Urology postoperative nursing management system and method

By using hollow pyramid network model and deep learning algorithm to extract wound multi-scale features in traditional nursing care management system, the problems of missed diagnosis and inefficiency in traditional nursing are solved, and the intelligentization of wound infection in patients after urology is achieved. Identify and prevent.

CN118521574BActive Publication Date: 2024-11-05JILIN UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410969751.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-05
Estimated Expiration
2044-07-19

AI Technical Summary

Technical Problem

Traditional postoperative care of urology relies on the experience of medical staff and regular examinations, which easily ignore subtle changes in wounds or early signs of infection, resulting in missed diagnosis or delayed diagnosis, and is inefficient in a large number of patients or patients who require long-term monitoring, and is not timely monitoring. and the care plan is not personalized enough.

Method used

A postoperative nursing management system for urology is designed to collect wound status images through cameras, extract multi-scale features using hollow pyramid network model and deep learning algorithms, and perform wound infection detection to achieve intelligent identification and monitoring.

Benefits of technology

It has realized intelligent identification of wound infections in patients with urology, avoided the defects of traditional inspections, helped medical staff to monitor the recovery process, manage nursing plans, and prevent wound infection problems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118521574B_ABST
    Figure CN118521574B_ABST
Patent Text Reader

Abstract

The present application discloses a postoperative nursing management system and method for urology surgery, which relates to the field of intelligent management technology. The system collects wound status images through a camera, and introduces an image processing and analysis algorithm based on artificial intelligence and deep learning at the back end to analyze the wound status images, so as to capture the multi-level and multi-scale fusion features of the patient's wound status, thereby making a judgment on wound infection. In this way, it is possible to realize intelligent identification of wound infection in urology surgery patients, thereby avoiding the defects caused by traditional reliance on medical staff for inspections, and can help medical staff monitor the patient's recovery process, manage the nursing plan for urology surgery patients, and prevent wound infection problems.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of intelligent management, and more specifically, to a urology postoperative care management system and method. Background Art

[0002] With the aging of the population and changes in lifestyle, the incidence of urological diseases has increased year by year, leading to a growing demand for urological surgery and postoperative care. Urological surgery is a common surgical procedure that involves the treatment of diseases of the urinary system, such as urinary tract stones, prostatic hyperplasia, urinary tract infections, etc. Postoperative nursing management of urological surgery is essential for the patient's recovery and prevention of complications. Good nursing management can help patients relieve pain, prevent infection, promote wound healing, and ultimately improve the patient's quality of life.

[0003] Wound infection is one of the common complications after surgery, which may lead to serious health problems and even prolong the patient's recovery time. However, traditional postoperative care relies on the experience and regular examinations of medical staff. This method is affected by the individual experience and ability of medical staff, and it is easy to ignore subtle changes in the wound or early signs of infection, resulting in missed or delayed diagnosis, leading to instability and inconsistency in the nursing process and quality. In addition, traditional nursing management requires medical staff to regularly visit patients, which consumes a lot of time and human resources. For a large number of patients or patients who need long-term monitoring, the traditional method will have problems such as inefficiency, untimely monitoring and insufficient personalization of care plans.

[0004] Therefore, an optimized urology postoperative care management system and method are desired. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed.

[0006] According to one aspect of the present application, a urology postoperative care management system is provided, comprising:

[0007] Patient information management module, used to manage patient information;

[0008] Nursing task scheduling module, used to manage nursing tasks;

[0009] Drug management module, used to manage drug information;

[0010] Complication monitoring module, used to monitor complications;

[0011] Patient education module, used to publish patient education content;

[0012] Wherein, the complication monitoring module includes:

[0013] A wound status image acquisition unit, used to acquire a wound status image acquired by a camera;

[0014] A wound status multi-scale feature extraction unit, used for inputting the wound status image into a wound status multi-scale feature extractor based on a hollow pyramid network model to obtain a wound status shallow feature map and a wound status deep feature map;

[0015] A feature collaborative attention unit, used for inputting the wound state shallow feature map and the wound state deep feature map into a feature collaborative attention module to obtain a collaboratively enhanced wound state shallow feature map and a collaboratively enhanced wound state deep feature map;

[0016] A multi-scale feature scanning and enhancement unit, used for performing feature multi-scale scanning on the synergistically enhanced wound state shallow feature map and the synergistically enhanced wound state deep feature map to obtain a synergistically enhanced wound state shallow scanning feature vector and a synergistically enhanced wound state deep scanning feature vector;

[0017] A wound state multi-scale fusion representation unit, used for inputting the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector into a balanced threshold feature vector adaptive fusion module to obtain a wound state multi-scale fusion representation vector;

[0018] The wound infection detection unit is used to input the multi-scale fusion representation vector of the wound state into a classifier-based monitoring module to obtain a monitoring result, and the monitoring result is used to indicate whether the wound is infected.

[0019] The feature collaborative attention unit is used to:

[0020] Performing maximum pooling processing and upsampling processing based on channel dimension on the shallow feature map of wound status to obtain a shallow collaborative attention pooling matrix of wound status;

[0021] Masking the wound state shallow collaborative attention pooling matrix to obtain a masked wound state shallow collaborative attention pooling matrix;

[0022] The masked wound status shallow collaborative attention pooling matrix is ​​used to weight each feature matrix along the channel dimension in the wound status shallow feature map to obtain the collaboratively enhanced wound status shallow feature map.

[0023] In the above-mentioned urology postoperative care management system, the multi-scale feature scanning enhancement unit is used to: input the synergistically enhanced wound state shallow feature map and the synergistically enhanced wound state deep feature map into the feature multi-scale scanning module respectively to obtain the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector.

[0024] In the above-mentioned urology postoperative care management system, the multi-scale feature scanning enhancement unit is used to:

[0025] Upsampling the synergistically enhanced wound state shallow feature map to obtain an upsampled wound state shallow feature map;

[0026] Performing global mean pooling processing on each feature matrix along the channel dimension in the upsampled wound status shallow feature map to obtain an upsampled wound status shallow channel feature vector;

[0027] Performing point convolution encoding on the upsampled wound status shallow channel feature vector to obtain a wound status shallow global feature vector;

[0028] Performing point convolution encoding on the upsampled wound state shallow feature map to obtain a channel modulated upsampled wound state shallow feature map;

[0029] Performing two-dimensional convolution encoding on the channel-modulated upsampled wound state shallow feature map to obtain a wound state shallow local feature vector;

[0030] The shallow global feature vector of the wound state and the shallow local feature vector of the wound state are fused to obtain the synergistically enhanced shallow scanning feature vector of the wound state.

