Dressing containing colorimetric area and wound monitoring method based on dressing

Abstract

The embodiment of the invention provides a wound dressing and a wound monitoring method based on the wound dressing containing a colorimetric area, and relates to the technical field of wound dressing technologies. The dressing comprises a backing and an acting piece, wherein the acting piece is fixedly connected with the backing; a colorimetric groove is formed in one side of the backing, a plurality of observation holes are formed in the colorimetric groove, a preset colorimetric area is inserted into the colorimetric groove, and colorimetric observation is carried out at the observation holes. According to the invention, the problem that the wound cannot be effectively observed is solved, and the effect of improving the wound observation precision is further achieved.

Description

Technical Field

[0001] This invention relates to the field of wound dressings, and more specifically, to a dressing containing a colorimetric region and a wound monitoring method based on the dressing. Background Technology

[0002] Effective management of chronic wounds is a significant challenge in clinical medicine, as the healing process is influenced by a complex interplay of various biochemical and physical factors, including infection, pH, temperature, and humidity.

[0003] Existing dressings can only cover wounds and promote healing, but they lack means to monitor the condition of the wound and can only be observed by the human eye. This method cannot effectively monitor the wound healing process and is prone to misjudgment. Summary of the Invention

[0004] This invention provides a dressing containing a colorimetric region and a wound monitoring method based on the dressing, in order to at least solve the problem of low wound observation accuracy in related technologies.

[0005] According to the present invention, a wound dressing comprising a colorimetric region is provided, comprising: The backing and the active component are fixedly connected to the backing; a colorimetric groove is provided on one side of the backing, and the colorimetric groove has several observation holes. A preset colorimetric area is inserted into the colorimetric groove and colorimetric observation is performed at the observation holes.

[0006] Specifically, the observation holes are arranged sequentially along the length of the colorimetric groove.

[0007] In an exemplary embodiment, the surface of the backing away from the actuating member is provided with positioning lines, which divide the surface of the backing away from the actuating member into several observation areas.

[0008] In particular, a dressing-based wound monitoring method is provided, comprising: Obtain the observation color development state matrix of the colorimetric region of the target observation hole, wherein the observation color development state matrix represents the original observation state of the color development array composed of multiple color development units; A distortion evaluation function is constructed, which is used to quantify and evaluate the color distortion of the color rendering state matrix; An iterative energy minimization correction process is performed, using the observed colorimetric state matrix as the initial input. The state of the colorimetric units in the colorimetric state matrix is ​​iteratively corrected to obtain the intrinsic physiological state map that makes the distortion evaluation function value converge to the minimum value. Based on the intrinsic physiological state map, the physiological state of the wound is determined.

[0009] Specifically, the distortion evaluation function includes a local distortion term and a coupled distortion term; the local distortion term is used to measure the fidelity between the state of each color unit in the color state matrix and the original observed state of the corresponding color unit in the observed color state matrix; the coupled distortion term is used to measure the state consistency between adjacent color units in the color state matrix based on the inter-unit coupling correlation model.

[0010] Specifically, the iterative energy minimization correction process includes: Set the initial color rendering state matrix as the observed color rendering state matrix; In each iteration, a candidate color rendering state matrix is ​​generated, and the candidate distortion degree corresponding to the candidate color rendering state matrix is ​​calculated; Based on the difference between the candidate distortion degree and the current distortion degree of the current color rendering state matrix, and the perturbation acceptance threshold, it is determined whether to accept the candidate color rendering state matrix as the new current color rendering state matrix according to the preset acceptance criteria. The disturbance acceptance threshold is iteratively attenuated.

[0011] Specifically, generating the candidate color state matrix includes: randomly selecting a color unit in the current color state matrix and applying a random perturbation to the state vector of the color unit.

[0012] Specifically, determining the physiological state of the wound based on the intrinsic physiological state map includes: Extract feature vectors from the intrinsic physiological state map; The feature vector is compared with a preset wound state database, which stores standard intrinsic physiological state maps or standard feature vectors corresponding to various known wound physiological states. The diagnostic result is one or more wound physiological states that have the highest matching degree with the feature vector.

