Intelligent application management method and application system based on flexible multi-mode sensor array

Abstract

The invention provides an intelligent application management method and application system based on a flexible multi-mode sensor array, relates to the technical field of medical treatment, and aims to solve the problems that an existing wound management method and intelligent dressing are insufficient in the aspects of real-time monitoring, objective evaluation and intelligent decision making. The method comprises the following steps: acquiring multi-dimensional physiological parameter data output by a flexible multi-modal sensor array; dividing the wound area into N grid units; processing the multi-dimensional physiological parameter data in the N grid units to obtain a characteristic value of each grid unit; constructing an N-order feature matrix reflecting the feature relationship among the N grid units; performing singular value decomposition on the N-order feature matrix, and extracting a maximum singular value as a main feature parameter; performing data alignment on the main characteristic parameters and a preset standard healing curve, and calculating to obtain a healing deviation index; determining a wound healing stage according to a preset threshold interval in which the healing deviation index is located; and generating a dressing replacement prompt signal.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of medical technology, and in particular to an intelligent patch management method and patch system based on a flexible multimodal sensor array. Background Art

[0002] With the global trend of aging population and the development of medical technology, the application of surgical operations is becoming increasingly popular. However, effective management of postoperative wounds remains a key link in affecting the quality of patient recovery and reducing the incidence of complications. Traditional wound management relies on medical staff's naked eye observation and experience judgment, supplemented by regular dressing changes and inspections. This method has many shortcomings. For example, frequent dressing changes can cause discomfort and pain to patients, especially for special wounds such as burns and chronic wounds. For example, manual assessment is subjective to a certain extent, making it difficult to achieve accurate quantitative analysis and real-time continuous monitoring of wound status. This may delay the detection of problems such as infection and abnormal exudation. In addition, frequent medical operations also increase the consumption of medical resources.

[0003] In recent years, smart dressing technology has garnered widespread attention as an emerging wound management solution. By integrating various sensors into dressings, real-time or near-real-time monitoring of wound microenvironmental parameters (such as temperature, humidity, pH, oxygen saturation, and biomarkers) can be achieved. Several wound dressings with sensor functionality have been disclosed in the prior art for monitoring wound status or wound microenvironmental parameters.

[0004] For example, Chinese patent application publication number CN109069712A discloses a sensor-enabled wound monitoring and treatment device, which includes a substantially flexible substrate supporting one or more sensors and a variety of sensor types. However, these existing technologies generally focus on the sensor itself or data acquisition, and there are still deficiencies in how to effectively and deterministically analyze and evaluate the collected multi-dimensional, high-throughput sensor data to extract key information to guide clinical decision-making. In particular, how to convert multimodal physiological parameter data with complex spatial distribution and temporal evolution into a single or a few indicators with clear physical or physiological meaning that can guide clinical practice, and ensure the automation and determinism of the evaluation process, is a technical challenge currently facing the field of intelligent wound management. Summary of the Invention

[0005] The embodiments of the present application provide an intelligent dressing management method and dressing system based on a flexible multimodal sensor array to improve the deficiencies of existing wound management methods and intelligent dressings in real-time monitoring, objective evaluation and intelligent decision-making.

[0006] To achieve the above objectives, the embodiments of the present application adopt the following technical solutions: In the first aspect, the present application provides an intelligent dressing management method based on a flexible multimodal sensor array, comprising the following steps: obtaining multidimensional physiological parameter data output by a flexible multimodal sensor array, the multidimensional physiological parameter data including temperature data, humidity data, pH value data and pressure distribution data; dividing the wound area into N grid units, where N is a preset integer; processing the multidimensional physiological parameter data within the N grid units to obtain the eigenvalues ​​of each grid unit; constructing an N-order characteristic matrix reflecting the characteristic relationship between the N grid units; performing singular value decomposition on the N-order characteristic matrix, and extracting the maximum singular value as the main characteristic parameter; aligning the main characteristic parameter with a preset standard healing curve to calculate a healing deviation index; determining the wound healing stage according to the preset threshold interval of the healing deviation index; and generating a dressing change reminder signal when the healing deviation index exceeds a first threshold.

[0007] By adopting the above method, the present application can integrate multi-dimensional sensor data, discretize spatial information using grid division, and extract the main characteristic parameters that characterize the overall wound status or key changes by constructing a feature matrix and performing singular value decomposition. The main characteristic parameters can effectively reduce the dimension and capture the core changes of multi-dimensional data. Furthermore, by aligning the data with the standard healing curve and calculating the healing deviation index, an objective quantitative evaluation of the wound healing process is achieved, and the healing stage is judged based on a determined threshold interval and a prompt signal is generated according to the deviation, so that the wound management process can achieve automated and intelligent decision support, improve the accuracy and timeliness of the evaluation, reduce the burden on medical staff, and improve the comfort of patients.

[0008] In a possible implementation of the first aspect, the step of dividing the wound area into N grid units specifically includes: establishing a polar coordinate system with the center point of the sensor array as the origin; dividing the wound area into K concentric annular areas, where K is an integer determined according to the total number of grid units N; and dividing fan-shaped grid units at equal angles in each of the annular areas.

[0009] In this way, the present application provides a specific grid division scheme suitable for circular or nearly circular wounds and sensor arrays arranged in polar coordinates, which can effectively map two-dimensional spatial data to discrete grid units, providing a basis for subsequent data processing and matrix construction.

