A method and system for determining wound infection care information

Abstract

The application provides a method and system for determining nursing information of a wound infection, and relates to the technical field of wound infection nursing. The method comprises the following steps: acquiring interface stress distribution evolution data of a pressure site; determining a plurality of pressure injury susceptible areas and a cumulative load injury prediction map of each pressure injury susceptible area based on the interface stress distribution evolution data of the pressure site; generating a deep injury evolution panoramic map of each pressure injury susceptible area based on tissue physical property data of each pathological feature sampling site of the pressure injury susceptible area and tissue physical property data of each advanced pathological feature sampling site; and determining a bed sore necrosis danger diffusion area based on tissue physical property data of a plurality of pathological feature sampling sites in a concealed deep tissue injury area and tissue physical property data of a plurality of advanced pathological feature sampling sites. The method can accurately identify a bed sore tissue injury spreading trend and determine a necrosis danger diffusion area.

Description

Technical Field

[0001] This invention relates to the field of wound infection care technology, and specifically to a method and system for determining wound infection care information. Background Technology

[0002] The prevention and care of pressure ulcers are crucial aspects of clinical nursing and rehabilitation, and the accurate acquisition of nursing information directly affects the wound healing process. Traditional pressure ulcer care and wound assessment mainly rely on comprehensive analysis by medical staff through visual observation combined with basic testing methods. However, this method has significant limitations in practical application. On the one hand, the manual analysis process is highly subjective, and the assessment results often depend heavily on the individual clinical experience of medical staff, leading to potential cognitive biases among different assessors regarding the same wound. On the other hand, the evolution of pressure ulcers involves complex data from multiple dimensions, such as interfacial stress and tissue physical properties, making it difficult for humans to deeply mine and correlate this massive amount of heterogeneous pathological information in a short period of time. This limitation not only limits the accuracy of identifying occult deep tissue injuries but also lacks objective quantitative standards, easily leading to missed diagnoses or misjudgments.

[0003] Therefore, accurately identifying the trend of tissue damage spread in pressure ulcers and determining the areas at risk of necrosis is an urgent problem to be solved. Summary of the Invention

[0004] The main technical problem addressed by this invention is how to accurately identify the spread trend of pressure ulcer tissue damage and determine the area at risk of necrosis.

[0005] According to a first aspect, the present invention provides a method for determining nursing information for wound infection, comprising: acquiring data on the evolution of interfacial stress distribution at pressure sites; determining multiple pressure injury susceptibility zones and a cumulative load injury prediction map for each pressure injury susceptibility zone based on the data on the evolution of interfacial stress distribution at pressure sites; acquiring surface morphology visualization images of multiple pressure injury susceptibility zones; determining multiple pathological feature sampling sites for each pressure injury susceptibility zone based on the surface morphology visualization images of the multiple pressure injury susceptibility zones, and acquiring tissue physical property data for each pathological feature sampling site; determining each pressure injury based on the tissue physical property data of each pathological feature sampling site and the cumulative load injury prediction map for each pressure injury susceptibility zone. Multiple advanced pathological feature sampling sites are used in the susceptible area, and the tissue physical characteristics data of each advanced pathological feature sampling site are obtained. Based on the tissue physical characteristics data of each pathological feature sampling site in each pressure injury susceptible area and the tissue physical characteristics data of each advanced pathological feature sampling site, a panoramic view of the deep injury evolution of each pressure injury susceptible area is generated. Based on the cumulative load injury prediction map of each pressure injury susceptible area and the panoramic view of the deep injury evolution of each pressure injury susceptible area, the hidden deep tissue injury area is identified. Based on the tissue physical characteristics data of multiple pathological feature sampling sites and multiple advanced pathological feature sampling sites within the hidden deep tissue injury area, the risk of pressure ulcer necrosis diffusion area is identified.

[0006] In one possible implementation, determining the pressure ulcer necrosis risk zone based on tissue physical property data from multiple pathological feature sampling sites and multiple advanced pathological feature sampling sites within a occult deep tissue injury area includes: constructing a wound knowledge graph, which includes multiple pathological feature sampling nodes and multiple advanced pathological feature sampling nodes. The node features of the pathological feature sampling nodes are tissue physical property data from the pathological feature sampling sites, and the node features of the advanced pathological feature sampling nodes are tissue physical property data from the advanced pathological feature sampling sites; processing the wound knowledge graph based on a graph neural network to determine multiple prognostic verification sites within each occult deep tissue injury area and obtaining the histopathological information of each prognostic verification site; and determining the pressure ulcer necrosis risk zone based on the histopathological information of each prognostic verification site.

[0007] In one possible implementation, determining the pressure ulcer necrosis risk spread area based on the histopathological information of each prognostic verification site includes: clustering the histopathological information of each prognostic verification site to obtain multiple clusters; determining multiple necrosis spread paths based on the multiple clusters; and determining the pressure ulcer necrosis risk spread area based on the multiple necrosis spread paths.

[0008] In one possible implementation, the input to the graph neural network is the wound knowledge graph, and the output of the graph neural network is multiple prognostic verification sites within each occult deep tissue injury area.

[0009] According to a second aspect, the present invention provides a system for determining nursing information on wound infection, comprising:

[0010] The data acquisition module is used to acquire data on the evolution of interfacial stress distribution in the compressed area;

[0011] The vulnerable area determination module is used to determine multiple pressure damage vulnerable areas and a cumulative load damage prediction map for each pressure damage vulnerable area based on the evolution data of the interface stress distribution of the pressure-bearing part.

[0012] The image acquisition module is used to acquire visual images of the surface morphology of multiple pressure-damage-prone areas;

[0013] The sampling analysis module is used to determine multiple pathological feature sampling sites for each pressure injury susceptible area based on the surface morphology visualization images of the multiple pressure injury susceptible areas, and to acquire tissue physical characteristic data for each pathological feature sampling site.

[0014] The advanced sampling module is used to determine multiple advanced pathological feature sampling sites for each pressure injury susceptible area based on the tissue physical characteristic data of each pathological feature sampling site and the cumulative load damage prediction map of each pressure injury susceptible area, and to acquire the tissue physical characteristic data of each advanced pathological feature sampling site.

[0015] The evolution generation module is used to generate a panoramic view of the deep injury evolution of each pressure injury susceptibility area based on the tissue physical property data of each pathological feature sampling site of each pressure injury susceptibility area and the tissue physical property data of each advanced pathological feature sampling site.

[0016] The damage area identification module is used to identify hidden deep tissue damage areas based on the cumulative load damage prediction map of each pressure injury susceptibility area and the panoramic map of deep damage evolution of each pressure injury susceptibility area.