[0013] This invention uses a colorimetric area to compare and judge the color changes of dressings, and then judges the wound condition based on the corresponding wound condition. Therefore, it can solve the problem of low wound judgment accuracy and improve the accuracy of wound judgment. Attached Figure Description

[0014] Figure 1 This is a three-dimensional structural schematic diagram of a dressing according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the planar structure of a dressing according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the colorimetric region structure of a dressing according to an embodiment of the present invention; Figure 4 This is a flowchart of a wound monitoring method based on wound dressing according to an embodiment of the present invention.

[0015] In the diagram: 1. Backing; 2. Functional component; 3. Limiting frame; 4. Observation hole; 5. Colorimetric area; 6. Positioning line; 7. Observation strip. Detailed Implementation

[0016] The technical solutions of the embodiments of this application will be described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0017] In the following description, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0018] Furthermore, in this application, directional terms such as "upper," "lower," "left," and "right" may be defined relative to the orientation of the components shown in the accompanying drawings. It should be understood that these directional terms can be relative concepts, used for relative description and clarification, and may change accordingly depending on the orientation of the components in the accompanying drawings.

[0019] In this application, unless otherwise expressly specified and limited, the term "connection" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral part; it can be a direct connection or an indirect connection through an intermediate medium. Furthermore, the term "coupled" can refer to an electrical connection that enables signal transmission.

[0020] As used herein, “about,” “approximately,” or “approximately” includes the stated value and the average value within an acceptable range of deviation from the given value, wherein the acceptable range of deviation is determined by a person skilled in the art taking into account the measurement under discussion and the error associated with the measurement of the given quantity (i.e., the limitations of the measurement system).

[0021] Example 1 This embodiment provides a dressing containing a colorimetric region, such as Figure 1-2As shown, it includes a backing 1 for fixing to the skin near the wound and an action member 2 for healing and covering the wound. The action member 2 is located on the lower surface of the backing 1 and is fixedly connected to the backing 1. A colorimetric area 5 is provided on the side of the backing 1 away from the surface of the action member 2. A limiting frame 3 is provided on the colorimetric area 5. An observation strip 7 is movably inserted in the limiting frame 3. In order to facilitate colorimetric observation, the observation strip of the colorimetric area 5 has several observation holes 4. The observation holes 4 are arranged sequentially along the color depth distribution direction of the colorimetric area to facilitate colorimetric observation under different conditions.

[0022] like Figure 3 As shown, in this embodiment, the colorimetric area is divided into several color regions, each representing a standard color collected at different stages of wound healing. For example... Figure 3 The color blocks in the middle, from left to right, represent the four areas of wound healing. They are the standard colors for the four different stages of wound healing. From the far left to the far right, they represent the first stage, the second stage, the third stage, and the fourth stage, respectively. Phase 1: The wound has just healed, and the scar will usually appear bright red or pink. Second stage: Over time, the color will gradually darken to a deep red or purplish-red; The third stage: When the scar has fully healed and matured, its color will gradually lighten. Phase 4: Eventually, it will gradually turn back to pink or skin tone.

[0023] Figure 3 This is merely one example of a scenario described in this embodiment, and is not limited to this one. Figure 3 The division of the colorimetric region can also be a gradient, formed by the gradual change of the standard colors of the above four regions.

[0024] Furthermore, to facilitate the location and observation of the wound and the changes in the wound of the active component 2, a positioning line 6 is provided on the surface of the backing 1 away from the active component 2. The positioning line 6 divides the surface of the backing 1 away from the active component 2 into several observation areas. By determining the color change of the active component 2 in different observation areas, the changes such as wound healing can be determined.

[0025] Example 2 The difference from Example 1 is that, in order to achieve automatic observation of wound changes and improve observation accuracy, this example provides a dressing-based wound monitoring method. Figure 4 This is a flowchart of a wound monitoring method based on wound dressing according to an embodiment of the present invention. The method determines wound changes by monitoring the colorimetric area 5 and the color changes of the active ingredient and the dressing. The specific process includes the following steps: Step S100: Obtain an observation colorimetric state matrix of the colorimetric region 5 of the target observation hole. The observation colorimetric state matrix represents the original observation state of the colorimetric array composed of multiple colorimetric units.