[0010] In a possible implementation of the first aspect, the multi-dimensional physiological parameter data within the N grid cells are processed to obtain the characteristic value of each grid cell, specifically including: normalizing the multi-dimensional physiological parameter data within each grid cell; linearly combining the normalized temperature data, the humidity data, the pH value data and the pressure distribution data according to a predetermined weight ratio to generate a composite characteristic value of the grid cell.

[0011] In this way, the present application eliminates the dimensional differences between different types of data through normalization processing, and generates a single composite eigenvalue through weighted linear combination, which simplifies the subsequent matrix construction and analysis process, and can assign different weights to different physiological parameters according to their importance to wound healing.

[0012] In a possible implementation of the first aspect, the step of constructing an N-order feature matrix reflecting the feature relationship between the N grid units specifically includes: calculating the spatial distance between each of the N grid units to obtain a spatial distance matrix D; calculating the absolute value of the composite eigenvalue difference between each of the N grid units to obtain a feature difference matrix V_diff; constructing the N-order matrix A, wherein the (i, j)th element A(i, j) of the matrix A is calculated by substituting the (i, j)th element D(i, j) of the spatial distance matrix D and the (i, j)th element V_diff(i, j) of the feature difference matrix V_diff into a preset function f, and making the matrix A a symmetric matrix, the function f is f(d, dv) = exp(-d / k) * dv, where d represents the spatial distance between two grid units i and j, dv represents the absolute value of the difference between the composite eigenvalues ​​of the two grid units i and j, and k is a preset scaling factor.

[0013] Thus, this application constructs a symmetric matrix by calculating the spatial distance and eigenvalue differences between grid cells and combining it with a preset function f (for example, the product of a Gaussian kernel function and eigenvalue differences). The elements of this matrix can simultaneously reflect the spatial proximity between grid cells and the degree of difference in physiological parameters, allowing singular value decomposition to more effectively capture the non-uniformity and spatial correlation of the distribution of physiological parameters within the wound area, thereby extracting more diagnostically meaningful primary characteristic parameters.

[0014] In a possible implementation of the first aspect, the step of aligning the data of the main characteristic parameter with a preset standard healing curve and calculating the healing deviation index specifically includes: obtaining the theoretical reference value of each time node in the standard healing curve; calculating the dynamic time warping distance between the main characteristic parameter sequence and the theoretical reference value sequence; and using the dynamic time warping distance as the healing deviation index.

[0015] In this way, this application uses a dynamic time warping algorithm to measure the similarity or deviation between the trajectory of the currently monitored main characteristic parameters changing over time and the standard healing curve trajectory. This can effectively handle nonlinear deformations between time series and make data alignment more accurate. At the same time, the time warping distance is directly used as a healing deviation indicator. Its numerical value directly reflects the degree of deviation between the current healing trajectory and the standard trajectory. The larger the value, the greater the deviation.

[0016] In a possible implementation of the first aspect, the method also includes a step of correcting the healing deviation index, specifically, including: obtaining the environmental parameter correction coefficient λ, the environmental parameters including the ambient temperature change rate and the ambient humidity change rate; calculating the environmental parameter correction coefficient λ according to the formula λ=1-αΔT_env-βΔH_env, where α and β are preset weight factors, ΔT_env is the ambient temperature change rate, and ΔH_env is the ambient humidity change rate; multiplying the healing deviation index by the environmental parameter correction coefficient λ to obtain the corrected healing deviation index.

[0017] Drastic changes in ambient temperature and humidity may affect sensor readings or the wound microenvironment, resulting in errors in the calculated healing deviation index. This application calculates a correction factor λ based on the rate of change of environmental parameters and applies it to the healing deviation index. This can, to a certain extent, reduce the interference of environmental factors on the evaluation results and improve the robustness of the evaluation. When the environment changes significantly, λ may be less than 1, which reduces the corrected deviation index and reduces the risk of false positives.

[0018] In a possible implementation of the first aspect, the step of determining the wound healing stage based on the preset threshold interval in which the healing deviation index is located specifically includes: presetting four continuous threshold intervals [0,θ1), [θ1,θ2), [θ2,θ3), [θ3,+∞), the threshold intervals corresponding to the inflammatory stage, the proliferative stage, the mature stage and the abnormal stage respectively; judging which threshold interval the healing deviation index falls into; and generating an evaluation report containing the name and characteristic parameters corresponding to the wound healing stage based on the judgment result.

[0019] In this way, this application maps continuous healing deviation indicators to discrete healing stages by presetting continuous threshold intervals with clear boundaries, making the evaluation results clearer and more operational. The generated evaluation report can provide medical staff with quantitative wound status information and stage judgment to assist clinical decision-making.

[0020] In a possible implementation of the first aspect, the method also includes a parameter self-calibration step, specifically, including: establishing a historical case database to store standard physiological parameter ranges for different healing stages; detecting the currently monitored multi-dimensional physiological parameter data; when the currently monitored multi-dimensional physiological parameter data deviates from the standard range by more than a second threshold, starting a parameter self-calibration mode; in the parameter self-calibration mode, matching historical data in the historical case database according to a nearest neighbor algorithm to adjust the sensor reference value.

[0021] Long-term sensor use can cause drift, and individual differences can also lead to different baseline values ​​for physiological parameters. This application establishes a historical case database and initiates self-calibration when current data deviates from the standard range. Using similar historical data to calibrate sensor readings, this improves the system's adaptability and accuracy for different individuals and long-term monitoring.