[0017] The diffusion prediction module is used to determine the risk zone for pressure ulcer necrosis diffusion based on the tissue physical characteristics data of multiple pathological feature sampling sites within the occult deep tissue injury area and the tissue physical characteristics data of multiple advanced pathological feature sampling sites.

[0018] In one possible implementation, the diffusion prediction module is further configured to: construct a wound knowledge graph, the wound knowledge graph including multiple pathological feature sampling nodes and multiple advanced pathological feature sampling nodes, wherein the node features of the pathological feature sampling nodes are tissue physical characteristic data of the pathological feature sampling sites, and the node features of the advanced pathological feature sampling nodes are tissue physical characteristic data of the advanced pathological feature sampling sites; process the wound knowledge graph based on a graph neural network to determine multiple prognostic verification sites within each occult deep tissue injury area, and obtain the histopathological information of each prognostic verification site; and determine the pressure ulcer necrosis risk diffusion area based on the histopathological information of each prognostic verification site.

[0019] In one possible implementation, determining the pressure ulcer necrosis risk spread area based on the histopathological information of each prognostic verification site includes: clustering the histopathological information of each prognostic verification site to obtain multiple clusters; determining multiple necrosis spread paths based on the multiple clusters; and determining the pressure ulcer necrosis risk spread area based on the multiple necrosis spread paths.

[0020] In one possible implementation, the input to the graph neural network is the wound knowledge graph, and the output of the graph neural network is multiple prognostic verification sites within each occult deep tissue injury area.

[0021] According to a third aspect, embodiments of the present invention provide an electronic device, including: a processor; a memory; and a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor to implement the method as described above, the method including: acquiring interfacial stress distribution evolution data of a pressure-bearing site; determining multiple pressure injury susceptibility zones and a cumulative load injury prediction map of each pressure injury susceptibility zone based on the interfacial stress distribution evolution data of the pressure-bearing site; acquiring surface morphology visualization images of the multiple pressure injury susceptibility zones; determining multiple pathological feature sampling sites of each pressure injury susceptibility zone based on the surface morphology visualization images of the multiple pressure injury susceptibility zones, and acquiring tissue physical property data of each pathological feature sampling site; based on the tissue physical property data of each pathological feature sampling site, the... The cumulative load damage prediction map of each pressure injury susceptible area is used to identify multiple advanced pathological feature sampling sites for each pressure injury susceptible area, and the tissue physical characteristic data of each advanced pathological feature sampling site are obtained. Based on the tissue physical characteristic data of each pathological feature sampling site and each advanced pathological feature sampling site in each pressure injury susceptible area, a panoramic view of the deep injury evolution of each pressure injury susceptible area is generated. Based on the cumulative load damage prediction map and the panoramic view of the deep injury evolution of each pressure injury susceptible area, the hidden deep tissue injury area is identified. Based on the tissue physical characteristic data of multiple pathological feature sampling sites and multiple advanced pathological feature sampling sites within the hidden deep tissue injury area, the risk zone for pressure ulcer necrosis spread is identified.

[0022] According to a fourth aspect, this embodiment provides a computer-readable storage medium storing a computer program thereon. When executed by a processor, the program implements the aforementioned method for determining nursing information on wound infection. The method includes: acquiring interfacial stress distribution evolution data of pressure sites; determining multiple pressure injury susceptibility zones and a cumulative load injury prediction map of each pressure injury susceptibility zone based on the interfacial stress distribution evolution data of the pressure sites; acquiring surface morphology visualization images of multiple pressure injury susceptibility zones; determining multiple pathological feature sampling sites for each pressure injury susceptibility zone based on the surface morphology visualization images of the multiple pressure injury susceptibility zones, and acquiring tissue physical property data of each pathological feature sampling site; and based on the tissue physical property data of each pathological feature sampling site and the cumulative load injury prediction map of each pressure injury susceptibility zone... The cumulative load damage prediction map of the susceptible area identifies multiple advanced pathological feature sampling sites for each pressure injury susceptible area, and acquires the tissue physical characteristic data of each advanced pathological feature sampling site; based on the tissue physical characteristic data of each pathological feature sampling site and each advanced pathological feature sampling site in each pressure injury susceptible area, a panoramic view of the deep injury evolution of each pressure injury susceptible area is generated; based on the cumulative load damage prediction map and the panoramic view of the deep injury evolution of each pressure injury susceptible area, the occult deep tissue injury area is identified; based on the tissue physical characteristic data of multiple pathological feature sampling sites and multiple advanced pathological feature sampling sites within the occult deep tissue injury area, the risk zone for pressure ulcer necrosis diffusion is identified.

[0023] This invention provides a method and system for determining nursing information related to wound infection. The method includes acquiring data on the evolution of interfacial stress distribution at pressure sites; determining multiple pressure injury susceptibility zones and a cumulative load damage prediction map for each pressure injury susceptibility zone based on the data on the evolution of interfacial stress distribution at pressure sites; acquiring surface morphology visualization images of the multiple pressure injury susceptibility zones; determining multiple pathological feature sampling sites for each pressure injury susceptibility zone based on the surface morphology visualization images of the multiple pressure injury susceptibility zones, and acquiring tissue physical property data for each pathological feature sampling site; and determining multiple advanced pathological feature sampling sites for each pressure injury susceptibility zone based on the tissue physical property data of each pathological feature sampling site and the cumulative load damage prediction map of each pressure injury susceptibility zone. The method acquires tissue physical characteristic data for each advanced pathological feature sampling site; based on the tissue physical characteristic data of each pathological feature sampling site in each pressure injury susceptible area and the tissue physical characteristic data of each advanced pathological feature sampling site, a panoramic view of the deep injury evolution in each pressure injury susceptible area is generated; based on the cumulative load injury prediction map and the panoramic view of the deep injury evolution in each pressure injury susceptible area, the hidden deep tissue injury area is identified; based on the tissue physical characteristic data of multiple pathological feature sampling sites and multiple advanced pathological feature sampling sites within the hidden deep tissue injury area, the risk of pressure ulcer necrosis spread is identified. This method can accurately identify the spread trend of pressure ulcer tissue damage and determine the risk of necrosis spread area. Attached Figure Description

[0024] Figure 1 A flowchart illustrating a method for determining nursing information for wound infection according to an embodiment of the present invention;

[0025] Figure 2 A schematic diagram of a pressure-bearing part provided in an embodiment of the present invention;

[0026] Figure 3 This is a schematic diagram of a process for determining the risk zone of pressure ulcer necrosis in an embodiment of the present invention;

[0027] Figure 4 A schematic diagram of a process for determining the risk zone of pressure ulcer necrosis based on histopathological information of each prognostic verification site, provided for an embodiment of the present invention;

[0028] Figure 5 This is a schematic diagram of a wound infection nursing information determination system provided in an embodiment of the present invention. Detailed Implementation

[0029] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings. Similar elements in different embodiments are referred to by associated similar element reference numerals. In the following embodiments, many details are described to facilitate a better understanding of the invention. However, those skilled in the art will readily recognize that some features may be omitted in different situations, or may be replaced by other elements, materials, or methods. In some cases, certain operations related to the present invention are not shown or described in the specification. This is to avoid obscuring the core parts of the invention with excessive description. For those skilled in the art, detailed description of these related operations is not necessary; they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.