[0031] In the above-mentioned urology postoperative care management system, the wound status multi-scale fusion representation unit includes:

[0032] A threshold constraint calculation subunit, used for performing threshold constraints on the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector based on different linear transformations and fusion methods to obtain a first activated linear transformation wound state multi-scale association feature vector, a second activated linear transformation wound state multi-scale association feature vector, and a third activated linear transformation wound state multi-scale association feature vector;

[0033] The wound status multi-scale feature constraint fusion representation subunit is used to use the respective position feature values ​​in the first activated linear transformation wound status multi-scale associated feature vector as the first threshold value, the respective position feature values ​​in the second activated linear transformation wound status multi-scale associated feature vector as the second threshold value, and the respective position feature values ​​in the third activated linear transformation wound status multi-scale associated feature vector as the third threshold value, and calculate the weighted sum of the synergistically enhanced wound status shallow scanning feature vector and the synergistically enhanced wound status deep scanning feature vector to obtain the wound status multi-scale fusion representation vector.

[0034] In the above-mentioned urology postoperative care management system, the threshold constraint calculation subunit is used to:

[0035] Cascading the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector to obtain a first wound state multi-scale correlation feature vector;

[0036] After calculating the matrix multiplication of the first wound state multi-scale associated feature vector and the first transformation matrix, the obtained feature vector is added to the first bias vector by position to obtain a first linearly transformed wound state multi-scale associated feature vector;

[0037] use An activation function activates the first linearly transformed wound state multi-scale associated feature vector to obtain the first activated linearly transformed wound state multi-scale associated feature vector;

[0038] Adding the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector according to position to obtain a second wound state multi-scale correlation feature vector;

[0039] After calculating the matrix multiplication of the second wound state multi-scale associated feature vector and the second transformation matrix, the obtained feature vector is added to the second bias vector by position to obtain a second linearly transformed wound state multi-scale associated feature vector;

[0040] use An activation function activates the second linearly transformed wound state multi-scale associated feature vector to obtain the second activated linearly transformed wound state multi-scale associated feature vector;

[0041] Multiplying the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector by position points to obtain a third wound state multi-scale correlation feature vector;

[0042] After calculating the matrix multiplication of the third wound state multi-scale associated feature vector and the third transformation matrix, the obtained feature vector is added to the third bias vector by position to obtain a third linearly transformed wound state multi-scale associated feature vector;

[0043] use The activation function activates the third linearly transformed wound state multi-scale associated feature vector to obtain the third activated linearly transformed wound state multi-scale associated feature vector.

[0044] In the above-mentioned urology postoperative care management system, the wound status multi-scale feature constraint fusion representation subunit is used to:

[0045] Calculating the average of the first threshold value, the second threshold value and the third threshold value to obtain a threshold average;

[0046] Calculate the product of the eigenvalue of each position in the synergistically enhanced wound state shallow scanning eigenvector and one minus the threshold mean to obtain a weighted synergistically enhanced wound state shallow scanning eigenvector;

[0047] Calculate the product of the characteristic value of each position in the synergistically enhanced wound state deep scanning characteristic vector and the threshold mean value to obtain a weighted synergistically enhanced wound state deep scanning characteristic vector;

[0048] The weighted item collaboratively enhanced wound state shallow scan feature vector and the weighted item collaboratively enhanced wound state deep scan feature vector are added by position to obtain the wound state multi-scale fusion representation vector.

[0049] In the above-mentioned urology postoperative care management system, it also includes a training module for training the feature collaborative attention module, the feature multi-scale scanning module, the balanced threshold feature vector adaptive fusion module and the classifier-based monitoring module;

[0050] Wherein, the training module includes:

[0051] A training data acquisition unit, used to acquire training data, wherein the training data includes a training wound state image acquired by a camera;

[0052] A training wound state multi-scale feature extraction unit, used for inputting the training wound state image into a wound state multi-scale feature extractor based on a hollow pyramid network model to obtain a training wound state shallow feature map and a training wound state deep feature map;

[0053] A training feature collaborative attention unit, used for inputting the training wound state shallow feature map and the training wound state deep feature map into a feature collaborative attention module respectively to obtain a training collaboratively enhanced wound state shallow feature map and a training collaboratively enhanced wound state deep feature map;

[0054] A training multi-scale feature scanning and strengthening unit is used to input the training collaboratively enhanced wound state shallow feature map and the training collaboratively enhanced wound state deep feature map into a feature multi-scale scanning module to obtain a training collaboratively enhanced wound state shallow scanning feature vector and a training collaboratively enhanced wound state deep scanning feature vector;

[0055] A training wound state multi-scale fusion representation unit is used to input the training collaboratively enhanced wound state shallow scanning feature vector and the training collaboratively enhanced wound state deep scanning feature vector into an equalization threshold feature vector adaptive fusion module to obtain a training wound state multi-scale fusion representation vector;

[0056] An optimization unit, configured to optimize the training wound state multi-scale fusion representation vector to obtain the optimized training wound state multi-scale fusion representation vector;

[0057] A classification loss unit, used for inputting the optimized training wound state multi-scale fusion representation vector into a classifier-based monitoring module to obtain a classification loss function value;

[0058] A training unit is used to train the feature collaborative attention module, the feature multi-scale scanning module, the equalization threshold feature vector adaptive fusion module and the classifier-based monitoring module based on the classification loss function value.