[0022] In a possible implementation of the first aspect, the method further includes: calculating the Euclidean distance between the parameter vector composed of the currently monitored multi-dimensional physiological parameter data and the parameter vector composed of each historical data in the historical case database; selecting k historical samples with the smallest Euclidean distance to construct a calibration matrix; calculating the column average of the calibration matrix, and using the column average as the new sensor reference value.

[0023] In this way, the present application enhances the effectiveness and certainty of calibration by adopting Euclidean distance as a similarity metric, selecting the most similar historical data samples through the k-nearest neighbor algorithm, and calculating the average value of these samples as a new benchmark parameter.

[0024] In the second aspect, the present application provides an intelligent dressing system, comprising: a data acquisition module, configured to acquire multi-dimensional physiological parameter data output by the flexible multimodal sensor array; a data processing module, configured to divide the wound area into grid units, process the sensor data in each grid unit to obtain the eigenvalue of each grid unit, construct a characteristic matrix, and perform singular value decomposition on the characteristic matrix to extract main characteristic parameters; a healing evaluation module, configured to align the main characteristic parameters with a preset standard healing curve to calculate a healing deviation index, and determine the wound healing stage according to the healing deviation index; and a signal generation module, configured to generate a dressing change reminder signal according to the healing deviation index. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A flowchart of a management method provided for some embodiments of the present application. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments.

[0027] Hereinafter, 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 the technical features being referred to. Thus, a feature identified with "first," "second," etc., may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0028] In this application, unless otherwise specified or limited, the term "connection" should be understood broadly. For example, "connection" can mean fixed connection, detachable connection, or integration; it can mean direct connection or indirect connection through an intermediate medium. In addition, the term "electrical connection" can refer to the method of electrical connection that enables signal transmission.

[0029] As used herein, “about,” “substantially,” or “approximately” includes the stated value and reference values ​​that are within an acceptable range of deviation from the particular value, as determined by one of ordinary skill in the art taking into account the measurements in question and the errors associated with the measurement of the particular quantity (i.e., the limitations of the measurement method).

[0030] This application provides an intelligent dressing management method and intelligent dressing system based on a flexible multimodal sensor array to improve the shortcomings of existing wound management methods and intelligent dressings in terms of real-time monitoring, objective evaluation, and intelligent decision-making.

[0031] This application provides a smart patch system based on a flexible multimodal sensor array for wound management. The smart patch is designed to adhere to the wound surface or cover the wound area, collecting multi-dimensional physiological parameter data from the wound microenvironment. The smart patch includes a substantially flexible substrate, such as a flexible printed circuit (FPC) formed from a flexible polymer or a substantially flexible non-conductive mesh. The substrate supports or integrates the flexible multimodal sensor array.

[0032] The sensor array includes multiple spatially distributed sensing units, each equipped with a sensor for measuring different physiological parameters. These sensors include temperature sensors, humidity sensors, pH sensors, and pressure distribution sensors. In some embodiments, they may also include conductivity sensors, multispectral optical measurement sensors, colorimetric sensors, optical sensors, ultraviolet (UV) sensors, infrared (IR) sensors, and the like.

[0033] The flexible substrate and sensor array are capable of bending and deforming to closely conform to the wound surface, ensuring that the sensors can effectively collect data. The smart dressing can have a multi-layer structure, for example, including a wound contact layer that contacts the wound, an absorbent layer located above it, and a backing layer. The sensor array can be integrated into the wound contact layer, located on the wound contact layer, or encapsulated in another layer of the wound dressing. The sensor array is electrically connected to a control module via a flexible connection, and the control module is responsible for collecting sensor data and communicating with an external processing device.

[0034] like Figure 1 As shown, the method includes: S101. Acquire multi-dimensional physiological parameter data output by a flexible multimodal sensor array.

[0035] The flexible multimodal sensor array is integrated into a smart dressing and applied to the wound surface. The array includes multiple types of sensors capable of measuring different physiological parameters of the wound microenvironment, such as temperature, humidity, pH, and pressure distribution sensors. The system periodically collects data from the sensor array, such as every hour or as determined by clinical needs.

[0036] The multi-dimensional physiological parameter data includes temperature data, humidity data, pH data, and pressure distribution data at different locations in the wound area. For example, if the sensor array includes 100 spatially distributed sensing units, each acquisition will obtain temperature, humidity, pH, and pressure distribution readings at 100 locations.

[0037] Exemplarily, at a certain acquisition moment, the system obtains sensor readings of M discrete spatial points in the wound area (corresponding to the positions of M sensing units), forming a temperature reading set {T1, T2, ..., TM}, a humidity reading set {H1, H2, ..., HM}, a pH value reading set {pH1, pH2, ..., pHM}, and a pressure distribution reading set {P1, P2, ..., PM}.

[0038] S102: Divide the wound area into N grid units.

[0039] Before assessing the wound area, the wound-covered area needs to be spatially divided into a plurality of discrete grid cells. The number of grid cells, N, is a predetermined integer, and its size depends on the desired spatial resolution and computational resources; for example, N can be set to 100, 200, or more. This division discretizes the continuous or nearly continuous area covered by the sensor array into N non-overlapping sub-regions covering the entire area, with each grid cell corresponding to one or more sensor readings.