[0030] In this embodiment of the invention, the following are provided: Figure 1 The method for determining nursing information of wound infection, as shown, includes steps S1 to S8:

[0031] Step S1: Obtain the evolution data of interfacial stress distribution at the pressure-bearing part.

[0032] The data on the evolution of interfacial stress distribution in the pressure-bearing area is a sequence of data reflecting the dynamic change of the pressure state between the surface of the pressure-bearing area of ​​the human body and the contact surface of the supporting medium over time. Figure 2 This is a schematic diagram of a pressure-bearing part provided in an embodiment of the present invention.

[0033] The data on the evolution of interfacial stress distribution in the compressed area includes the pressure intensity values, pressure gradient directions, and displacement of the compressed area at different time points.

[0034] In some embodiments, data can be collected by an array of flexible pressure sensors laid on the mattress surface, and the physical deformation sensed by the sensors can be converted into electrical signals to obtain data on the evolution of interfacial stress distribution at the pressure point.

[0035] Step S2: Based on the evolution data of the interfacial stress distribution of the pressure-bearing part, determine multiple pressure damage susceptibility zones and a cumulative load damage prediction map for each pressure damage susceptibility zone.

[0036] In some embodiments, a susceptible area determination model can be used to determine multiple pressure damage susceptible areas and a cumulative load damage prediction map for each pressure damage susceptible area. The susceptible area determination model is a deep neural network model. The input to the susceptible area determination model is the interfacial stress distribution evolution data of the pressure-bearing site, and the output of the susceptible area determination model is multiple pressure damage susceptible areas and a cumulative load damage prediction map for each pressure damage susceptible area.

[0037] Deep neural network models include deep neural networks (DNNs). A deep neural network is a computational model that can simulate a biological nervous system and can map features of complex nonlinear relationships by constructing multiple hidden layers. Through the weight connections between neurons and the nonlinear transformation of activation functions, deep neural networks can learn high-dimensional abstract features from massive amounts of data and achieve accurate classification or regression predictions.

[0038] Pressure injury susceptibility zones are candidate areas in the human body that have potential injury risk, identified by a susceptibility zone determination model.

[0039] Pressure injury susceptibility zones are used to indicate the tendency of soft tissue in this area to experience biomechanical imbalance due to sustained load.

[0040] Areas susceptible to pressure injuries can serve as target monitoring zones for subsequent pathological sampling and in-depth evolution analysis.

[0041] The cumulative load damage prediction map for each pressure injury susceptibility zone is a distribution map output by the susceptibility zone determination model, which quantitatively predicts the damage risk of each spatial location within the pressure injury susceptibility zone. The cumulative load damage prediction map for each pressure injury susceptibility zone includes the predicted damage level for each pixel location and the corresponding confidence probability distribution.

[0042] The data on the evolution of interfacial stress distribution at the pressure site records the duration of pressure action, intensity fluctuations, and shear force components. These physical quantities directly determine the biomechanical response of deep tissues. The model can identify the evolution patterns of pressure peaks and cyclic loading through time series analysis.

[0043] Deep neural networks receive data on the evolution of interfacial stress distribution at the pressure site through the input layer. Then, using hidden layers, they can extract the spatial topological features of stress concentration areas and capture the dynamic trend of pressure accumulation over time. Based on learned tissue tolerance thresholds, the model can automatically identify areas where interfacial stress continuously exceeds the critical point and cannot be alleviated by minor postural changes, thus determining multiple pressure injury susceptibility zones. For each identified region, the deep neural network can further calculate the degree of tissue attenuation at each coordinate point under the current load history through regression layers. Combining this with damage evolution paths from historical samples, it assigns a damage probability value and damage level to each site. Finally, the model uses backprojection technology to generate a cumulative load damage prediction map for each pressure injury susceptibility zone.

[0044] In some embodiments, determining multiple pressure damage susceptibility zones and a cumulative load damage prediction map for each pressure damage susceptibility zone based on the interfacial stress distribution evolution data of the pressure-bearing portion includes steps S21 to S23:

[0045] Step S21: Based on the evolution data of the interface stress distribution of the compressed part, determine the coordinates of multiple interface stress concentration points, the duration of the pressure peak at each interface stress concentration point, and the shear force gradient data around each interface stress concentration point.

[0046] In some embodiments, a deep neural network can be used to determine the coordinates of multiple interface stress concentration points, the duration of the peak pressure at each interface stress concentration point, and the shear force gradient data around each interface stress concentration point.

[0047] The coordinates of multiple interface stress concentration points were determined by a deep neural network, which identified the physical locations on the skin surface with the highest pressure values ​​and the most prominent risks.

[0048] The duration of the peak pressure at each interface stress concentration point is determined by a deep neural network, which measures the cumulative duration during which the pressure intensity at the interface stress concentration point remains above a preset high-risk threshold.

[0049] The shear force gradient data around each interface stress concentration point is determined by a deep neural network. It is a quantitative index of the rate at which the shear force changes with spatial location in and around the interface stress concentration point.

[0050] The shear force gradient data around each interface stress concentration point includes the direction of the shear force change vector and the rate of change of force per unit length.

[0051] Deep neural networks can capture the frequency of pressure fluctuations at various spatial locations in the evolution data of interfacial stress distribution at stressed sites, thereby identifying local extrema where stress values ​​remain at a sustained peak. Deep neural networks can deduce the spatial evolution trend of interfacial shear force by calculating the difference in tangential force between adjacent sensor units. Through feature mapping layers, deep neural networks can transform physical signals into mechanical feature vectors and, combined with pre-defined threshold filtering logic, accurately locate all high-risk physical coordinates, thereby determining the coordinates of multiple interfacial stress concentration points, the duration of the pressure peak at each stress concentration point, and the shear force gradient data around each stress concentration point.

