodium detection spectral signal and the denoised information of the heparin sodium detection electrochemical signal according to a preset heparin sodium detection signal feature extraction matrix to obtain a plurality of sub-feature information of the heparin sodium detection spectral signal and a plurality of sub-feature information of the heparin sodium detection electrochemical signal; According to multiple preset heparin sodium detection signal sub-feature enhancement matrices, the multiple heparin sodium detection spectral signal sub-feature information and the multiple heparin sodium detection electrochemical signal sub-feature information are interactively enhanced to obtain heparin sodium detection spectral signal feature information and heparin sodium detection electrochemical signal feature information.

3. The heparin sodium detection method based on deep learning according to claim 2, characterized in that: The plurality of preset heparin sodium detection signal sub-feature enhancement matrices include a preset first heparin sodium detection signal sub-feature enhancement matrix, a preset second heparin sodium detection signal sub-feature enhancement matrix, and a preset third heparin sodium detection signal sub-feature enhancement matrix; The step of interactively enhancing the plurality of heparin sodium detection spectral signal sub-feature information and the plurality of heparin sodium detection electrochemical signal sub-feature information according to the plurality of preset heparin sodium detection signal sub-feature enhancement matrices to obtain the heparin sodium detection spectral signal feature information and the heparin sodium detection electrochemical signal feature information specifically includes: Performing feature interactive fusion processing on the plurality of heparin sodium detection spectral signal sub-feature information and the plurality of heparin sodium detection electrochemical signal sub-feature information to obtain heparin sodium detection signal sub-feature fusion information; Obtaining first heparin sodium detection signal sub-feature enhancement information according to the heparin sodium detection signal sub-feature fusion information and a preset first heparin sodium detection signal sub-feature enhancement matrix; Obtaining second heparin sodium detection signal sub-feature enhancement information according to the heparin sodium detection signal sub-feature fusion information and a preset second heparin sodium detection signal sub-feature enhancement matrix; Obtaining third heparin sodium detection signal sub-feature enhancement information according to the heparin sodium detection signal sub-feature fusion information and a preset third heparin sodium detection signal sub-feature enhancement matrix; Obtaining heparin sodium detection signal sub-feature interactive reinforcement information according to the first heparin sodium detection signal sub-feature reinforcement information and the second heparin sodium detection signal sub-feature reinforcement information; Obtaining heparin sodium detection signal feature information according to the heparin sodium detection signal sub-feature interaction enhancement information and the third heparin sodium detection signal sub-feature enhancement information; The heparin sodium detection signal characteristic information is separated and processed to obtain heparin sodium detection spectral signal characteristic information and heparin sodium detection electrochemical signal characteristic information.

4. The heparin sodium detection method based on deep learning according to claim 1, characterized in that The step of generating a plurality of heparin sodium detection signal optimization feature information according to the heparin sodium detection spectral signal feature information, the heparin sodium detection electrochemical signal feature information, the preset heparin sodium detection signal feature optimization iteration number, and the preset heparin sodium detection signal feature optimization amplitude information specifically includes: performing fusion encoding processing on the heparin sodium detection spectral signal characteristic information and the heparin sodium detection electrochemical signal characteristic information to generate a plurality of heparin sodium detection signal characteristic individual information; Based on the preset heparin sodium detection signal feature optimization boundary, and according to the plurality of heparin sodium detection signal feature individual information, a plurality of initial heparin sodium detection signal feature individuals are generated; Calculating a plurality of heparin sodium detection signal characteristic fitnesses based on the plurality of initial heparin sodium detection signal characteristic individuals; taking the initial heparin sodium detection signal characteristic individual corresponding to the maximum value of the fitness of the plurality of heparin sodium detection signal characteristics as the reference heparin sodium detection signal characteristic individual; generating a plurality of target heparin sodium detection signal feature individuals according to the plurality of initial heparin sodium detection signal feature individuals, the reference heparin sodium detection signal feature individuals, the preset number of heparin sodium detection signal feature optimization iterations, and the preset heparin sodium detection signal feature optimization amplitude information; A plurality of heparin sodium detection signal optimization feature information is generated based on the plurality of target heparin sodium detection signal feature individuals.

5. The heparin sodium detection method based on deep learning according to claim 4, characterized in that: The step of calculating the fitness of multiple heparin sodium detection signal characteristics based on the multiple initial heparin sodium detection signal characteristic individuals specifically includes: performing dimensionality reduction processing on the plurality of individual features of the initial heparin sodium detection signal to obtain dimensionality reduction information of the plurality of initial heparin sodium detection signal features; Extracting information from the plurality of initial heparin sodium detection signal feature dimensionality reduction information to obtain a plurality of heparin sodium detection spectral signal feature dimensionality reduction information and heparin sodium detection electrochemical signal feature dimensionality reduction information; Calculating a plurality of heparin sodium detection signal attenuation information according to the plurality of heparin sodium detection spectral signal feature dimensionality reduction information, the heparin sodium detection electrochemical signal feature dimensionality reduction information, and a preset heparin sodium detection signal attenuation calculation function; Multiple heparin sodium detection signal feature fitnesses are calculated based on the multiple heparin sodium detection spectral signal feature dimensionality reduction information, the heparin sodium detection electrochemical signal feature dimensionality reduction information, and the preset heparin sodium detection signal feature fitness weight information.