[0040] Exemplarily, the division step can be performed using a polar coordinate system. A polar coordinate system is established with the geometric center point of the sensor array or the approximate center point of the wound as the origin. The wound area is divided into K concentric annular areas. The integer K can be determined based on the total number of grid cells N, for example, K=round(log2(N)), which means that the number of annular areas is logarithmically related to the number of grid cells, so that a higher radial density can be used in the central area. Within each annular area, sectors are divided at equal angles to form grid cells. For example, if K=4, and the inner ring is divided into 10 sectors, the second ring is divided into 20 sectors, the third ring is divided into 30 sectors, and the outermost ring is 40 sectors, then the total number of grid cells N=10+20+30+40=100. Each grid cell i (i=1toN) corresponds to the position of one or more sensor units. The smart patch system can record the spatial position information of each grid cell, such as the polar coordinates (ri,θi) or Cartesian coordinates (xi,yi) of its center point.

[0041] S103: Process the multi-dimensional physiological parameter data in the N grid cells to obtain characteristic values ​​of each grid cell. For each grid cell, the sensor readings contained therein need to be integrated into one or a few representative characteristic values.

[0042] Exemplarily, the processing includes normalizing the multi-dimensional physiological parameter data within each grid unit and then performing linear combination. For example, for each grid unit i, the readings of all sensors falling into the unit are collected. If there are multiple sensors of the same type in a unit, their average or median can be taken as the representative value of the unit on the parameter. Different types of parameters are normalized, for example, using Z-score normalization or other min-max normalization methods to map readings such as temperature, humidity, pH value, pressure, etc. to a similar numerical range to eliminate dimensional effects. The normalized values ​​are Ti', Hi', pHi', Pi'. The normalized parameters are linearly combined according to a predetermined weight ratio to generate a composite eigenvalue v_i of the grid unit i. Exemplarily, the composite eigenvalue v_i can be calculated as: v_i=w_T*Ti'+w_H*Hi'+w_pH*pHi'+w_P*Pi' Here, w_T, w_H, w_pH, and w_P are preset weighting factors used to assess the relative importance of each physiological parameter in wound healing assessment. These weighting factors can be determined through clinical experience or machine learning methods, and their sum is usually 1 (for example, w_T, w_H, w_pH, and w_P can be 0.4, 0.1, 0.3, and 0.3, respectively, where w_T + w_H + w_pH + w_P = 1). This yields a vector V = [v_1, v_2, ..., v_N]ᵀ containing N composite eigenvalues, where v_i represents the comprehensive physiological state of the i-th grid cell.

[0043] S104: Construct an N-order feature matrix reflecting the feature relationship between the N grid cells.

[0044] For example, the N-order feature matrix A can be constructed as a symmetric matrix, whose element A(i,j) reflects the feature relationship between the i-th grid unit and the j-th grid unit. Specifically, the construction method includes: Obtain the spatial distance D(i,j) between any two grid cells i and j to obtain the spatial distance matrix D. The spatial distance can be the Euclidean distance between the center points of the two grid cells or other suitable distance metrics. Obtain the absolute value Δv(i,j)=|v_i-v_j| of the difference between the composite eigenvalues ​​of any two grid cells i and j to obtain the feature difference matrix V_diff. Construct the N-order feature matrix A, where the (i,j)th element A(i,j) of the matrix A can be calculated by substituting the (i,j)th element D(i,j) of the spatial distance matrix D and the (i,j)th element V_diff(i,j) of the feature difference matrix V_diff into the preset function f, and make the matrix A a symmetric matrix. The function f is: f(d,dv)=exp(-d / k)*dv Wherein, d represents the spatial distance between two grid units i and j, dv represents the absolute value of the difference between the composite eigenvalues ​​of the two grid units i and j, and k is a preset scaling factor.

[0045] Before constructing matrix A, the spatial distances between all grid cells must be calculated to form a spatial distance matrix D. d is the element D(i,j) in matrix D, representing the distance between the i-th and j-th grid cells. d quantifies the spatial proximity of two grid cells in the wound area. The closer the distance, the smaller the value of d, while the farther the distance, the larger the value of d.

[0046] After processing the multidimensional physiological parameter data within the grid cells and obtaining the composite eigenvalue vector V for each grid cell, the absolute value of the difference between the composite eigenvalues ​​v_i and v_j of any two grid cells i and j, |v_i - v_j|, is calculated. dv is the calculated |v_i - v_j|. dv quantifies the degree of similarity or difference in the comprehensive physiological parameter status of the two grid cells. Greater differences correspond to larger dv values, while smaller differences (greater similarity) correspond to smaller dv values.

[0047] k is a preset scaling factor or attenuation constant used to control the degree of influence of spatial distance d on the function value, especially the speed of spatial attenuation. In the exponential term exp(-d / k), d / k is a dimensionless value. If k is larger, then for the same distance d, the value of d / k is smaller, and the decay of exp(-d / k) is slower, indicating that the mutual influence or correlation between grid cells at a longer distance is still relatively strong. If k is smaller, then for the same distance d, the value of d / k is larger, and the decay of exp(-d / k) is faster, indicating that only very close grid cells have significant mutual influence or correlation.