[0052] Step S22: Based on the coordinates of the multiple interface stress concentration points, the duration of the pressure peak at each interface stress concentration point, and the shear force gradient data around each interface stress concentration point, determine the preliminary boundary range of multiple pressure damage susceptibility zones, the distribution location of stress cores in each zone, the damage contribution index of each stress core, and the proportion of stress concentration points with high damage contribution index in each zone.

[0053] In some embodiments, a deep neural network can be used to determine the initial boundary range of multiple pressure damage susceptibility zones, the location of stress cores in each zone, the damage contribution index of each stress core, and the proportion of stress concentration points with high damage contribution index in each zone.

[0054] The preliminary boundary ranges of multiple pressure injury susceptibility zones are candidate geometric regions with injury risk determined by deep neural networks.

[0055] The location of the stress core distribution within each region is a set of coordinates of the key stress center points that exert the strongest destructive force on the tissue structure within the pressure damage susceptibility zone, determined by a deep neural network.

[0056] The damage contribution index of each stress core is a risk assessment value determined by a deep neural network, which quantifies the degree of impact of each specific stress core on the occurrence of substantial damage to soft tissue.

[0057] The percentage of stress concentration points with high damage contribution index in each region is the ratio of the number of stress concentration points in that region whose damage contribution index exceeds a preset threshold to the total number of stress concentration points in that region.

[0058] Deep neural networks, by constructing a spatial topological correlation matrix, can analyze the geometric distribution patterns of stress concentration points, thereby identifying physically adjacent and characteristically continuous sites as a set of features with a common evolutionary trend. The model utilizes nonlinear regression units to integrate pressure duration and shear force gradient to calculate the damage contribution index of each point. Subsequently, the model can filter out sites with high damage potential based on preset thresholds, and calculate the proportion of stress concentration points with high damage contribution indices based on the distribution density of such sites within a local area. Deep neural networks can also identify the spatial connectivity of stress cores and achieve spatial closure through boundary extrapolation algorithms, ultimately accurately determining the preliminary boundary range, stress core location, damage contribution index, and site proportion.

[0059] Step S23: Based on the interface stress distribution evolution data of the pressure-bearing part, the preliminary boundary range of the multiple pressure damage susceptibility zones, the stress core distribution location in each zone, the damage contribution index of each stress core, and the proportion of stress concentration points with high damage contribution index in each zone, determine the multiple pressure damage susceptibility zones and the cumulative load damage prediction map of each pressure damage susceptibility zone.

[0060] In some embodiments, deep neural networks can be used to determine multiple stress injury susceptibility zones and a cumulative load damage prediction map for each stress injury susceptibility zone.

[0061] Deep neural networks perform multi-dimensional correlation mapping on input data through multi-channel input layers. The model can simulate the biomechanical decay process of soft tissue under current cumulative load by using initial boundary ranges as spatial constraints, and infer the fatigue state of deep tissues based on the damage contribution index of the stress core. Then, the model can dynamically correct the region edges by incorporating the proportion of high damage index points within the region. Through a back-projection algorithm, the deep neural network can map the calculated damage risk value to the corresponding spatial coordinate system to generate a prediction field with continuous gradient characteristics, thereby ultimately determining multiple pressure injury susceptibility zones and a cumulative load damage prediction map for each pressure injury susceptibility zone.

[0062] Step S3: Obtain surface morphology visualization images of multiple pressure-sensitive areas.

[0063] The surface morphology visualization images of multiple pressure injury-prone areas are obtained through non-contact scanning of identified pressure injury-prone areas using a high-resolution digital imaging device under multispectral light source conditions, resulting in multidimensional visual data of the skin surface. These images contain epidermal texture topology information of the pressure-affected skin, local color deviation data caused by capillary dilation, and information on differences in skin surface reflectivity due to tissue edema.

[0064] Step S4: Based on the surface morphology visualization images of the multiple pressure injury susceptible areas, determine multiple pathological feature sampling sites for each pressure injury susceptible area, and obtain tissue physical characteristic data for each pathological feature sampling site.

[0065] In some embodiments, a sampling analysis model can be used to determine multiple pathological feature sampling sites for each pressure injury susceptibility zone. The sampling analysis model is a convolutional neural network. The input to the sampling analysis model is a surface morphology visualization image of the multiple pressure injury susceptibility zones, and the output of the sampling analysis model is multiple pathological feature sampling sites for each pressure injury susceptibility zone.

[0066] Convolutional Neural Networks (CNNs) are neural network models capable of processing grid-structured data such as images. CNNs extract local spatial features using convolutional kernels with shared weights through convolutional layers, pooling layers reduce feature dimensionality while preserving translation invariance, and finally, fully connected layers aggregate and output the features.

[0067] Multiple pathological feature sampling sites in each pressure injury susceptibility zone are determined by a sampling analysis model and are specific locations within the pressure injury susceptibility zone used to collect tissue physical property data.

[0068] The tissue physical properties data for each pathological feature sampling site were collected using biomechanical and humidity sensors at the corresponding pathological feature sampling site. These data are quantitative indicators describing the biomechanical and physiological state of a specific site. The tissue physical properties data for each pathological feature sampling site include tissue elastic modulus, tissue stiffness, and local water content.

[0069] Visualizing the surface morphology of multiple pressure-sensitive areas can reveal early signs of skin lesions, such as localized color differences caused by capillary dilation or texture changes due to tissue edema. These visual features reveal the projection of subcutaneous damage onto the surface, thus guiding the model to set sampling points at the coordinates where the lesion is most likely to be severe.

[0070] Convolutional neural networks (CNNs) can perform pixel-by-pixel scanning of surface morphology visualization images of multiple pressure injury-prone areas through multi-layer filters to extract subtle gradient changes in skin texture and color feature vectors. By comparing the feature distribution of normal skin, CNNs can identify local areas in the image that exhibit morphological distortion or color abnormalities. CNNs can also automatically mark the most diagnostically valuable coordinates in high-contrast regions and areas with abrupt edge changes using saliency detection algorithms, thereby determining multiple pathological feature sampling sites for each pressure injury-prone area.

[0071] Step S5: Based on the tissue physical characteristics data of each pathological feature sampling site and the cumulative load damage prediction map of each pressure injury susceptible area, determine multiple advanced pathological feature sampling sites for each pressure injury susceptible area, and obtain the tissue physical characteristics data of each advanced pathological feature sampling site.

[0072] In some embodiments, a second sampling analysis model can be used to determine multiple advanced pathological feature sampling sites for each pressure injury susceptibility zone. The second sampling analysis model is a convolutional neural network. The inputs to the second sampling analysis model are the tissue physical property data of each pathological feature sampling site and the cumulative load damage prediction map of each pressure injury susceptibility zone. The output of the second sampling analysis model is multiple advanced pathological feature sampling sites for each pressure injury susceptibility zone.