6. The heparin sodium detection method based on deep learning according to claim 4, characterized in that: The step of generating a plurality of target heparin sodium detection signal characteristic individuals based on the plurality of initial heparin sodium detection signal characteristic individuals, the reference heparin sodium detection signal characteristic individuals, the preset heparin sodium detection signal characteristic optimization iteration number, and the preset heparin sodium detection signal characteristic optimization amplitude information specifically includes: performing iterative optimization processing on the plurality of initial heparin sodium detection signal characteristic individuals according to the reference heparin sodium detection signal characteristic individual and the preset heparin sodium detection signal characteristic optimization amplitude information, to obtain a plurality of intermediate heparin sodium detection signal characteristic individuals and the number of iterative optimization times of the heparin sodium detection signal characteristic individual; Determining whether the number of iterative optimizations of the individual heparin sodium detection signal feature is equal to a preset number of iterative optimizations of the heparin sodium detection signal feature; If yes, taking the plurality of intermediate heparin sodium detection signal characteristic individuals as a plurality of target heparin sodium detection signal characteristic individuals; If not, calculating multiple heparin sodium detection signal feature iterative optimization fitnesses based on the multiple intermediate heparin sodium detection signal feature individuals; The plurality of intermediate heparin sodium detection signal characteristic individuals are used as initial heparin sodium detection signal characteristic individuals, and the intermediate heparin sodium detection signal characteristic individuals corresponding to the maximum values ​​of the iterative optimization fitness of the plurality of heparin sodium detection signal characteristics are used as reference heparin sodium detection signal characteristic individuals. The process returns to the step of performing iterative optimization processing on the plurality of initial heparin sodium detection signal characteristic individuals based on the reference heparin sodium detection signal characteristic individuals and the preset heparin sodium detection signal characteristic optimization amplitude information to obtain a plurality of intermediate heparin sodium detection signal characteristic individuals and the number of iterative optimization times of the heparin sodium detection signal characteristic individuals.

7. The heparin sodium detection method based on deep learning according to claim 1, characterized in that: The step of classifying and processing the plurality of heparin sodium detection signal optimization feature information based on the preset heparin sodium detection signal feature classification convergence threshold information to obtain heparin sodium detection information specifically includes: According to the preset heparin sodium detection signal optimization feature extraction quantity information, a plurality of heparin sodium detection signal optimization feature information is randomly extracted to obtain a plurality of central heparin sodium detection signal optimization feature information; calculating the logical distances of the plurality of heparin sodium detection signal optimization feature information and the plurality of central heparin sodium detection signal optimization feature information to obtain a plurality of heparin sodium detection signal optimization feature distance information; According to the plurality of heparin sodium detection signal optimization feature distance information and the central heparin sodium detection signal optimization feature information, a plurality of heparin sodium detection signal optimization feature class information is obtained; the heparin sodium detection signal optimization feature class information corresponds one-to-one to the central heparin sodium detection signal optimization feature information; one heparin sodium detection signal optimization feature class information includes a plurality of heparin sodium detection signal optimization feature information; Calculating the average value of each of the heparin sodium detection signal optimization feature information to obtain a plurality of heparin sodium detection signal optimization feature average value information; Calculating the average value information of the plurality of heparin sodium detection signal optimization feature information and the convergence distance value of the central heparin sodium detection signal optimization feature information to obtain the heparin sodium detection signal optimization feature classification convergence value information; Determining whether the heparin sodium detection signal optimization feature classification convergence value information is less than a preset heparin sodium detection signal feature classification convergence threshold information; If yes, optimizing the characteristic information according to the plurality of heparin sodium detection signals to obtain heparin sodium detection information; If not, the average value information of the plurality of heparin sodium detection signal optimization characteristics is used as the central heparin sodium detection signal optimization characteristic information, and the process returns to the step of calculating the logical distance between the plurality of heparin sodium detection signal optimization characteristic information and the plurality of central heparin sodium detection signal optimization characteristic information to obtain the plurality of heparin sodium detection signal optimization characteristic distance information.

8. A heparin sodium detection system based on deep learning, characterized in that: include: A heparin sodium detection signal information acquisition module is used to obtain heparin sodium detection spectral signal information and heparin sodium detection electrochemical signal information; a heparin sodium detection signal characteristic information generation module, configured to extract characteristics of the heparin sodium detection spectral signal information and the heparin sodium detection electrochemical signal information to obtain heparin sodium detection spectral signal characteristic information and heparin sodium detection electrochemical signal characteristic information; a heparin sodium detection signal optimization characteristic information generation module, configured to generate a plurality of heparin sodium detection signal optimization characteristic information based on the heparin sodium detection spectral signal characteristic information, the heparin sodium detection electrochemical signal characteristic information, a preset heparin sodium detection signal characteristic optimization iteration number, and a preset heparin sodium detection signal characteristic optimization amplitude information; The heparin sodium detection information generating module is used to classify and process the plurality of heparin sodium detection signal optimization feature information based on the preset heparin sodium detection signal feature classification convergence threshold information to obtain heparin sodium detection information.

9. A terminal device, characterized in that: The terminal device includes a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images