[0059] According to another aspect of the present application, a method for postoperative care management in urology surgery is provided, comprising:

[0060] Manage patient information;

[0061] managing nursing tasks;

[0062] Manage medication information;

[0063] Monitor for complications;

[0064] publish patient education content;

[0065] Among other things, monitor for complications, including:

[0066] Acquire a wound status image captured by a camera;

[0067] Inputting the wound status image into a wound status multi-scale feature extractor based on a hollow pyramid network model to obtain a wound status shallow feature map and a wound status deep feature map;

[0068] Inputting the wound status shallow feature map and the wound status deep feature map into a feature collaborative attention module respectively to obtain a collaboratively enhanced wound status shallow feature map and a collaboratively enhanced wound status deep feature map;

[0069] Performing feature multi-scale scanning on the synergistically enhanced wound state shallow feature map and the synergistically enhanced wound state deep feature map to obtain a synergistically enhanced wound state shallow scanning feature vector and a synergistically enhanced wound state deep scanning feature vector;

[0070] The synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector are input into the equalization threshold feature vector adaptive fusion module to obtain a wound state multi-scale fusion representation vector; a wound infection detection unit is used to input the wound state multi-scale fusion representation vector into a classifier-based monitoring module to obtain a monitoring result, and the monitoring result is used to indicate whether the wound is infected.

[0071] Compared with the prior art, the present application provides a urology postoperative care management system and method, which uses a camera to collect wound status images, and introduces an image processing and analysis algorithm based on artificial intelligence and deep learning at the back end to analyze the wound status images, so as to capture the multi-level and multi-scale fusion features of the patient's wound status, thereby making a judgment on wound infection. In this way, it is possible to realize intelligent identification of wound infection in urology patients, thereby avoiding the defects caused by traditional reliance on medical staff for inspections, and can help medical staff monitor the patient's recovery process, manage the nursing plan for urology patients after surgery, and prevent wound infection problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0073] Figure 1 is a block diagram of a urology postoperative care management system according to an embodiment of the present application;

[0074] Figure 2 This is a system architecture diagram of a urology postoperative care management system according to an embodiment of the present application;

[0075] Figure 3 is a block diagram of a training module of a urology postoperative care management system according to an embodiment of the present application;

[0076] Figure 4 is a block diagram of a complication monitoring module in a postoperative care management system for urology according to an embodiment of the present application;

[0077] Figure 5 4 is a block diagram of a wound status multi-scale fusion representation unit in a urology postoperative care management system according to an embodiment of the present application;

[0078] Figure 6 The present invention is a flowchart of a method for postoperative care management of urology surgery according to an embodiment of the present application. DETAILED DESCRIPTION

[0079] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0080] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0081] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0082] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0083] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0084] Traditional nursing management requires medical staff to regularly visit patients, which consumes a lot of time and human resources. For a large number of patients or patients who require long-term monitoring, the traditional method is inefficient, monitoring is not timely, and the nursing plan is not personalized enough.

[0085] Therefore, an optimized urology postoperative care management system and method are desired.

[0086] In the technical solution of the present application, a urology postoperative care management system is proposed. Figure 1 It is a block diagram of a urology postoperative care management system according to an embodiment of the present application. Figure 2 FIG. 1 is a system architecture diagram of a urology postoperative care management system according to an embodiment of the present application. Figure 1 and Figure 2As shown, the urology postoperative care management system 300 according to an embodiment of the present application includes: a patient information management module 310 for managing patient information; a nursing task scheduling module 320 for managing nursing tasks; a drug management module 330 for managing drug information; a complication monitoring module 340 for monitoring complications; and a patient education module 350 for publishing patient education content.

[0087] In particular, the patient information management module 310 is used to manage patient information. The patient information management module is a vital component of the healthcare system and is used to collect, store, update and manage the patient's personal and medical information. The patient's appointment information, including date, time, doctor, clinic, etc., is recorded to effectively arrange the appointment. Through the patient information management module, medical institutions can better manage and utilize patient information, improve work efficiency, optimize the quality of medical services, and protect the security and privacy of patient information.

[0088] In particular, the nursing task scheduling module 320 and the drug management module 330 are used to manage nursing tasks and drug information respectively. Effective management of the nursing task scheduling module and the drug management module can improve the work efficiency of medical institutions, ensure that patients receive timely and correct care, and ensure the safety and effectiveness of drug management.

[0089] In particular, the complication monitoring module 340 is used to monitor complications. In particular, in a specific example of the present application, Figure 4As shown, the complication monitoring module 340 includes: a wound status image acquisition unit 341, which is used to obtain a wound status image acquired by a camera; a wound status multi-scale feature extraction unit 342, which is used to input the wound status image into a wound status multi-scale feature extractor based on a hollow pyramid network model to obtain a wound status shallow feature map and a wound status deep feature map; a feature collaborative attention unit 343, which is used to input the wound status shallow feature map and the wound status deep feature map into a feature collaborative attention module respectively to obtain a collaboratively enhanced wound status shallow feature map and a collaboratively enhanced wound status deep feature map; a multi-scale feature scanning and strengthening unit 344, which is used to scan and strengthen the collaboratively enhanced wound status. The wound state shallow feature map and the synergistically enhanced wound state deep feature map are subjected to feature multi-scale scanning to obtain a synergistically enhanced wound state shallow scanning feature vector and a synergistically enhanced wound state deep scanning feature vector; a wound state multi-scale fusion representation unit 345 is used to input the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector into an equalization threshold feature vector adaptive fusion module to obtain a wound state multi-scale fusion representation vector; a wound infection detection unit 346 is used to input the wound state multi-scale fusion representation vector into a classifier-based monitoring module to obtain a monitoring result, and the monitoring result is used to indicate whether the wound is infected.

[0090] Specifically, the wound state image acquisition unit 341 and the wound state multi-scale feature extraction unit 342 are used to obtain the wound state image collected by the camera; and the wound state image is input into the wound state multi-scale feature extractor based on the hollow pyramid network model to obtain the wound state shallow feature map and the wound state deep feature map. It should be understood that the wound state image may contain feature information of different scales and different levels about the patient's wound, such as the subtle texture of the wound edge, the overall color distribution, and the deep state semantics. Therefore, the wound state image is further input into the wound state multi-scale feature extractor based on the hollow pyramid network model to obtain the wound state shallow feature map and the wound state deep feature map. In particular, since the hollow convolution can increase the receptive field without increasing the number of parameters, it is very useful for extracting multi-scale features in the image. In wound image analysis, features of different scales may correspond to different wound healing stages or degrees of infection. Therefore, through the processing of the wound status multi-scale feature extractor based on the hollow pyramid network model, the wound status can be analyzed at different scales and levels, which helps to more comprehensively understand the wound status and situation and provide a basis for wound infection detection.