[0073] Advanced pathological feature sampling sites are determined by the second sampling analysis model. These sampling locations can reflect the potential damage of deep tissues more deeply based on the initial sampling.

[0074] The tissue physical properties data for each advanced pathological feature sampling site were collected using biomechanical and humidity sensors at the corresponding advanced pathological feature sampling site. These data represent quantitative indicators reflecting the biomechanical and physiological state of that site. The tissue physical properties data for each advanced pathological feature sampling site include tissue elastic modulus, tissue stiffness, and local water content.

[0075] The cumulative load damage prediction map of each pressure injury-prone area provides a global risk prediction trend, and the tissue physical property data of each pathological feature sampling site provides the measured values ​​of the existing sites. This can make up for the blind spots of single visual analysis and guide the model to perform precise positioning and reinforcement sampling of advanced pathological feature sampling sites in spatial coordinates where the predicted risk value is high and there is a lack of measured data support.

[0076] Convolutional neural networks (CNNs) can construct a multi-channel input layer that integrates spatial probability maps and discrete physical features. This allows the CNN to use the cumulative load damage prediction map of each pressure injury susceptibility zone as a spatial guiding template and map the tissue physical property data of each pathological feature sampling site to the corresponding coordinate grid. By calculating the residual between the predicted damage degree and the measured physical parameters, the CNN can identify transitional zones where data inconsistencies or risk gradient changes are most dramatic. In these zones, the CNN can use feature activation maps to locate new key points of interest, thereby determining multiple advanced pathological feature sampling sites for each pressure injury susceptibility zone.

[0077] Step S6: Based on the tissue physical characteristic data of each pathological feature sampling site in each pressure injury susceptible area and the tissue physical characteristic data of each advanced pathological feature sampling site, generate a panoramic view of the deep injury evolution of each pressure injury susceptible area.

[0078] In some embodiments, an evolutionary generative model can be used to generate a panoramic view of the deep injury evolution of each pressure injury susceptibility zone. The evolutionary generative model is a deep neural network. The input to the evolutionary generative model is the tissue physical property data of each pathological feature sampling site and the tissue physical property data of each advanced pathological feature sampling site in each pressure injury susceptibility zone. The output of the evolutionary generative model is a panoramic view of the deep injury evolution of each pressure injury susceptibility zone.

[0079] The panoramic view of deep injury evolution in each pressure injury-prone area is a digital three-dimensional map reflecting the temporal evolution and spatial spread of soft tissue injury beneath the skin surface. The panoramic view of deep injury evolution in each pressure injury-prone area includes the density gradient distribution of tissue necrosis at different depths, the geometric boundaries of deep hematoma spread, and the degree of fascial layer injury.

[0080] The tissue necrosis density gradient distribution map is a quantitative distribution map that represents the continuous change in the degree of cell inactivation within soft tissue in three-dimensional space.

[0081] Tissue necrosis density gradient maps use different color intensities and numerical densities to represent the risk of irreversible tissue damage at specific coordinate points. These maps can demonstrate the smooth transition of damage from the necrotic core to the surrounding ischemic penumbra.

[0082] The geometric boundary of deep hematoma spread represents the extent of the spread of a hematoma formed by bleeding within deep tissues in three-dimensional space.

[0083] The fascia layer damage grade is a grading assessment indicator for the degree of damage to fascial tissue.

[0084] The tissue physical property data of each pathological feature sampling site in each pressure injury-prone area, along with the tissue physical property data of each advanced pathological feature sampling site, constitute a sparse, multidimensional sample of tissue health status. This data includes fundamental parameters such as tissue elastic modulus, tissue stiffness, and local water content, which can serve as boundary conditions and constraints for generating deep pathological distributions.

[0085] Deep neural networks (DNNs) can map tissue physical property data from multiple sites into a high-dimensional feature space through fully connected layers, and simulate the biomechanical response of tissues under long-term pressure loading using recursive structures. DNNs can infer stress transmission paths and inflammation diffusion dynamics within tissues based on differences in elastic modulus, tissue stiffness distribution, and local water content variations at different sites. DNNs utilize a generative architecture to spatially interpolate and semantically complete discrete sampling site features, and combine this with pre-defined anatomical constraints to generate a continuous feature field of deep tissue states. Through multi-scale deconvolution operations, the model can transform the feature field into a visual representation with spatial depth, thus displaying the damage gradient from the dermis to the periosteum, ultimately generating a panoramic view of the deep injury evolution of each pressure injury-prone area.

[0086] Step S7: Based on the cumulative load damage prediction map of each pressure injury susceptible area and the panoramic map of the deep damage evolution of each pressure injury susceptible area, determine the hidden deep tissue damage area.

[0087] In some embodiments, a damage area identification model can be used to determine occult deep tissue damage areas. The damage area identification model is a convolutional neural network. The inputs to the damage area identification model are a cumulative load damage prediction map of each pressure injury susceptibility zone and a panoramic view of the deep damage evolution of each pressure injury susceptibility zone; the output of the damage area identification model is the occult deep tissue damage area.

[0088] The occult deep tissue injury zone is a high-risk core area identified through further screening within the pressure injury susceptibility zone using an injury zone identification model. It represents areas where deep tissues have undergone severe ischemia, necrosis, or structural changes. The occult deep tissue injury zone is used to characterize the precise spatial extent to which lesions have penetrated into deep fascia or muscle tissue.

[0089] The cumulative load damage prediction map for each pressure injury susceptibility zone provides the probability distribution of damage caused by external stress, while the panoramic view of deep damage evolution in each pressure injury susceptibility zone provides the structural pathological features of deep tissues. The spatial overlay of the two enhances the identification accuracy, enabling the model to locate occult deep tissue injuries where the pressure load and tissue damage degree are highly matched.

[0090] Convolutional neural networks (CNNs) can take the cumulative load damage prediction map of each pressure injury-prone area and the panoramic view of the deep damage evolution of each pressure injury-prone area as dual inputs, and map them in the same anatomical coordinate system. The CNN can use parallel convolutional layers to extract external stress features and deep pathological features separately, and find correlations through feature fusion layers. The model focuses on connected subgraphs that show high necrosis density gradients in the panoramic view and belong to high-load areas in the prediction map. By setting multi-level threshold judgments and edge detection, the CNN can determine the core boundary of deep tissue necrosis, thereby identifying occult deep tissue damage areas.

[0091] Step S8: Based on the tissue physical characteristics data of multiple pathological feature sampling sites in the occult deep tissue injury area and the tissue physical characteristics data of multiple advanced pathological feature sampling sites, determine the risk zone for pressure ulcer necrosis spread.