[0048] In this way, the obtained A(i,j) value comprehensively reflects the spatial proximity between grid cells i and j and the degree of difference in physiological parameters. The matrix A constructed based on A(i,j), whose elements A(i,j) highlight those pairs of regions within the wound area that are spatially adjacent but have significantly different physiological states. Singular value decomposition of this matrix, especially extracting the maximum singular value, can capture the main spatial non-uniformity and spatial correlation patterns of the distribution of physiological parameters in the wound area. As the wound heals, inflammation decreases, and the tissue tends to be homogenized, this spatial non-uniformity and difference will theoretically decrease, which is reflected in the overall decrease in the element values ​​of matrix A, resulting in a decrease in the maximum singular value σ_max. Therefore, the maximum singular value σ_max is the key main characteristic parameter that characterizes the overall healing state or change of the wound.

[0049] This function f ensures that closer grid cells with more distinct characteristics contribute more to the matrix elements, while their contribution decreases with greater distance. The diagonal elements A(i,i) of matrix A are typically defined as 0 or a value related to the eigenvalue of cell i, for example, f(0,0)=0 or A(i,i)=v_i. Here, according to the function definition, D(i,i)=0, Δv(i,i)=0, then A(i,i)=f(0,0)=exp(0)*0=0.

[0050] Therefore, the constructed matrix A is a symmetric matrix whose off-diagonal elements quantify the feature differences and spatial correlations between grid cells at different spatial locations.

[0051] For example, assume that there are 3 grid cells, and their composite eigenvalue vector V = [v1, v2, v3] = [0.5, 0.8, 0.3]. Assume that the spatial distance matrix D (Euclidean distance) is: D=[[0,1.0,1.5], [1.0,0,2.0], [1.5,2.0,0]] The feature difference matrix V_diff=|Vᵀ-V| is: V_diff=[[0,|0.5-0.8|,|0.5-0.3|], [|0.8-0.5|,0,|0.8-0.5|], [|0.3-0.5|,|0.3-0.8|,0]] =[[0,0.3,0.2], [0.3,0,0.5], [0.2,0.5,0]] Take k=1.0 and calculate the characteristic matrix A according to A(i,j)=exp(-D(i,j) / k)*V_diff(i,j): A(1,2)=exp(-1.0 / 1.0)*0.3≈0.368*0.3≈0.110 A(1,3)=exp(-1.5 / 1.0)*0.2≈0.223*0.2≈0.045 A(2,3)=exp(-2.0 / 1.0)*0.5≈0.135*0.5≈0.068 A=[[0,0.110,0.045], [0.110,0,0.068], [0.045,0.068,0]] In this way, a characteristic difference symmetric matrix reflecting the spatial attenuation between grid cells can be obtained.

[0052] S105 , performing singular value decomposition on the N-order feature matrix, and extracting the maximum singular value as the main feature parameter.

[0053] Perform singular value decomposition (SVD) on the N-order feature matrix A. SVD is a matrix decomposition method that can decompose any matrix A into A = UΣVᵀ, where U and V are orthogonal matrices, and Σ is a diagonal matrix whose diagonal elements are called singular values, typically arranged in descending order: σ1 ≥ σ2 ≥ ... ≥ σn ≥ 0. The largest singular value σ1 captures the most dominant variance or energy in matrix A, that is, the main pattern of variation in the data. In this application, matrix A characterizes the spatial differences and correlations of physiological parameters within the wound area. Therefore, the largest singular value σ1 can represent the overall heterogeneity and intensity of variation in the spatial distribution of physiological parameters under the current wound state and is used as the main feature parameter σ_max. As the wound heals, inflammation decreases, and the tissue becomes more uniform, this spatial variation decreases, so the value of the main feature parameter σ_max also changes in a specific trend over time.

[0054] For example, the singular value decomposition of the matrix A is performed as A=UΣ , extract the maximum singular value σ1 from the diagonal matrix Σ. For example, if the calculated singular values ​​are [0.187, 0.042, 0.000], then the maximum singular value σ_max = 0.187, which is used as the primary characteristic parameter at the current moment. Repeating the above steps multiple times can obtain a time-varying primary characteristic parameter sequence {σ_max(t1),σ_max(t2),σ_max(t3),...}. This primary characteristic parameter sequence constitutes the primary characteristic parameter trajectory of the current wound healing process.

[0055] S106: Align the main characteristic parameters with a preset standard healing curve, and calculate a healing deviation index.

[0056] Wound healing is a dynamic process, and the healing speed and specific trajectory may vary between individuals and different types of wounds. In order to objectively assess the current healing status of the wound, it is necessary to compare the time trajectory of its main characteristic parameters with a curve representing the standard healing process. The standard healing curve is a theoretical trajectory of the change of a preset main characteristic parameter over time, which can be obtained by averaging or modeling a large amount of historical case data. Since the actual main characteristic parameter sequence collected and the standard curve may have a nonlinear correspondence on the time axis (for example, actual healing is slower or faster than the standard), data alignment is required.

[0057] Exemplarily, the data alignment employs a dynamic time warping (DTW) algorithm. A theoretical reference value sequence Y = [y1, y2, ..., ym] for each time node in the standard healing curve is obtained. A primary characteristic parameter sequence X = [σ_max(t1), σ_max(t2), ..., σ_max(tn)] is obtained. The dynamic time warping distance DTW(X, Y) between sequences X and Y is calculated.