[0026] In one embodiment, S100 is the foundation for all subsequent data processing and analysis, aiming to transform image information from the physical world into structured, mathematically processable data entities, namely, the observed colorimetric state matrix. To ensure that the matrix can reflect the original state of colorimetric region 5 as accurately as possible, and to provide high-quality input for subsequent correction algorithms, this step involves a series of sophisticated image processing and data extraction operations.

[0027] First, image acquisition is performed; the user operates the built-in camera to take a picture of an observation hole 4 area on the colorimetric area 5 of the dressing 101 covering the wound; in order to reduce the interference of ambient light on subsequent analysis, some standardized lighting control measures can be adopted during the shooting; optionally, a standard color reference block (such as a standard gray block) can be placed next to the colorimetric area 5 during the shooting, or the device can record the ambient light parameters (such as color temperature and brightness) during the shooting, which can be used for subsequent color correction.

[0028] Next, image preprocessing is performed. Since the acquired raw digital images often differ from the ideal colorimetric region 5 image due to factors such as shooting angle, lens distortion, and lighting conditions, the processor performs a series of preprocessing operations on the raw images to eliminate these differences. These operations include white balance correction, geometric correction, and image registration. White balance correction uses white areas in the image or preset standard color reference blocks to adjust the overall color cast of the image, making its color representation closer to reality. Geometric correction aims to eliminate perspective distortion caused by the tilt of the shooting angle, as well as barrel or pincushion distortion caused by the camera lens. Through affine transformations or more complex nonlinear transformations, the colorimetric region 5 in the image is restored to a standard rectangle. Image registration precisely aligns the geometrically corrected colorimetric region 5 image with a pre-stored standard colorimetric region 5 template in the device, ensuring that each colorimetric unit is located at the expected coordinate position. This provides a precise spatial reference for subsequent unit segmentation and data extraction.

[0029] Subsequently, colorimetric unit segmentation and state extraction are performed; after image registration is completed, the processor precisely segments the colorimetric region 5 image region into a segment based on the layout information defined in the standard template. A grid composed of individual colorimetric units is called a "colorimetric array"; for each colorimetric unit in the array, its index is... .

[0030] Next, state information is extracted from the image sub-region corresponding to each color unit; the most common state information is color; specifically, the average RGB color value of all pixels in each sub-region is calculated to obtain a three-dimensional state vector. The The values ​​are usually normalized to The floating-point range or representation is The state vector is an integer. In other embodiments, the state vector may also include components of other color spaces (such as HSV, Lab), or include texture features, to more comprehensively describe the state of the color unit.

[0031] Finally, construct the observed color state matrix. The processor will display the state vectors of all color units. Based on their spatial positions within the colorimetric array, they are organized into a three-dimensional tensor or matrix, namely the observed colorimetric state matrix. The dimension of this matrix is... ,in and These are the number of rows and columns in the colorimetric array. It is the feature dimension that describes the state of each unit (for RGB color, ),this The matrix is ​​a direct numerical representation of the color relaxation phenomenon. It contains the true color changes caused by the biochemical information of the wound, and also superimposed the noise and distortion caused by the various physical and chemical disturbances mentioned above. It is the sole input data source for all subsequent correction and analysis steps.