[0092] In some embodiments, Figure 3 This is a schematic flowchart illustrating the process for determining the risk zone of pressure ulcer necrosis according to an embodiment of the present invention. The determination of the risk zone includes steps S31-S33:

[0093] Step S31: Construct a wound knowledge graph. The wound knowledge graph includes multiple pathological feature sampling nodes and multiple advanced pathological feature sampling nodes. The node features of the pathological feature sampling nodes are the tissue physical characteristics data of the pathological feature sampling sites, and the node features of the advanced pathological feature sampling nodes are the tissue physical characteristics data of the advanced pathological feature sampling sites.

[0094] The wound knowledge graph consists of multiple nodes and multiple edges connecting them. The nodes include multiple pathological feature sampling nodes and multiple advanced pathological feature sampling nodes. Each node has specific node characteristics.

[0095] Pathological feature sampling nodes represent the sampling locations initially selected within the pressure injury-prone area. The node features of pathological feature sampling nodes are the tissue physical characteristics data of the pathological feature sampling sites.

[0096] Advanced pathological feature sampling nodes represent deeper sampling locations determined based on the initial sampling. The node features of advanced pathological feature sampling nodes are the tissue physical characteristics data of the advanced pathological feature sampling sites.

[0097] An edge represents the relationship between two nodes in the wound knowledge graph, and the feature of an edge is the similarity between the tissue physical characteristic data corresponding to the two nodes.

[0098] In some embodiments, a deep neural network can be used to determine the similarity between corresponding tissue physical property data between two nodes.

[0099] The evolution of wound tissue damage is influenced by the biomechanical and physiological states of multiple sampling sites. By constructing a wound knowledge graph, the correlation of tissue physical properties between sampling points at different depths and locations can be clearly represented, including the consistency of distribution of elastic modulus, tissue stiffness, and local water content, thus providing a detailed spatial topological basis for subsequent identification of damage propagation paths.

[0100] Step S32: Process the wound knowledge graph based on graph neural network to determine multiple prognostic verification sites in each hidden deep tissue injury area, and obtain the histopathological information of each prognostic verification site.

[0101] Graph Neural Networks (GNNs) are deep learning models that operate directly on knowledge graphs. Through message passing and feature aggregation mechanisms between neighboring nodes, GNNs can learn the embedded representations of nodes and their local neighborhoods, capturing spatial relationships and dynamic dependencies within the knowledge graph. The input to the GNN is the wound knowledge graph, and the output is multiple prognostic validation sites within each occult deep tissue injury area.

[0102] Multiple prognostic validation sites within each area of ​​occult deep tissue injury were determined using a graph neural network. These sites are key detection locations used to verify the trend of injury spread and to conduct prognostic assessment.

[0103] The histopathological information for each prognostic validation site is data describing the microscopic lesions and metabolic activity of the tissue, collected at the corresponding prognostic validation site using a microdialysis sampling analyzer and biochemical sensors. The histopathological information for each prognostic validation site includes local lactate concentration, inflammatory factor levels, and fibroblast activity parameters.

[0104] Fibroblast activity parameters are physiological indicators used to quantitatively assess the proliferation efficiency, metabolic level, and matrix synthesis capacity of fibroblasts in human connective tissue.

[0105] Fibroblast activity parameters include cell proliferation index, cell migration speed, and secretion of healing-promoting factors.

[0106] Wound knowledge graphs can represent the pathological spatial knowledge of pressure ulcer wounds and visually display the associations between various sampling sites and their tissue physical properties. This structured representation facilitates graph neural networks' understanding and processing of complex tissue damage propagation relationships. Through node features and edge information, graph neural networks can better capture the physiological similarities and physical differences between sampling points at different depths and locations. For example, tissue stiffness and water content in node features provide the substantial damage state of the soft tissue at that site, while the similarity on edges reflects the continuity and diffusion trend of these states in the anatomical structure.

[0107] The construction of a wound knowledge graph allows for the effective organization of scattered sampling points and their physiological relationships, facilitating aggregation computation by graph neural networks. This organization reduces computational complexity and improves the training and inference efficiency of the model when determining prognostic validation sites. Because the wound knowledge graph provides structured node and edge information, graph neural networks can better handle multidimensional physical parameters, thus avoiding the data sparsity and curse of spatial dimensionality problems caused by discrete sampling points in traditional methods. This structural advantage makes graph neural networks more effective in handling complex spatial relationships in cases of hidden, deep tissue injuries.

[0108] By comprehensively analyzing the physical characteristics and similarity relationships of nodes in the wound knowledge graph, graph neural networks can accurately identify key nodes downstream of the necrosis chain reaction, thereby finding the best verification site in complex, hidden injury areas that can both represent the direction of lesion spread and verify the prognostic trend.

[0109] Graph neural networks (GNNs) aggregate the tissue physical characteristics of each node in the wound knowledge graph through feature propagation in each iteration, thereby calculating the importance score of each node in the local microenvironment. By analyzing the gradient differences and connectivity of physiological parameters between nodes, GNNs can identify the node paths with the strongest damage diffusion signals. GNNs utilize attention mechanisms to focus on regions with abnormal metabolic trends at the edges of occult deep tissue injuries, selecting central coordinates that represent the future course of the disease, thus determining multiple prognostic validation sites within each occult deep tissue injury area.

[0110] Step S33: Determine the risk zone for pressure ulcer necrosis spread based on the histopathological information of each prognostic verification site.

[0111] In some embodiments, Figure 4 This invention provides a schematic flowchart for determining the risk zone of pressure ulcer necrosis based on histopathological information at each prognostic verification site. The determination of the risk zone includes steps S41-S43:

[0112] Step S41: Clustering is performed based on the histopathological information of each prognostic verification site to obtain multiple clusters.

[0113] The clustering method used is K-means clustering. K-means clustering is an unsupervised learning algorithm. Its core is to divide the dataset into K clusters and iteratively calculate the centroids of the clusters and update the data point affiliations, thereby maximizing the similarity of data points within the same cluster and minimizing the similarity between different clusters.

[0114] In some embodiments, the value of K can be determined using a preset table relating the value of K to the area of ​​the hidden deep tissue injury zone. The larger the area of ​​the hidden deep tissue injury zone, the larger the value of K. The preset table relating the value of K to the area of ​​the hidden deep tissue injury zone is artificially constructed in advance.

[0115] Multiple clusters are datasets divided using the K-means clustering algorithm based on the pathological similarity of prognostic validation sites. Each cluster represents a specific level of tissue necrosis risk and prognostic validation sites with similar pathological features.

[0116] Histopathological information at each prognostic validation site can exhibit a non-uniform distribution. Clustering can discretize complex pathological states to identify a typical subset of samples with high-risk characteristics.