[0058] The DTW algorithm can calculate the distance between two time series by finding the optimal nonlinear time bending path between them. Even if they are stretched or offset on the time axis, a reasonable similarity measure can be calculated. The calculation of the DTW distance is deterministic. Two sequences are input and a unique distance value is output. In this application, the calculated dynamic time warping distance is used as the healing deviation index. The larger the numerical value of the healing deviation index, the greater the deviation of the main characteristic parameter trajectory of the current wound from the standard healing curve, and the more likely the healing state is abnormal or slower than expected.

[0059] For example, suppose the main characteristic parameter sequence monitored over a certain period of time is X = [0.2, 0.18, 0.15, 0.12, 0.10], and the corresponding standard healing curve theoretical value is Y = [0.25, 0.20, 0.15, 0.10, 0.05]. Calculating DTW(X, Y) yields a distance value, for example, 0.15. The healing deviation index at the current calculation moment is 0.15.

[0060] S107 : Determine the wound healing stage according to the preset threshold range in which the healing deviation indicator is located.

[0061] In order to convert the continuous healing deviation index into a clinically meaningful healing stage judgment, the system presets multiple continuous threshold intervals.

[0062] For example, four consecutive threshold intervals can be preset: [0, θ1), [θ1, θ2), [θ2, θ3), [θ3, +∞). The thresholds θ1, θ2, and θ3 can be preset, clinically meaningful values. For example, θ1 = 0.05, θ2 = 0.15, and θ3 = 0.30 can be set. These threshold intervals correspond to different stages or states of wound healing. For example: [0, θ1) corresponds to the mature / stable period, indicating that the healing trajectory is very close to the standard curve and the healing progress is smooth or complete.

[0063] [θ1,θ2) corresponds to the proliferative phase, indicating that the healing trajectory is within the standard range and is in the active tissue repair stage.

[0064] [θ2,θ3) corresponds to the inflammatory phase / protracted phase, indicating that the healing trajectory deviates from the standard curve, which may indicate mild inflammation or slow healing.

[0065] [θ3,+∞) corresponds to the abnormal period, indicating that the healing trajectory deviates seriously from the standard curve, and there may be serious problems such as infection, hematoma, and abnormal dressing function.

[0066] The system determines which threshold interval the currently calculated healing deviation index falls within, thereby determining the current wound healing stage. Based on this determination, an assessment report is generated containing the wound healing stage name (e.g., "proliferative phase," "abnormal phase") and the current healing deviation index value. This report can be displayed on a mobile app or medical terminal associated with the smart patch.

[0067] For example, if the calculated healing deviation index is 0.18, which falls within the interval [0.15, 0.30), the system determines that the wound is in the "inflammatory stage / protracted stage" and generates an evaluation report containing information such as "current stage: inflammatory stage / protracted stage, deviation index: 0.18".

[0068] S108. When the healing deviation index exceeds a first threshold, generate a dressing change prompt signal.

[0069] In addition to determining the stage of healing, the system can also trigger a dressing change or other intervention based on the degree of abnormality in the healing state. The healing deviation index quantifies the overall deviation between the primary characteristic parameter sequence of the current wound healing process and the standard healing curve sequence. When this overall deviation reaches a certain level (i.e., a large value for the healing deviation index), it indicates that there may be abnormalities or risks in wound healing, requiring attention. Therefore, the system presets a first threshold (denoted as Y1) to determine when to generate a dressing change reminder signal.

[0070] For example, the first threshold value Y1 is set to a predetermined value, such as 0.20. At the end of the current calculation cycle, the system obtains the healing deviation index value and determines whether the healing deviation index value is greater than the first threshold value Y1. If the healing deviation index is greater than Y1, the wound healing state is determined to be abnormal, and a dressing change prompt signal is generated. This signal can be sent to the mobile application or medical terminal associated with the smart dressing in the form of a pop-up window, a push notification message, an audible alarm, etc., prompting medical staff or patients to check or change the dressing.

[0071] For example, if the calculated healing deviation index is 0.25, since 0.25>0.20, the system determines that the healing deviation index exceeds the threshold and generates a dressing change prompt signal.

[0072] In some embodiments, the method further includes S109, correcting the healing deviation index. Exemplarily, S109 includes: After calculating the initial healing deviation index, the healing deviation index can be corrected based on environmental parameters, including the rate of change of ambient temperature (ΔT_env) and ambient humidity (ΔH_env) outside the patch, which can be obtained through additional environmental sensors on the patch or external devices.

[0073] Exemplarily, the ambient temperature change rate ΔT_env and the ambient humidity change rate ΔH_env in the current time period are obtained. According to the preset weight factors α and β, the environmental parameter correction coefficient λ=1-αΔT_env-βΔH_env is calculated. The weight factors α and β are positive values ​​determined by experiments or experience, which reflect the degree of influence of changes in ambient temperature and humidity on sensor readings or wound healing assessments. Among them, the α quantifies the degree of influence of a unit degree Celsius temperature change on the correction coefficient λ, and the selection of its value and unit makes the αΔT_env term a dimensionless value. The β quantifies the degree of influence of a unit percentage humidity change on the correction coefficient λ, and the selection of its value and unit makes the βΔH_env term a dimensionless value. For example, α can be set to 0.01 and β can be set to 0.02. If the ambient temperature rises by 5°C (ΔT_env = 5) and the humidity decreases by 10% (ΔH_env = -10) over a period of time, then λ = 1-0.015-0.02(-10) = 1-0.05+0.20 = 1.15. If the ambient temperature falls by 5°C (ΔT_env = -5) and the humidity increases by 10% (ΔH_env = 10), then λ = 1-0.01*(-5)-0.02*10 = 1+0.05-0.20 = 0.85.