[0091] It is worth noting that the Pyramid Network is a deep neural network structure whose design is inspired by the structure of a pyramid, that is, the bottom is a larger feature map and the top is a smaller feature map. Pyramid networks are usually used to process computer vision tasks such as image classification, object detection, and semantic segmentation.

[0092] Specifically, the feature collaborative attention unit 343 is used to input the wound status shallow feature map and the wound status deep feature map into the feature collaborative attention module to obtain a collaboratively enhanced wound status shallow feature map and a collaboratively enhanced wound status deep feature map. Considering that the wound status shallow feature map and the wound status deep feature map respectively contain shallow features such as texture, edge and color of the patient's wound status and deep semantic feature information such as the morphology and structure of the wound, these features have different importance in performing wound status recognition and infection detection tasks. Therefore, in order to be able to respectively identify the important parts of the shallow features and deep semantic features of the wound status in the wound status image and strengthen these important parts, in the technical solution of the present application, the wound status shallow feature map and the wound status deep feature map are respectively input into the feature collaborative attention module to obtain a collaboratively enhanced wound status shallow feature map and a collaboratively enhanced wound status deep feature map. Through the processing of the feature collaborative attention module, the internal autocorrelation feature information in the shallow feature map of the wound status and the deep feature map of the wound status can be automatically learned and captured respectively, and the key features with greater relevance to the subsequent wound infection detection tasks can be highlighted by the feature collaboration and importance semantics, thereby improving the system's ability to grasp key information and helping to more accurately identify and understand the wound status.

[0093] In a specific example of the present application, the wound status shallow feature map and the wound status deep feature map are respectively input into the feature collaborative attention module to obtain a collaboratively enhanced wound status shallow feature map and a collaboratively enhanced wound status deep feature map, including: performing maximum pooling and upsampling processing based on the channel dimension on the wound status shallow feature map to obtain a wound status shallow collaborative attention pooling matrix; performing masking processing on the wound status shallow collaborative attention pooling matrix to obtain a masked wound status shallow collaborative attention pooling matrix; using the masked wound status shallow collaborative attention pooling matrix, weighting each feature matrix along the channel dimension in the wound status shallow feature map to obtain the collaboratively enhanced wound status shallow feature map.

[0094] In summary, in the above embodiment, the wound state shallow feature map and the wound state deep feature map are respectively input into the feature collaborative attention module to obtain the collaboratively enhanced wound state shallow feature map and the collaboratively enhanced wound state deep feature map, including: inputting the wound state shallow feature map into the feature collaborative attention module to process it with the following collaborative attention formula to obtain the collaboratively enhanced wound state shallow feature map; wherein the collaborative attention formula is:

[0095] ;

[0096] ;

[0097] ;

[0098] in, is the superficial characteristic map of the wound status, It is the first on the channel The eigenvalues ​​of the positions, Indicates that the maximum value pooling is performed on the feature map along the channel dimension. represents upsampling processing, The shallow collaborative attention pooling matrix is ​​used to focus on the eigenvalues ​​of each position in the wound status. To mask the wound status, the shallow layer collaboratively focuses on the eigenvalues ​​of each position in the pooling matrix. is the masked wound state shallow co-attention pooling matrix, In order to use the feature matrix to perform weighted multiplication calculations on each feature matrix along the channel dimension in the feature map, This is a superficial characteristic map of the synergistically enhanced wound state.

[0099] Specifically, the multi-scale feature scanning and strengthening unit 344 is used to perform feature multi-scale scanning on the synergistically enhanced wound state shallow feature map and the synergistically enhanced wound state deep feature map to obtain a synergistically enhanced wound state shallow scanning feature vector and a synergistically enhanced wound state deep scanning feature vector. That is, after the key features in the wound state shallow feature map and the wound state deep feature map are respectively enhanced to obtain the synergistically enhanced wound state shallow feature map and the synergistically enhanced wound state deep feature map, these key features related to the wound state have correlation relationships of different scales, such as subtle texture association and overall structure association. Therefore, in order to scan and analyze the enhanced wound state features at different levels at different scales in more detail, in the technical solution of the present application, the synergistically enhanced wound state shallow feature map and the synergistically enhanced wound state deep feature map are further input into the feature multi-scale scanning module to obtain the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector. Through the processing of the feature multi-scale scanning module, feature scanning based on different scales can be performed on the enhanced shallow feature map of the wound status and the deep feature map of the wound status, so as to more comprehensively capture the feature information in the image that is more closely related to the representation of the wound status. This integration of global and local information helps the system to understand the wound status more comprehensively, thereby improving the ability to identify and analyze the wound status.

[0100] In a specific example of the present application, a feature multi-scale scan is performed on the synergistically enhanced wound state shallow feature map and the synergistically enhanced wound state deep feature map to obtain a synergistically enhanced wound state shallow scanning feature vector and a synergistically enhanced wound state deep scanning feature vector, including: inputting the synergistically enhanced wound state shallow feature map and the synergistically enhanced wound state deep feature map into a feature multi-scale scanning module respectively to obtain the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector. More specifically: the synergistically enhanced shallow feature map of wound status is upsampled to obtain an upsampled shallow feature map of wound status; each feature matrix along the channel dimension in the upsampled shallow feature map of wound status is globally mean pooled to obtain an upsampled shallow channel feature vector of wound status; the upsampled shallow channel feature vector of wound status is point convolutionally encoded to obtain a shallow global feature vector of wound status; the upsampled shallow feature map of wound status is point convolutionally encoded to obtain a channel-modulated upsampled shallow feature map of wound status; the channel-modulated upsampled shallow feature map of wound status is two-dimensionally convolutionally encoded to obtain a shallow local feature vector of wound status; the shallow global feature vector of wound status and the shallow local feature vector of wound status are fused to obtain the synergistically enhanced shallow scanning feature vector of wound status.