[0032] For example, suppose a colorimetric region 5 is designed as a The colorimetric array, i.e. After the camera captures images and the processing terminal performs preprocessing and segmentation, nine color units are obtained. The processing terminal extracts the average RGB value (range 0-255) of each unit; for example, the color unit located in the center... The corresponding image area may have slightly uneven lighting, and its true ideal color should be... However, the actual observed average RGB value is The processing terminal performs this operation on all nine units, ultimately building a... Observational color state matrix ;For example, The specific values ​​can be shown below (for example only): This matrix, containing nine three-dimensional vectors, is the final output of the S100 step. It captures the complete visual information of colorimetric region 5 at a specific point in time, but this information is the "relaxed" state, not the "intrinsic" state. For example, because the light is coming from the upper left corner, the R, G, and B values ​​of the first row may be systematically higher than the ideal values, and due to seepage diffusion, the color of the central cell may affect its neighboring cells. These factors all contribute to the final output. This is reflected in the numerical values; the core task of the subsequent steps in this embodiment is to extract this information. In this process, computational methods are used to reconstruct the ideal, undisturbed "intrinsic physiological state map." The quality of this step directly determines the upper limit of the final diagnostic result. A precise... It can provide a solid data foundation for subsequent iterative corrections, ensuring that the correction process can converge to the correct state.

[0033] S200: Construct a distortion evaluation function, which is used to quantitatively evaluate the color distortion of a color state matrix.

[0034] After obtaining the observed colorimetric state matrix Subsequently, the core task of S200 is to establish a mathematical model that can theoretically determine the color state matrix for any given color state. The "irrationality" or "instability" is quantitatively assessed. This assessment result is termed "color distortion," denoted as... The construction of this function is one of the core innovations of this application. It transforms physical intuition (such as "colors should transition smoothly" and "results should not deviate too far from observations") into rigorous and computable mathematical formulas, providing a theoretical basis for finding the most "reasonable" and "stable" state through optimization methods.

[0035] In a preferred embodiment, the distortion evaluation function It is designed as a weighted sum of two key components. This structure embodies a trade-off: respecting the original observational data while imposing a priori physical constraint (such as spatial continuity). Its mathematical form is defined as: In this function, Represents any one to be evaluated 3D color state matrix. It is a local distortion term. It is a coupling distortion term, and It is a coupling weight coefficient used to balance these two importances.

[0036] The local distortion term Its function is to ensure that the corrected result does not deviate indefinitely from the original observation data; it plays the role of "data fidelity." This item measures the matrix to be evaluated. The state of each colorimetric unit in the matrix corresponds to the matrix we actually observe in S100. The smaller the difference between the states of corresponding units, the better. The higher the "loyalty" to the original observation, the smaller the penalty for this item. Its specific mathematical expression is: in, and These are the row and column indices of the color unit in the matrix, and are dimensionless integers. The matrix to be evaluated Middle position The state vector of the color unit (a k-dimensional vector), whose components are in units of color intensity (e.g., integers from 0 to 255). Is The observed state vector at the corresponding position in the middle, its unit is . same; This represents the calculation of the square of the Euclidean distance between two vectors, with the result in the square of the color intensity unit. A total local distortion degree is obtained by summing the squared distances of all cells in the entire array.

[0037] The coupling distortion term Its function is to introduce a priori knowledge of the physical world, namely that changes in physical quantities are usually continuous or smooth in space; for example, changes in light intensity are usually gradual, and the diffusion of exudate concentration also follows a gradient law. This term aims to penalize matrix states that exhibit drastic and illogical color jumps in space. It measures the consistency of states between adjacent color-forming units through an "inter-unit coupling correlation model," resulting in a smooth and internally self-consistent matrix. ,That The value will be very low; its specific mathematical expression can be defined as: in, It is a color rendering unit The set of neighborhoods, for example, can be the simplest 4-neighborhood (top, bottom, left, right) or 8-neighborhood. It is the index of a neighboring cell in the neighborhood set. It is the neighborhood coupling weight, a dimensionless scalar used to represent the coupling strength between different neighbors; in the simplest case, all weights can be set to 1; in more complex models, the weights can be inversely proportional to the distance. The degree of difference between the state of the central cell and its neighboring cells was calculated, with the unit being the square of the color intensity unit; by weighted summation of the differences between all cells and all their neighbors, we obtained a total coupling distortion degree.

[0038] The coupling weight coefficient It is a key hyperparameter, a dimensionless scalar, that controls the balance between "data fidelity" and "spatial smoothness." If the value is large, the optimization process will tend to produce a very smooth result, but may sacrifice similarity to the original observations; conversely, if... If the value is small, the result will be very close to the original observation matrix, but it may not be effective in removing noise. The optimal value needs to be determined based on the specific dressing material, the characteristics of the developer, and typical application scenarios, through experimental methods such as cross-validation. For example, its value range can be... between.