[0074] The calculated correction coefficient λ is multiplied by the healing deviation index to obtain a corrected healing deviation index = healing deviation index*λ.

[0075] For example, if the calculated initial healing deviation index is 0.15 and the current environmental correction factor λ = 0.85, the corrected healing deviation index is 0.15 * 0.85 = 0.1275. This corrected index is used for stage judgment and signal generation. This allows the calculated deviation to be appropriately lowered when the environment fluctuates drastically (λ may be less than 1), thereby reducing misjudgments caused by environmental noise. In contrast, when the environment is stable (λ is close to 1), the correction effect is minimal.

[0076] In some embodiments, the method may also pre-establish a historical case database. This database stores a large amount of multi-dimensional physiological parameter monitoring data from different wound types and different patients during standard healing processes (or different healing stages). Each historical data record may include a multi-dimensional physiological parameter data vector (e.g., raw temperature, humidity, pH, and pressure readings for all grid cells) for a specific time point or time period, along with corresponding healing stage or clinical assessment information. The database may store standard physiological parameter ranges or representative data for different healing stages (e.g., inflammatory, proliferative, and maturation stages).

[0077] The system periodically detects a parameter vector composed of the currently monitored multi-dimensional physiological parameter data. When the parameter vector composed of the currently monitored multi-dimensional physiological parameter data deviates from the standard physiological parameter range corresponding to the healing stage in the historical database by more than a second threshold value (for example, the current average reading of all sensors or the reading of a key parameter exceeds ±10% (second threshold value = 10%) of the standard range), the system initiates a parameter self-calibration mode.

[0078] In the parameter self-calibration mode, the system searches the historical case database for historical data samples that are similar to the current preliminary judgment of the healing stage, and searches for historical data records that are most similar to the parameter vector composed of the currently monitored multi-dimensional physiological parameter data.

[0079] Exemplarily, the nearest neighbor algorithm is used for matching. Calculate the Euclidean distance between the parameter vector consisting of the currently monitored multi-dimensional physiological parameter data (for example, a vector consisting of the normalized original sensor readings of all grid cells) and the parameter vector corresponding to each historical data record in the historical case database. Select k historical samples with the smallest Euclidean distance. The k is a preset integer, for example, k=5 or 10. These k historical samples constitute a calibration data set, which can be represented as a calibration matrix, each column representing a physiological parameter type (for example, normalized temperature, humidity, pH, pressure), and each row representing the corresponding reading of a historical sample. Calculate the column average of the calibration matrix to obtain a parameter vector, each element of which is the average value of the corresponding physiological parameter in the k most similar historical samples. Use the column average vector as the new sensor reference value.

[0080] For example, if the original pH sensor's baseline value is theoretically 7.4, but the new pH baseline value calculated through calibration is 7.38, then subsequent pH readings will be relatively calibrated based on 7.38 as the zero point (for example, subtract 0.02 from all pH readings). In this way, calibration based on historical similar data can effectively correct for overall deviations caused by individual sensor differences or drift.

[0081] The intelligent patch system of the present application may further include a data acquisition module, a data processing module, a healing assessment module, and a signal generation module.

[0082] The data acquisition module, comprising a flexible multimodal sensor array and supporting data acquisition circuitry and communication interfaces, is responsible for collecting and transmitting raw physiological parameter data. The data processing module, which can be a microcontroller, embedded system, or external computing device communicating with the patch, receives raw data and performs computationally intensive tasks such as gridding, data processing within the grid cells (normalization, linear combination), feature matrix construction, and singular value decomposition, outputting a sequence of primary characteristic parameters. The healing assessment module, which can be implemented on an external computing device communicating with the patch (such as a smartphone, tablet, or cloud server), receives the sequence of primary characteristic parameters and performs functions such as data alignment with a standard healing curve (DTW), healing deviation index calculation, threshold determination, healing stage determination, and assessment report generation. Optionally, this module also implements environmental parameter correction and parameter self-calibration logic, including accessing a historical case database. The signal generation module, integrated with or in communication with the healing assessment module, generates and transmits a dressing change reminder signal to a user interface or other system based on the assessment results (e.g., if the healing deviation index exceeds a threshold).

[0083] Those skilled in the art will appreciate that the functions of the above modules can be implemented by hardware, software, or a combination thereof. The connections and data flows between the modules are performed in the order of the steps described in this method.

[0084] Although the present application has been described in conjunction with preferred embodiments, it will be understood by those skilled in the art that various changes or modifications may be made to the aforementioned embodiments without departing from the principles and spirit of the present application, and such changes and modifications also fall within the scope of protection of the present application. For example, the method of grid division is not limited to the polar coordinate system, and a rectangular coordinate system or other division methods adapted to the shape of the wound may also be used. The method of generating composite eigenvalues ​​is not limited to linear combinations, and other feature fusion technologies may also be used. The method of constructing the feature matrix may also adopt other matrix forms that can reflect spatial relationships or feature distribution patterns. In addition to DTW, the method of data alignment may also adopt other time series similarity measurement methods. The division of the threshold interval may be adjusted or the number of stages may be increased according to specific clinical needs.