[0117] In some embodiments, the process of clustering the histopathological information of each prognostic validation site using the K-means clustering algorithm is as follows: K histopathological information from prognostic validation sites are randomly selected as initial cluster centers. For each histopathological information from a prognostic validation site, its Euclidean distance to all initial cluster centers is calculated, and each histopathological information from a prognostic validation site is assigned to the nearest cluster center, thus forming K clusters. For each cluster, the average value of all histopathological information vectors within the cluster is calculated, and this average value is used as the new cluster center. The above steps are repeated until the cluster centers no longer change or a predetermined number of iterations is reached, at which point the clustering is complete.

[0118] Clustering effectively integrates the complex histopathological information of prognostic validation sites. Since the histopathological information of each prognostic validation site within a hidden deep tissue injury area involves multidimensional indicators and diverse characteristics, direct analysis of this information is quite challenging. However, clustering can group histopathological information with similar characteristics into one category and form multiple clusters, thus greatly simplifying the data structure and facilitating the extraction of valuable patterns in the evolution of necrosis risk. By dividing the histopathological information of each prognostic validation site into multiple clusters, the different degrees and distributions of tissue metabolic disorders within a specific injury area can be visually displayed. Analysis of these clusters can quickly determine the main characteristics and physiological conditions of tissue necrosis within that area. Further comparison of multiple clusters at different injury stages can clearly reveal the impact of the pressure ulcer necrosis process on the tissue microenvironment.

[0119] Step S42: Determine multiple necrosis propagation paths based on the multiple clusters.

[0120] In some embodiments, a diffusion trend prediction model can be used to determine multiple necrosis diffusion paths. The diffusion trend prediction model is a deep neural network. The input to the diffusion trend prediction model is the plurality of clusters, and the output of the diffusion trend prediction model is the multiple necrosis diffusion paths.

[0121] Multiple necrosis spread pathways are the expected trajectories of necrotic lesions spreading from the occult lesion core to surrounding tissues, determined by a spread trend prediction model. Each necrosis spread pathway includes a sequence of geometric coordinates and the expected spread rate.

[0122] Multiple clusters represent pathological heterogeneity in different regions, and the spatial distribution and risk levels of these clusters determine the potential direction of damage spread. The distribution of high-risk clusters reflects the active frontier of tissue destruction, and thus the dynamic evolution path of necrosis in physical space can be predicted through diffusion trend prediction models.

[0123] Deep neural networks can analyze the physiological potential differences between different clusters. They can calculate the diffusion tensor generated by each risk cluster in the surrounding space and use the shortest path algorithm to search for the continuous coordinate sequence of the most severe pathological deterioration in the spatial grid, thereby determining multiple necrosis diffusion paths.

[0124] Step S43: Determine the risk zone for pressure ulcer necrosis spread based on the multiple necrosis spread paths.

[0125] In some embodiments, a diffusion region determination model can be used to determine the risk diffusion zone of pressure ulcer necrosis. The diffusion region determination model is a Transformer model. The input to the diffusion region determination model is the multiple necrosis diffusion paths, and the output of the diffusion region determination model is the risk diffusion zone of pressure ulcer necrosis.

[0126] The Transformer model is a deep learning architecture based on a self-attention mechanism, enabling deep fusion of sequence features. Through multi-head attention, the Transformer model can calculate the correlation strength between any two elements in the input sequence. The Transformer model includes encoder and decoder structures.

[0127] The risk zone for pressure ulcer necrosis is defined by a diffusion area determination model, and is a complete coverage area that includes occult deep tissue damage and where the necrosis process is spreading or about to spread to surrounding healthy tissue.

[0128] Multiple necrosis propagation paths can represent the dynamic trend and boundary dynamics of damage spread. These path sequences contain temporal and spatial information about necrosis propagation, providing crucial vector basis for the model to delineate the ultimately threatened closed region.

[0129] The Transformer model maps multiple necrosis propagation paths into feature vectors with location information and time-series attributes through positional encoding, and uses a multi-head self-attention mechanism to analyze the mutual influence and spatial overlap between the paths. The Transformer model globally weights features at path intersections, points of anomalous propagation velocity, and surrounding susceptible areas to extract features reflecting the distribution of necrosis energy. Based on a learned propagation threshold, the Transformer model assesses the risk to soft tissues affected by the paths and performs semantic segmentation and edge closure on the geometric space encompassed and affected by multiple paths. Through the region mask output by the decoder, the Transformer model can accurately delineate all danger zones where necrosis may occur under the current trend, ultimately identifying the dangerous propagation zone of pressure ulcer necrosis.

[0130] Based on the same inventive concept Figure 5 This is a schematic diagram of a wound infection nursing information determination system provided in an embodiment of the present invention. The wound infection nursing information determination system includes:

[0131] The data acquisition module 51 is used to acquire data on the evolution of interfacial stress distribution in the compressed part;

[0132] The susceptible area determination module 52 is used to determine multiple pressure damage susceptible areas and a cumulative load damage prediction map of each pressure damage susceptible area based on the interface stress distribution evolution data of the pressure-bearing part.

[0133] Image acquisition module 53 is used to acquire surface morphology visualization images of multiple pressure-damage-prone areas;

[0134] The sampling analysis module 54 is used to determine multiple pathological feature sampling sites for each pressure injury susceptible area based on the surface morphology visualization images of the multiple pressure injury susceptible areas, and to obtain tissue physical characteristic data for each pathological feature sampling site.

[0135] The advanced sampling module 55 is used to determine multiple advanced pathological feature sampling sites for each pressure injury susceptible area based on the tissue physical characteristic data of each pathological feature sampling site and the cumulative load damage prediction map of each pressure injury susceptible area, and to acquire the tissue physical characteristic data of each advanced pathological feature sampling site.

[0136] Evolution generation module 56 is used to generate a panoramic view of the deep injury evolution of each pressure injury susceptibility zone based on the tissue physical property data of each pathological feature sampling site of each pressure injury susceptibility zone and the tissue physical property data of each advanced pathological feature sampling site.

[0137] The damage area identification module 57 is used to identify hidden deep tissue damage areas based on the cumulative load damage prediction map of each pressure injury susceptibility area and the panoramic map of deep damage evolution of each pressure injury susceptibility area.

[0138] The diffusion prediction module 58 is used to determine the risk diffusion area of ​​pressure ulcer necrosis based on the tissue physical characteristics data of multiple pathological feature sampling sites in the occult deep tissue injury area and the tissue physical characteristics data of multiple advanced pathological feature sampling sites.