[0101] Specifically, the wound state multi-scale fusion representation unit 345 is used to input the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector into the balanced threshold feature vector adaptive fusion module to obtain the wound state multi-scale fusion representation vector. It should be understood that since the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector respectively contain the shallow scanning features and deep semantic scanning features related to the wound state after the feature enhancement expression of different levels and scales, these features have their own advantages respectively, and there is a correlation relationship between these features, and this correlation relationship can better characterize the patient's wound state. Based on this, in order to be able to integrate the wound state features of different levels and scales to make up for the deficiencies of their respective feature expressions, so as to further improve the richness and ability of the wound state feature expression, so as to make the system more reliable in identifying and monitoring the wound state, in the technical solution of the present application, the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector are further input into the balanced threshold feature vector adaptive fusion module to obtain the wound state multi-scale fusion representation vector. Through the processing of the balanced threshold feature vector adaptive fusion module, the hidden correlation and importance semantics between the features of different levels and scales of the wound state after collaborative enhancement can be automatically learned and captured, so that these key features can be adaptively weighted and highlighted in the process of fusing the collaborative enhancement wound state shallow scanning feature vector and the collaborative enhancement wound state deep scanning feature vector, so as to improve the feature richness and expression ability, so as to obtain a more comprehensive and accurate wound state representation, and provide a more reliable basis for subsequent wound state analysis and detection. In particular, in a specific example of the present application, such as Figure 5As shown, the wound state multi-scale fusion representation unit 345 includes: a threshold constraint calculation subunit 3451, which is used to perform threshold constraints on the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector based on different linear transformations and fusion methods to obtain a first activated linear transformation wound state multi-scale associated feature vector, a second activated linear transformation wound state multi-scale associated feature vector and a third activated linear transformation wound state multi-scale associated feature vector; a wound state multi-scale feature constraint fusion representation subunit 3452, which is used to use each position feature value in the first activated linear transformation wound state multi-scale associated feature vector as a first threshold value, each position feature value in the second activated linear transformation wound state multi-scale associated feature vector as a second threshold value, and each position feature value in the third activated linear transformation wound state multi-scale associated feature vector as a third threshold value to calculate the weighted sum of the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector to obtain the wound state multi-scale fusion representation vector.

[0102] More specifically, the threshold constraint calculation subunit 3451 is used to perform threshold constraints on the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector based on different linear transformations and fusion methods to obtain a first activated linear transformation wound state multi-scale correlation feature vector, a second activated linear transformation wound state multi-scale correlation feature vector and a third activated linear transformation wound state multi-scale correlation feature vector. In a specific example of the present application, the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector are cascaded to obtain a first wound state multi-scale correlation feature vector; after calculating the matrix multiplication of the first wound state multi-scale correlation feature vector and the first transformation matrix, the obtained feature vector is added to the first bias vector by position to obtain a first linear transformation wound state multi-scale correlation feature vector; and the first linear transformation wound state multi-scale correlation feature vector is used. The activation function activates the first linear transformation wound state multi-scale correlation feature vector to obtain the first activated linear transformation wound state multi-scale correlation feature vector; the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector are added by position to obtain a second wound state multi-scale correlation feature vector; after calculating the matrix multiplication of the second wound state multi-scale correlation feature vector and the second transformation matrix, the obtained feature vector is added by position to the second bias vector to obtain a second linear transformation wound state multi-scale correlation feature vector; using The activation function activates the second linear transformation wound state multi-scale associated feature vector to obtain the second activated linear transformation wound state multi-scale associated feature vector; the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector are multiplied by position to obtain a third wound state multi-scale associated feature vector; after calculating the matrix multiplication of the third wound state multi-scale associated feature vector and the third transformation matrix, the obtained feature vector is added to the third bias vector by position to obtain the third linear transformation wound state multi-scale associated feature vector; using The activation function activates the third linearly transformed wound state multi-scale associated feature vector to obtain the third activated linearly transformed wound state multi-scale associated feature vector.

[0103] More specifically, the wound status multi-scale feature constrained fusion representation subunit 3452 is used to use the respective position feature values ​​in the first activated linear transformation wound status multi-scale associated feature vector as the first threshold value, the respective position feature values ​​in the second activated linear transformation wound status multi-scale associated feature vector as the second threshold value, and the respective position feature values ​​in the third activated linear transformation wound status multi-scale associated feature vector as the third threshold value, to calculate the weighted sum of the synergistically enhanced wound status shallow scanning feature vector and the synergistically enhanced wound status deep scanning feature vector to obtain the wound status multi-scale fusion representation vector. In a specific example of the present application, the mean among the first threshold value, the second threshold value and the third threshold value is calculated to obtain the threshold mean; the product of the eigenvalue of each position in the synergistically enhanced wound state shallow scanning feature vector and one minus the threshold mean is calculated to obtain a weighted synergistically enhanced wound state shallow scanning feature vector; the product of the eigenvalue of each position in the synergistically enhanced wound state deep scanning feature vector and the threshold mean is calculated to obtain a weighted synergistically enhanced wound state deep scanning feature vector; the weighted synergistically enhanced wound state shallow scanning feature vector and the weighted synergistically enhanced wound state deep scanning feature vector are added by position to obtain the wound state multi-scale fusion representation vector.

[0104] In summary, in the above embodiment, the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector are input into the balanced threshold feature vector adaptive fusion module to obtain the wound state multi-scale fusion representation vector, including: inputting the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector into the balanced threshold feature vector adaptive fusion module to process them with the following adaptive balanced fusion formula to obtain the wound state multi-scale fusion representation vector; wherein the adaptive balanced fusion formula is:

[0105] ;

[0106] in, and are respectively the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector, and They are respectively the shallow scanning feature vectors of the synergistically enhanced wound state and the synergistically enhanced wound state deep scanning feature vector The eigenvalues ​​at each position in , is the characteristic value of each position in the multi-scale fusion representation vector of the wound state, represents the concatenation of vectors, , and is the transformation matrix, , and is the bias vector, is the first activated linear transformation wound state multi-scale correlation feature vector, represents vector addition by position, It means vector multiplication by position. is the second activated linear transformation wound state multi-scale correlation feature vector, is the third activated linear transformation wound state multi-scale correlation feature vector, is the first activation linear transformation wound state multi-scale correlation feature vector The threshold value of each position in is the second activation linear transformation wound state multi-scale correlation feature vector The threshold value of each position in is the third activated linear transformation wound state multi-scale correlation feature vector The threshold value of each position in , express Activation function.