[0085] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of modules or units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0086] The units described as separate components may or may not be physically separate, and the components shown as units may be one physical unit or multiple physical units, that is, they may be located in one place or distributed in multiple places. Some or all of the units shown in this embodiment may be selected according to actual needs to achieve the purpose of this embodiment.

[0087] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated units may be implemented in the form of hardware.

[0088] The above content is only a specific embodiment of this application, but the scope of protection of this application is not limited to this. Any changes or replacements within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. An intelligent patch management method based on a flexible multimodal sensor array, characterized in that: include: Acquire multi-dimensional physiological parameter data output by flexible multimodal sensor array; Dividing the wound area into N grid units, where N is a preset integer; Processing the multi-dimensional physiological parameter data in the N grid cells to obtain a characteristic value of each grid cell; Constructing an N-order feature matrix reflecting the feature relationship between the N grid cells; Performing singular value decomposition on the N-order characteristic matrix and extracting the maximum singular value as the main characteristic parameter; Aligning the main characteristic parameters with a preset standard healing curve to calculate a healing deviation index; determining the wound healing stage according to a preset threshold range in which the healing deviation index is located; as well as When the healing deviation indicator exceeds a first threshold, a dressing change reminder signal is generated.

2. The method according to claim 1, characterized in that The dividing the wound area into N grid units includes: Establishing a polar coordinate system with the center point of the sensor array as the origin; Dividing the wound area into K concentric annular areas, where K is an integer determined according to the total number N of grid cells; In each of the annular regions, fan-shaped grid units are divided at equal angles.

3. The method according to claim 1, characterized in that The processing of the multi-dimensional physiological parameter data in the N grid cells to obtain a characteristic value of each grid cell includes: Normalizing the multi-dimensional physiological parameter data within each grid unit, the multi-dimensional physiological parameter data including temperature data, humidity data, pH value data, and pressure distribution data; The normalized temperature data, the humidity data, the pH value data, and the pressure distribution data are linearly combined according to a predetermined weight ratio to generate a composite eigenvalue of the grid unit.

4. The method according to claim 3, characterized in that The constructing of an N-order feature matrix reflecting the feature relationship between the N grid cells includes: Obtaining a spatial distance matrix according to the spatial distances between each of the N grid cells; Obtaining a feature difference matrix according to the absolute values ​​of the composite eigenvalue differences between any two of the N grid cells; Construct the N-order matrix, wherein the (i, j)th element A(i, j) of the N-order matrix is ​​obtained based on the (i, j)th element D(i, j) of the spatial distance matrix and the (i, j)th element of the feature difference matrix, and the matrix A is a symmetric matrix.

5. The method according to claim 1, wherein The data alignment of the main characteristic parameter with the preset standard healing curve and calculation of the healing deviation index include: Obtaining theoretical reference values ​​for each time node in the standard healing curve; Calculating the dynamic time warping distance between the main characteristic parameter sequence and the theoretical reference value sequence; The dynamic time warping distance is used as the healing deviation indicator.

6. The method according to claim 5, characterized in that The method further includes correcting the healing deviation index, including: Obtaining the environmental parameter correction coefficient, wherein the environmental parameters include an ambient temperature change rate and an ambient humidity change rate; Obtaining the environmental parameter correction coefficient according to the ambient temperature change rate and the ambient humidity change rate; A corrected healing deviation index is obtained according to the healing deviation index and the environmental parameter correction coefficient.

7. The method according to claim 1, characterized in that Determining the wound healing stage according to the preset threshold range of the healing deviation indicator includes: Four continuous threshold intervals [0, θ1), [θ1, θ2), [θ2, θ3), and [θ3, +∞) are preset, and the threshold intervals correspond to the inflammatory stage, the proliferative stage, the mature stage, and the abnormal stage respectively; Determining the threshold range within which the healing deviation indicator falls; An evaluation report including the name and characteristic parameters of the corresponding wound healing stage is generated according to the judgment result.

8. The method according to claim 1, characterized in that The method further comprises: Establish a historical case database to store standard physiological parameter ranges at different healing stages; Detecting the multi-dimensional physiological parameter data currently monitored; When the currently monitored multi-dimensional physiological parameter data deviates from the standard range and exceeds a second threshold, starting a parameter self-calibration mode; In the parameter self-calibration mode, the sensor reference value is adjusted by matching historical data in the historical case database using a nearest neighbor algorithm.

9. The method according to claim 8, characterized in that The parameter self-calibration mode includes: Calculating the Euclidean distance between the parameter vector composed of the currently monitored multi-dimensional physiological parameter data and the parameter vector composed of each historical data in the historical case database; Selecting k historical samples with the smallest Euclidean distance to construct a calibration matrix; Calculate the column average of the calibration matrix and use the column average as the new sensor reference value.

10. An intelligent patch system, characterized in that: include: a data acquisition module configured to acquire multi-dimensional physiological parameter data output by the flexible multimodal sensor array; a data processing module configured to divide the wound area into grid cells, process sensor data within each grid cell to obtain an eigenvalue of each grid cell, construct a feature matrix, and perform singular value decomposition on the feature matrix to extract main feature parameters; a healing assessment module configured to align the main characteristic parameter with a preset standard healing curve to calculate a healing deviation index, and determine the wound healing stage according to the healing deviation index; as well as A signal generating module is configured to generate a dressing change prompt signal according to the healing deviation index.

Images