[0139] It should be noted that, in order to simplify the descriptions disclosed herein and thus aid in the understanding of one or more embodiments of the invention, the foregoing description of embodiments of this specification may sometimes combine multiple features into a single embodiment, drawing, or description thereof. However, this method of disclosure does not imply that the subject matter of this specification requires more features than those mentioned in the claims. In fact, the embodiments contain fewer features than all the features of a single embodiment disclosed above.

[0140] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.

Claims

1. A method for determining nursing information for wound infection, characterized in that, include: Obtain data on the evolution of interfacial stress distribution in the compressed area; Based on the evolution data of interfacial stress distribution at the pressure-bearing site, multiple pressure damage susceptibility zones and a cumulative load damage prediction map for each pressure damage susceptibility zone are determined. Obtain surface morphology visualization images of multiple pressure-damage-prone areas; Based on the surface morphology visualization images of the multiple pressure injury susceptibility areas, multiple pathological feature sampling sites for each pressure injury susceptibility area are determined, and tissue physical property data for each pathological feature sampling site are obtained. Based on the tissue physical property data of each pathological feature sampling site and the cumulative load damage prediction map of each pressure injury susceptible area, multiple advanced pathological feature sampling sites for each pressure injury susceptible area are determined, and the tissue physical property data of each advanced pathological feature sampling site are obtained. Based on the tissue physical characteristics data of each pathological feature sampling site in each pressure injury susceptible area and the tissue physical characteristics data of each advanced pathological feature sampling site, a panoramic view of the deep injury evolution of each pressure injury susceptible area is generated. Based on the cumulative load damage prediction map of each pressure injury susceptibility zone and the panoramic map of the deep damage evolution of each pressure injury susceptibility zone, the hidden deep tissue damage zone is identified. Based on the tissue physical characteristics data of multiple pathological feature sampling sites in the occult deep tissue injury area and the tissue physical characteristics data of multiple advanced pathological feature sampling sites, the risk zone for pressure ulcer necrosis diffusion was determined.

2. The method for determining nursing information for wound infection as described in claim 1, characterized in that, The determination of the pressure ulcer necrosis risk zone based on tissue physical characteristic data from multiple pathological feature sampling sites within the occult deep tissue injury area and tissue physical characteristic data from multiple advanced pathological feature sampling sites includes: A wound knowledge graph is constructed, which includes multiple pathological feature sampling nodes and multiple advanced pathological feature sampling nodes. The node features of the pathological feature sampling nodes are the tissue physical characteristics data of the pathological feature sampling sites, and the node features of the advanced pathological feature sampling nodes are the tissue physical characteristics data of the advanced pathological feature sampling sites. The wound knowledge graph is processed using a graph neural network to identify multiple prognostic verification sites within each occult deep tissue injury area, and histopathological information of each prognostic verification site is obtained. Based on the histopathological information of each prognostic verification site, the risk zone for pressure ulcer necrosis spread is determined.

3. The method for determining nursing information for wound infection as described in claim 2, characterized in that, The determination of the risk zone for pressure ulcer necrosis spread based on the histopathological information of each prognostic verification site includes: Multiple clusters were obtained by clustering based on the histopathological information of each prognostic validation site; Multiple necrosis propagation paths are determined based on the aforementioned clusters; Based on the aforementioned multiple necrosis spread pathways, the risk zone for pressure ulcer necrosis spread is determined.

4. The method for determining nursing information for wound infection as described in claim 2, characterized in that, The input to the graph neural network is the wound knowledge graph, and the output of the graph neural network is multiple prognostic verification sites within each hidden deep tissue injury area.

5. A system for determining nursing information related to wound infection, characterized in that, include: The data acquisition module is used to acquire data on the evolution of interfacial stress distribution in the compressed area; The vulnerable area determination module is used to determine multiple pressure damage vulnerable areas and a cumulative load damage prediction map for each pressure damage vulnerable area based on the evolution data of the interface stress distribution of the pressure-bearing part. The image acquisition module is used to acquire visual images of the surface morphology of multiple pressure-damage-prone areas; The sampling analysis module is used to determine multiple pathological feature sampling sites for each pressure injury susceptible area based on the surface morphology visualization images of the multiple pressure injury susceptible areas, and to acquire tissue physical characteristic data for each pathological feature sampling site. The advanced sampling module is used to determine multiple advanced pathological feature sampling sites for each pressure injury susceptible area based on the tissue physical characteristic data of each pathological feature sampling site and the cumulative load damage prediction map of each pressure injury susceptible area, and to acquire the tissue physical characteristic data of each advanced pathological feature sampling site. The evolution generation module is used to generate a panoramic view of the deep injury evolution of each pressure injury susceptibility area based on the tissue physical property data of each pathological feature sampling site of each pressure injury susceptibility area and the tissue physical property data of each advanced pathological feature sampling site. The damage area identification module is used to identify hidden deep tissue damage areas based on the cumulative load damage prediction map of each pressure injury susceptibility area and the panoramic map of deep damage evolution of each pressure injury susceptibility area. The diffusion prediction module is used to determine the risk zone for pressure ulcer necrosis diffusion based on the tissue physical characteristics data of multiple pathological feature sampling sites within the occult deep tissue injury area and the tissue physical characteristics data of multiple advanced pathological feature sampling sites.

6. The system for determining nursing information of wound infection as described in claim 5, characterized in that, The diffusion prediction module is also used for: A wound knowledge graph is constructed, which includes multiple pathological feature sampling nodes and multiple advanced pathological feature sampling nodes. The node features of the pathological feature sampling nodes are the tissue physical characteristics data of the pathological feature sampling sites, and the node features of the advanced pathological feature sampling nodes are the tissue physical characteristics data of the advanced pathological feature sampling sites. The wound knowledge graph is processed using a graph neural network to identify multiple prognostic verification sites within each occult deep tissue injury area, and histopathological information of each prognostic verification site is obtained. Based on the histopathological information of each prognostic verification site, the risk zone for pressure ulcer necrosis spread is determined.

7. The system for determining nursing information of wound infection as described in claim 6, characterized in that, The determination of the risk zone for pressure ulcer necrosis spread based on the histopathological information of each prognostic verification site includes: Multiple clusters were obtained by clustering based on the histopathological information of each prognostic validation site; Multiple necrosis propagation paths are determined based on the aforementioned clusters; Based on the aforementioned multiple necrosis spread pathways, the risk zone for pressure ulcer necrosis spread is determined.

8. The system for determining nursing information of wound infection as described in claim 6, characterized in that, The input to the graph neural network is the wound knowledge graph, and the output of the graph neural network is multiple prognostic verification sites within each hidden deep tissue injury area.

Images