[0107] It is worth mentioning that in other specific examples of the present application, the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector can also be input into the balanced threshold feature vector adaptive fusion module in other ways to obtain a multi-scale fusion representation vector of the wound state, for example: input the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector; input the shallow scanning feature vector and the deep scanning feature vector into the balanced threshold feature vector module; the function of the balanced threshold feature vector module is to establish a balance and threshold between the feature vectors so as to better fuse the shallow and deep features; input the feature vector processed by the balanced threshold into the adaptive fusion module; the function of the adaptive fusion module is to dynamically adjust the fusion mode according to the input feature vector to adapt to different wound state characteristics; the adaptive fusion module fuses the shallow scanning feature vector and the deep scanning feature vector to obtain the multi-scale fusion representation vector of the wound state.

[0108] Specifically, the wound infection detection unit 346 is used to input the multi-scale fusion representation vector of the wound state into a monitoring module based on a classifier to obtain a monitoring result, and the monitoring result is used to indicate whether the wound is infected. That is, in the technical solution of the present application, the multi-scale fusion representation feature of the wound state is used for classification processing to determine the wound infection. In this way, it is possible to realize intelligent identification of wound infection in urology patients, thereby avoiding the defects caused by the traditional reliance on medical staff for inspections, and can help medical staff monitor the patient's recovery process, manage the nursing plan for urology patients after surgery, and prevent wound infection problems.

[0109] In particular, the patient education module 350 is used to publish patient education content. Through the patient education module, patients can better understand their diseases and treatment options, improve their self-management ability, and thus better participate in the treatment process and promote recovery and health.

[0110] It should be understood that before using the above neural network model for inference, it is necessary to train the feature collaborative attention module, the feature multi-scale scanning module, the balanced threshold feature vector adaptive fusion module and the classifier-based monitoring module. That is, according to the urology postoperative care management system 300 of the present application, it also includes a training phase for training the feature collaborative attention module, the feature multi-scale scanning module, the balanced threshold feature vector adaptive fusion module and the classifier-based monitoring module.

[0111] Figure 3 FIG. 1 is a block diagram of the training phase of the postoperative care management system for urology surgery according to an embodiment of the present application. Figure 3As shown, according to the embodiment of the present application, the urology postoperative care management system 300 includes: a training module 400, including: a training data acquisition unit 410, used to acquire training data, wherein the training data includes a training wound state image collected by a camera; a training wound state multi-scale feature extraction unit 420, used to input the training wound state image into a wound state multi-scale feature extractor based on a hollow pyramid network model to obtain a training wound state shallow feature map and a training wound state deep feature map; a training feature collaborative attention unit 430, used to input the training wound state shallow feature map and the training wound state deep feature map into a feature collaborative attention module respectively to obtain a training collaboratively enhanced wound state shallow feature map and a training collaboratively enhanced wound state deep feature map; a training multi-scale feature scanning and enhancement unit 440, used to input the training collaboratively enhanced wound state shallow feature map and the training collaboratively enhanced wound state deep feature map into a feature multi-scale scanning module respectively. To obtain a training collaboratively enhanced wound state shallow scanning feature vector and a training collaboratively enhanced wound state deep scanning feature vector; a training wound state multi-scale fusion representation unit 450 is used to input the training collaboratively enhanced wound state shallow scanning feature vector and the training collaboratively enhanced wound state deep scanning feature vector into the equalization threshold feature vector adaptive fusion module to obtain a training wound state multi-scale fusion representation vector; an optimization unit 460 is used to optimize the training wound state multi-scale fusion representation vector to obtain the optimized training wound state multi-scale fusion representation vector; a classification loss unit 470 is used to input the optimized training wound state multi-scale fusion representation vector into a classifier-based monitoring module to obtain a classification loss function value; a training unit 480 is used to train the feature collaborative attention module, the feature multi-scale scanning module, the equalization threshold feature vector adaptive fusion module and the classifier-based monitoring module based on the classification loss function value.

[0112] In one example, inputting the training wound state multi-scale fusion representation vector into a classifier-based monitoring module to obtain a monitoring result includes:

[0113] Multiplying the training wound state multi-scale fusion representation vector by the length of the training wound state multi-scale fusion representation vector and the square root of the length of the training wound state multi-scale fusion representation vector respectively to obtain a training wound state multi-scale fusion full-width representation vector and a wound state multi-scale fusion half-width representation vector;

[0114] Performing point subtraction on the training wound state multi-scale fusion full-width representation vector and the sum of the norms of the training wound state multi-scale fusion representation vector, and calculating the square root of the absolute value of each position of the point subtraction result vector to obtain the training wound state multi-scale fusion full-width semantic change vector;

[0115] Performing point subtraction on the training wound state multi-scale fusion half-width representation vector and the bi-norm of the wound state multi-scale fusion representation vector, and calculating the square root of the absolute value of each position of the point subtraction result vector to obtain the training wound state multi-scale fusion half-width semantic change vector;

[0116] Respectively calculating the logarithmic value with base 2 of each eigenvalue of the training wound state multi-scale fusion full-width semantic change vector and the training wound state multi-scale fusion half-width semantic change vector to obtain the training wound state multi-scale fusion full-width semantic change information vector and the training wound state multi-scale fusion half-width semantic change information vector;

[0117] Using the balanced hyperparameter as a weight, a weighted sum of the training wound state multi-scale fusion full-width semantic change information vector and the training wound state multi-scale fusion half-width semantic change information vector is calculated to obtain an optimized training wound state multi-scale fusion representation vector;

[0118] The optimized training wound state multi-scale fusion representation vector is input into a classifier-based monitoring module to obtain a monitoring result.