[0039] For example, we continue to use the definition in S100 Examples of colorimetric arrays.

[0040] Suppose we need to evaluate a candidate state matrix The degree of distortion, especially in the central unit Contribution to total distortion; assuming in middle, In the candidate matrix In the middle, the value of the central unit is corrected to So, this unit is for The contributions are as follows: (Unit: squared unit of color intensity) Now calculate the unit pair . contributions.

[0041] Consider its relationship with the neighbors above. The coupling relationship, assuming in middle, Assuming neighborhood coupling weights If the value is 1, then this pair of neighbors... The contributions are as follows: (Unit: squared unit of color intensity) The processing terminal needs to repeat this calculation for all 9 cells and all their neighbors, and then... The sum and The sum according to (For example, suppose) Only by performing a weighted summation can the entire matrix be obtained. Total distortion ,this The smaller the value, the more it represents The more "reasonable" this state is in our model, the better; the completion of S200 leads to the search for the minimum value in S300. The matrix of values ​​laid the mathematical foundation.

[0042] S300: Perform an iterative energy minimization correction process, using the observed colorimetric state matrix as the initial input, and iteratively correct the state of the colorimetric units in the colorimetric state matrix to obtain an intrinsic physiological state map that makes the distortion evaluation function value converge to a minimum value.

[0043] After defining the function for evaluating the system's "distortion degree," the goal of S300 is to find a color rendering state matrix. , making The value reaches the global minimum; this matrix that reaches the minimum distortion degree is defined as the intrinsic physiological state map. ;because The function form is complex and may have a large number of local minima. Directly using methods such as gradient descent can easily lead to local optima. Therefore, this embodiment adopts a global optimization strategy inspired by the simulated annealing algorithm, namely the iterative energy minimization correction process, to systematically explore the solution space and find the global optimum with a high probability.

[0044] This process can be understood as a simulated physical process: a system starts from a high-energy (high-distortion) initial state, and through a series of random perturbations and controlled cooling processes, eventually stabilizes in a ground state with the lowest energy (lowest distortion).

[0045] First, initialization is performed. The processor initializes the current color state matrix. Initialized to the observed colorimetric state matrix obtained in S100 This is a natural starting point because the observational data is the only source of information. Subsequently, the degree of distortion of this initial state is calculated. .

[0046] At the same time, a key control parameter is initialized: the disturbance acceptance threshold. This is commonly referred to as "temperature". Set it to a relatively high initial value. ,For example, (units and) The same, i.e., the square of the color intensity unit. A high initial temperature means that in the early stages of the process, the system has a higher probability of accepting a "worse" state (i.e., a state with higher distortion), which allows the algorithm to "escape" the trap of local minima and perform a broader global search. In addition, a temperature decay rate needs to be set. It is a positive real number slightly less than 1 (for example, (Dimensionless) is used to reduce the temperature after each iteration. Finally, the stopping condition for the iteration is set, which can be a preset maximum number of iterations. (For example (times), or when the temperature Reduce to a very small value (For example Stop when ).

[0047] Next, the core iterative loop begins. This loop continues to execute until the stopping condition is met. In each iteration, the following operations are performed: Generate candidate states: from Randomly select a colorimetric unit The state vector of this unit Apply a small random perturbation to generate a new candidate state vector. For example, a perturbation vector can be sampled from a Gaussian distribution with a mean of 0 and a small standard deviation (e.g., a standard deviation of 5 color intensity units) and added to the original vector; Replace the vector at the corresponding position in the middle with This yields a new candidate state matrix. .

[0048] Evaluate candidate states: Call the distortion evaluation function defined in S200 to calculate the distortion of this new candidate state matrix. Since only the state of one cell is changed at a time, this calculation can be optimized by simply recalculating the local distortion and coupling distortion terms associated with that cell, without having to recalculate the entire matrix, thus greatly improving efficiency.