[0119] Here, for the cross-scale and cross-depth image semantic representation differences of the training wound status shallow feature map and the training wound status deep feature map based on the wound status image, after further enhancement through feature collaborative attention and feature multi-scale scanning, the training collaboratively enhanced wound status shallow scanning feature vector and the training collaboratively enhanced wound status deep scanning feature vector are input into the equalization threshold feature vector adaptive fusion module to obtain the training wound status multi-scale fusion representation vector. There will be a problem of insufficient aggregation of the adaptive fusion distribution of cross-scale and cross-depth image semantic features, thereby affecting the classification convergence efficiency of the classifier, that is, affecting the efficiency of classification training and the accuracy of monitoring results.

[0120] Based on this, the applicant of this application takes the training wound state multi-scale fusion representation vector as a feature set, wherein the change semantic representation is based on the feature value of the training wound state multi-scale fusion representation vector. In order to dynamically aggregate the semantic set composed of different change semantics of the training wound state multi-scale fusion representation vector without ignoring the individual semantic change information, the individual features of the training wound state multi-scale fusion representation vector and the aggregated scale of the training wound state multi-scale fusion representation vector are expressed as the feature full amplitude and half amplitude, and the low-rank negative correlation of different dimensions of the overall semantics of the feature set of the training wound state multi-scale fusion representation vector is used as the phase and scaled to dynamically adjust the change relationship of the semantic content of the training wound state multi-scale fusion representation vector, thereby improving the overall semantic information expression aggregation of the feature set of the training wound state multi-scale fusion representation vector, and improving the classification convergence efficiency of the training wound state multi-scale fusion representation vector when it is classified by the monitoring module based on the classifier. In this way, the wound infection of urology patients can be identified more accurately and intelligently, so as to help medical staff monitor the patient's recovery process, manage the nursing plan of urology patients after surgery, and prevent wound infection problems.

[0121] As described above, the urology postoperative care management system 300 according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with a urology postoperative care management algorithm. In a possible implementation, the urology postoperative care management system 300 according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the urology postoperative care management system 300 can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the urology postoperative care management system 300 can also be one of the many hardware modules of the wireless terminal.

[0122] Alternatively, in another example, the urology postoperative care management system 300 and the wireless terminal may also be separate devices, and the urology postoperative care management system 300 may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0123] Furthermore, a method for postoperative nursing management in urology surgery is also provided.

[0124] Figure 6 FIG. 1 is a flow chart of a method for postoperative nursing management of urology surgery according to an embodiment of the present application. Figure 6As shown, the urology postoperative care management method according to the embodiment of the present application includes the steps of: S1, managing patient information; S2, managing nursing tasks; S3, managing drug information; S4, monitoring complications; S5, publishing patient education content.

[0125] In summary, the postoperative nursing management method for urology surgery according to the embodiment of the present application is explained, which collects wound status images through a camera, and introduces an image processing and analysis algorithm based on artificial intelligence and deep learning at the back end to analyze the wound status image, so as to capture the multi-level and multi-scale fusion characteristics of the patient's wound status, thereby making a judgment on wound infection. In this way, it is possible to realize intelligent identification of wound infection in urology surgery patients, thereby avoiding the defects caused by traditional reliance on medical staff for inspections, and can help medical staff monitor the patient's recovery process, manage the nursing plan for urology surgery patients, and prevent wound infection problems.

[0126] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A postoperative nursing management system for urology surgery, characterized in that: include: Patient information management module, used to manage patient information; Nursing task scheduling module, used to manage nursing tasks; Drug management module, used to manage drug information; Complication monitoring module, used to monitor complications; Patient education module, used to publish patient education content; Wherein, the complication monitoring module includes: A wound status image acquisition unit, used to acquire a wound status image acquired by a camera; A wound status multi-scale feature extraction unit, used for inputting the wound status image into a wound status multi-scale feature extractor based on a hollow pyramid network model to obtain a wound status shallow feature map and a wound status deep feature map; A feature collaborative attention unit, used for inputting the wound state shallow feature map and the wound state deep feature map into a feature collaborative attention module to obtain a collaboratively enhanced wound state shallow feature map and a collaboratively enhanced wound state deep feature map; A multi-scale feature scanning and enhancement unit, used for performing feature multi-scale scanning on the synergistically enhanced wound state shallow feature map and the synergistically enhanced wound state deep feature map to obtain a synergistically enhanced wound state shallow scanning feature vector and a synergistically enhanced wound state deep scanning feature vector; A wound state multi-scale fusion representation unit, used for inputting the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector into a balanced threshold feature vector adaptive fusion module to obtain a wound state multi-scale fusion representation vector; A wound infection detection unit, used for inputting the wound state multi-scale fusion representation vector into a classifier-based monitoring module to obtain a monitoring result, wherein the monitoring result is used to indicate whether the wound is infected; Wherein, the feature collaborative attention unit is used to: Performing maximum pooling processing and upsampling processing based on channel dimension on the shallow feature map of wound status to obtain a shallow collaborative attention pooling matrix of wound status; Masking the wound state shallow collaborative attention pooling matrix to obtain a masked wound state shallow collaborative attention pooling matrix; Using the masked wound state shallow collaborative attention pooling matrix, weighting each feature matrix along the channel dimension in the wound state shallow feature map to obtain the collaboratively enhanced wound state shallow feature map; Wherein, the multi-scale feature scanning and strengthening unit is used to: Upsampling the synergistically enhanced wound state shallow feature map to obtain an upsampled wound state shallow feature map; Performing global mean pooling processing on each feature matrix along the channel dimension in the upsampled wound status shallow feature map to obtain an upsampled wound status shallow channel feature vector; Performing point convolution encoding on the upsampled wound status shallow channel feature vector to obtain a wound status shallow global feature vector; Performing point convolution encoding on the upsampled wound state shallow feature map to obtain a channel modulated upsampled wound state shallow feature map; Performing two-dimensional convolution encoding on the channel-modulated upsampled wound state shallow feature map to obtain a wound state shallow local feature vector; Fusion of the shallow global feature vector of the wound state and the shallow local feature vector of the wound state to obtain the synergistically enhanced shallow scanning feature vector of the wound state; Wherein, the wound status multi-scale fusion representation unit includes: A threshold constraint calculation subunit, used for performing threshold constraints on the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector based on different linear transformations and fusion methods to obtain a first activated linear transformation wound state multi-scale association feature vector, a second activated linear transformation wound state multi-scale association feature vector, and a third activated linear transformation wound state multi-scale association feature vector; A wound state multi-scale feature constraint fusion representation subunit, used to use each position feature value in the first activated linear transformation wound state multi-scale associated feature vector as a first threshold value, each position feature value in the second activated linear transformation wound state multi-scale associated feature vector as a second threshold value, and each position feature value in the third activated linear transformation wound state multi-scale associated feature vector as a third threshold value, to calculate the weighted sum of the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector to obtain the wound state multi-scale fusion representation vector; Wherein, the threshold constraint calculation subunit is used to: Cascading the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector to obtain a first wound state multi-scale correlation feature vector; After calculating the matrix multiplication of the first wound state multi-scale associated feature vector and the first transformation matrix, the obtained feature vector is added to the first bias vector by position to obtain a first linearly transformed wound state multi-scale associated feature vector; use An activation function activates the first linearly transformed wound state multi-scale associated feature vector to obtain the first activated linearly transformed wound state multi-scale associated feature vector; Adding the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector according to position to obtain a second wound state multi-scale correlation feature vector; After calculating the matrix multiplication of the second wound state multi-scale associated feature vector and the second transformation matrix, the obtained feature vector is added to the second bias vector by position to obtain a second linearly transformed wound state multi-scale associated feature vector; use An activation function activates the second linearly transformed wound state multi-scale associated feature vector to obtain the second activated linearly transformed wound state multi-scale associated feature vector; Multiplying the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector by position points to obtain a third wound state multi-scale correlation feature vector; After calculating the matrix multiplication of the third wound state multi-scale associated feature vector and the third transformation matrix, the obtained feature vector is added to the third bias vector by position to obtain a third linearly transformed wound state multi-scale associated feature vector; use An activation function activates the third linearly transformed wound state multi-scale associated feature vector to obtain the third activated linearly transformed wound state multi-scale associated feature vector; Wherein, the wound status multi-scale feature constraint fusion representation subunit is used for: Calculating the average of the first threshold value, the second threshold value and the third threshold value to obtain a threshold average; Calculate the product of the eigenvalue of each position in the synergistically enhanced wound state shallow scanning eigenvector and one minus the threshold mean to obtain a weighted synergistically enhanced wound state shallow scanning eigenvector; Calculate the product of the characteristic value of each position in the synergistically enhanced wound state deep scanning characteristic vector and the threshold mean value to obtain a weighted synergistically enhanced wound state deep scanning characteristic vector; The weighted item collaboratively enhanced wound state shallow scan feature vector and the weighted item collaboratively enhanced wound state deep scan feature vector are added by position to obtain the wound state multi-scale fusion representation vector.

2. The urology postoperative care management system according to claim 1, characterized in that: The multi-scale feature scanning and enhancement unit is used to: input the synergistically enhanced wound state shallow feature map and the synergistically enhanced wound state deep feature map into the feature multi-scale scanning module respectively to obtain the synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector.

3. The urology postoperative care management system according to claim 2, characterized in that: It also includes a training module for training the feature collaborative attention module, the feature multi-scale scanning module, the equalization threshold feature vector adaptive fusion module and the classifier-based monitoring module; Wherein, the training module includes: A training data acquisition unit, used to acquire training data, wherein the training data includes a training wound state image acquired by a camera; A training wound state multi-scale feature extraction unit, used for inputting the training wound state image into a wound state multi-scale feature extractor based on a hollow pyramid network model to obtain a training wound state shallow feature map and a training wound state deep feature map; A training feature collaborative attention unit, used for inputting the training wound state shallow feature map and the training wound state deep feature map into a feature collaborative attention module respectively to obtain a training collaboratively enhanced wound state shallow feature map and a training collaboratively enhanced wound state deep feature map; A training multi-scale feature scanning and strengthening unit is used to input the training collaboratively enhanced wound state shallow feature map and the training collaboratively enhanced wound state deep feature map into a feature multi-scale scanning module to obtain a training collaboratively enhanced wound state shallow scanning feature vector and a training collaboratively enhanced wound state deep scanning feature vector; A training wound state multi-scale fusion representation unit is used to input the training collaboratively enhanced wound state shallow scanning feature vector and the training collaboratively enhanced wound state deep scanning feature vector into an equalization threshold feature vector adaptive fusion module to obtain a training wound state multi-scale fusion representation vector; An optimization unit, configured to optimize the training wound state multi-scale fusion representation vector to obtain the optimized training wound state multi-scale fusion representation vector; A classification loss unit, used for inputting the optimized training wound state multi-scale fusion representation vector into a classifier-based monitoring module to obtain a classification loss function value; A training unit is used to train the feature collaborative attention module, the feature multi-scale scanning module, the equalization threshold feature vector adaptive fusion module and the classifier-based monitoring module based on the classification loss function value.

4. A method for postoperative nursing management of urology surgery, using the postoperative nursing management system of urology surgery according to claim 1, characterized in that: include: Manage patient information; managing nursing tasks; Manage medication information; Monitor for complications; publish patient education content; Among other things, monitor for complications, including: Acquire a wound status image captured by a camera; Inputting the wound status image into a wound status multi-scale feature extractor based on a hollow pyramid network model to obtain a wound status shallow feature map and a wound status deep feature map; Inputting the wound status shallow feature map and the wound status deep feature map into a feature collaborative attention module respectively to obtain a collaboratively enhanced wound status shallow feature map and a collaboratively enhanced wound status deep feature map; Performing feature multi-scale scanning on the synergistically enhanced wound state shallow feature map and the synergistically enhanced wound state deep feature map to obtain a synergistically enhanced wound state shallow scanning feature vector and a synergistically enhanced wound state deep scanning feature vector; The synergistically enhanced wound state shallow scanning feature vector and the synergistically enhanced wound state deep scanning feature vector are input into the equalization threshold feature vector adaptive fusion module to obtain a wound state multi-scale fusion representation vector; a wound infection detection unit is used to input the wound state multi-scale fusion representation vector into a classifier-based monitoring module to obtain a monitoring result, and the monitoring result is used to indicate whether the wound is infected.