[0049] Calculate the distortion change: Calculate the difference in distortion between the old and new states. .

[0050] Execute the acceptance criterion judgment: This is the core of the algorithm.

[0051] if This means that if the candidate state is better than the current state (lower distortion), then the system unconditionally accepts this new state. That is, updating... and .

[0052] if This means the candidate state is a "worse" state. In this case, the system doesn't reject it outright, but accepts it with a certain probability. This acceptance probability... As given by the Metropolis criterion: .

[0053] The processing terminal generates a in Uniformly distributed random numbers within the interval ,if Even if the new state is worse, the system will still accept it and update. and Otherwise, if If the perturbation is rejected, the system state remains unchanged. This probabilistic acceptance mechanism is key to the algorithm's ability to escape local optima; in Even at very high levels Larger, and more likely to be accepted; with As the probability of accepting a "worse" state decreases, the algorithm gradually converges to the optimal solution in the later stages.

[0054] Update the perturbation acceptance threshold: At the end of each iteration, decay the temperature. .

[0055] Finally, the loop terminates when the iteration process meets the stopping condition; at this point... The matrix, which is the state matrix with the lowest distortion found throughout the search process, is defined as the intrinsic physiological state map. It is a kind of The data structure has the same dimensions, but the values ​​inside it have been systematically corrected, and theoretically the effects of various physical and chemical disturbances have been eliminated. It has been transformed from a noisy observation state 301 to a smooth and self-consistent intrinsic state 302.

[0056] For example, continuing from the above An example of an array. Suppose the current position is the array... Next iteration, current temperature , The processing terminal randomly selects a unit. And by applying a perturbation, a [generated] ; calculation yielded ,therefore, .

[0057] because It needs to be accepted according to probability; the probability of acceptance is... The processor generates a random number. .because Therefore, even though this is a worse state, the system still accepts it and updates. , Then update the temperature. ; Proceed to the next iteration, assuming it is at the th iteration. After the second iteration and further perturbation, we obtain... ,but ;because The system unconditionally accepts and updates. and This process is repeated until... It becomes very small, the system state tends to stabilize, and no longer undergoes large changes.

[0058] S400: Determine the physiological state of the wound based on the intrinsic physiological state map.

[0059] The intrinsic physiological state map was successfully calculated using the S300. S400's next task is to transform this purified and corrected high-quality data into conclusions with clinical diagnostic significance. It is a numerical matrix, and in order to achieve automated diagnosis, it needs to be correlated with known wound conditions.

[0060] First, feature extraction is performed. This involves directly extracting the entire... Using matrices as input for comparison is computationally intensive and may contain redundant information. Therefore, it is usually necessary to start from... Extract a more compact and representative set of feature vectors. There are various methods for feature extraction, including: Statistical characteristics: calculation The mean, variance, skewness, kurtosis, etc. of each color channel.

[0061] Spatial characteristics: calculation The gradient spectrum can be used to analyze the direction and intensity of color changes, or to extract the coefficients in a specific transform domain (such as the Fourier domain or wavelet domain).

[0062] Dimensionality reduction features: Using techniques such as principal component analysis (PCA) and linear discriminant analysis (LDA), high-dimensional data is reduced in dimensionality. The data is projected into a low-dimensional feature space to obtain a low-dimensional but information-rich feature vector.

[0063] Based on model characteristics: For a specific chromogenic unit (e.g., a pH-sensitive unit), its value can be directly taken from... The state values ​​in the data are used as features.

[0064] Next, database matching is performed. For diagnosis, a "wound state-intrinsic atlas" database needs to be pre-constructed. The construction of this database is itself an important offline process, involving: collecting a large number of clinical wound samples and determining the precise physiological state of each sample using gold standard methods (such as microbial culture and biochemical testing) (e.g., "normal healing period," "Staphylococcus aureus infection," "Pseudomonas aeruginosa infection," "pH=7.0," etc.). Simultaneously, the colorimetric region images corresponding to these samples are processed using the method described in this application to obtain their respective "standard" intrinsic physiological state atlases or the "standard" feature vectors extracted from them. These "state-atlas / feature" pairs are stored in the database.

[0065] Finally, perform a status diagnosis. When a new, diagnosable wound is obtained... After extracting its feature vector, the processor compares this feature vector with all standard feature vectors stored in the database. Various pattern recognition algorithms can be used for this comparison.

[0066] The nearest k-NN algorithm calculates the distances (such as Euclidean distance and Mahalanobis distance) between the feature vector to be tested and all standard feature vectors in the database, and finds the closest one. A neighbor. According to this The diagnosis of the wound to be tested is determined by voting on the wound status tags of each neighbor.

[0067] Support Vector Machine (SVM): One or more SVM classifiers are pre-trained using data from a database. Each classifier is used to distinguish between two or more wound states. The feature vector to be tested is input into the trained SVM model, and the model outputs its class.

[0068] Neural Networks: Construct a deep neural network (DNN) or convolutional neural network (CNN) to... Alternatively, the extracted features can be used as input, and the network's output layers correspond to different wound state categories. By training on a database, the network can learn a complex nonlinear mapping from the graph to the state.

[0069] The diagnostic output can be a single most probable state, or multiple possible states and their corresponding confidence scores. For example, the output could be: "Staphylococcus aureus infection (confidence 92%), Pseudomonas aeruginosa infection (confidence 5%)".

[0070] For example, suppose that after feature extraction, from of A 10-dimensional feature vector was obtained. The database stores the standard feature vector centers for two types of wound states ("State A: Infected", "State B: Healed"), which are respectively and Using the simplest nearest neighbor algorithm (k=1), the processor calculates... and and The Euclidean distance.

[0071] because The system determines that the state of the wound to be tested is closer to "State A: Infection" and outputs this diagnosis result.

[0072] Based on the core process described above, this method can be further developed into a method that can predict the future state of a wound and provide prospective guidance for clinical intervention.

[0073] Specifically, it includes the following steps S501-S503: S501: Time Series Data Construction For the same wound, steps S100-S300 were repeated at multiple consecutive time points (e.g., every 12 hours) to obtain a series of intrinsic physiological state maps ordered by time: This sequence data records the "trajectory" of the wound's state evolving over time.

[0074] S502: Healing Trajectory Modeling A dynamic prediction model is trained using a large amount of historically accumulated time-series data. Due to the long-term temporal dependence of the wound healing process, recurrent neural networks (RNNs) and their variants, such as long short-term memory networks (LSTMs) or gated recurrent units (GRUs), are ideal for this type of modeling. The model's task is to learn a mapping function. This function can be based on past spectral sequence at each time point This is used to predict the next time point. map Here, the model takes a sequence of graphs as input and outputs a predicted graph.

[0075] S503: Trend Forecasting and Early Warning In actual monitoring, the latest continuous data collected will be used. Each intrinsic physiological state map sequence is input into a pre-trained dynamic model, which outputs a prediction of the future map. The system makes a prediction; subsequently, the S400 diagnostic steps can be performed on this predicted pattern. If the diagnosis shows a tendency for the wound condition to worsen (e.g., a significant increase in the characteristic value representing infection in the prediction), the system can issue an early warning to healthcare professionals; this prediction-based feedforward control allows for earlier clinical intervention, rather than waiting until the wound has shown obvious signs of deterioration, thus buying valuable treatment time for the patient.

[0076] For example, suppose the model is a simplified LSTM that receives data from the past two time points ( The graph of ) predicts A timeline. In At that moment, the system obtained and These two matrices are flattened into vectors and fed into the LSTM model. The gating units and hidden states inside the model are updated based on these two inputs, and the final output is a prediction vector, which is then reshaped into the predicted graph. Assuming The diagnosis was "mild inflammation," After diagnosis, its similarity to the standard pattern of "Staphylococcus aureus infection" increased sharply from 30% to 70%. At this point, the system will trigger an alert, indicating that "the wound has a high risk of infection, and it is recommended to have a follow-up examination or adjust the treatment plan."

[0077] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0078] Embodiments of the present invention also provide a computer-readable storage medium storing a computer program, wherein the computer program is configured to perform the steps in any of the above method embodiments when executed.

[0079] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0080] Embodiments of the present invention also provide an electronic device including a memory and a processor, the memory storing a computer program and the processor being configured to run the computer program to perform the steps in any of the above method embodiments.

[0081] In one exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor and the input / output device is connected to the processor.

[0082] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0083] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0084] The units described as separate components may or may not be physically separate. A component shown as a unit can be one or more physical units; that is, it can be located in one place or distributed in multiple different locations. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0085] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0086] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, in essence, or the parts that contribute to the prior art, or all or part of the technical solutions, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0087] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A dressing comprising a colorimetric region, characterized in that, The application relates to a wound dressing, comprising: a backing and an action element fixedly connected to the backing; one side of the backing is provided with a colorimetric area, a limiting frame and an observation strip, a plurality of observation holes are formed in the observation strip, the colorimetric area limiting frame is arranged at two ends of the colorimetric area, and the observation strip is movably inserted into the limiting frame.

2. The dressing comprising a colorimetric region of claim 1, wherein, The observation holes are arranged in sequence along a color depth arrangement direction of the colorimetric area.

3. The dressing comprising a colorimetric region of claim 1, wherein, A positioning line is arranged on a surface of the backing away from the action element, and the positioning line divides the surface of the backing away from the action element into a plurality of observation areas.

4. The dressing comprising a colorimetric region of claim 1, wherein, The colorimetric area is divided into a plurality of color areas, and each color area is a standard color of a different period of wound healing.

5. The dressing comprising a colorimetric region of claim 4, wherein, The standard color of each color area is a fixed color or a gradient color.

6. A dressing-based wound monitoring method, characterized by, The application is applied to the wound dressing in claims 1-3, comprising: an observation color development state matrix of a colorimetric area of a target observation hole is obtained, the observation color development state matrix represents an original observation state of a color development array composed of a plurality of color development units; a distortion degree evaluation function is constructed, the distortion degree evaluation function is used for quantitatively evaluating color development distortion degrees of the color development state matrix; an iterative energy minimization correction process is performed, the color development state matrix is taken as initial input, the states of the color development units in the color development state matrix are iteratively corrected, and thus an intrinsic physiological state atlas making a distortion degree evaluation function value converge to a minimum value is obtained; a physiological state of the wound is determined based on the intrinsic physiological state atlas.

7. The dressing-based wound monitoring method of claim 4, wherein, The distortion degree evaluation function comprises a local distortion term and a coupling distortion term; the local distortion term is used for measuring fidelity between the states of the color development units in the color development state matrix and original observation states of corresponding color development units in the observation color development state matrix; the coupling distortion term is used for measuring state consistency between adjacent color development units in the color development state matrix based on a unit coupling correlation model.

8. The dressing-based wound monitoring method of claim 5, wherein, The application relates to a wound dressing, comprising: The iterative energy minimization correction process comprises: an initial color development state matrix is set as the observation color development state matrix; in each iteration, a candidate color development state matrix is generated, and a candidate distortion degree corresponding to the candidate color development state matrix is calculated; based on a difference value between the candidate distortion degree and a current distortion degree of a current color development state matrix and a perturbation acceptance threshold, whether the candidate color development state matrix is accepted as a new current color development state matrix is determined according to a preset acceptance criterion; the perturbation acceptance threshold is iteratively attenuated.

9. The dressing-based wound monitoring method of claim 6, wherein, The generation of the candidate color development state matrix comprises: a color development unit is randomly selected in the current color development state matrix, and a random perturbation is applied to a state vector of the color development unit.

10. The dressing-based wound monitoring method of claim 4, wherein, The determination of the physiological state of the wound based on the intrinsic physiological state atlas comprises: a feature vector is extracted from the intrinsic physiological state atlas; the feature vector is compared with a preset wound state database, the wound state database stores standard intrinsic physiological state atlases or standard feature vectors corresponding to a plurality of known wound physiological states; one or more wound physiological states with the highest matching degree with the feature vector are output as a diagnosis result.

Images