Wound infection assessment method and system based on pH monitoring model

Abstract

The invention discloses a wound infection assessment method and system based on a pH monitoring model, and relates to the technical field of wound infection assessment, a pH data set and a wound feature set are extracted based on historical wound samples, data processing is performed on the pH data set to obtain a pH feature parameter set, the pH monitoring model is constructed according to the pH feature parameter set, and the wound infection assessment method and system based on the pH monitoring model are obtained. Carrying out model training on the pH monitoring model through the wound feature set; a to-be-detected wound sample is obtained, wound infection evaluation of the to-be-detected wound sample is carried out through the pH monitoring model after model training is completed, and then an infection preliminary indication of a wound corresponding to the current to-be-detected wound sample is obtained; the wounds are divided into different infection grades according to the initial infection indications, infection trend follow-up judgment is carried out on the wounds of the different infection grades in a set monitoring period, and then the infection evolution condition of the wounds is obtained.

Description

Technical Field

[0001] The present invention relates to the technical field of wound infection assessment, and in particular to a wound infection assessment method and system based on a pH monitoring model. Background Art

[0002] Wound infection is a common complication after surgery, which seriously affects the patient's recovery process. Traditional wound infection assessment methods mainly rely on the experience and judgment of clinicians and the results of laboratory bacterial culture, which are highly subjective and time-consuming. pH value is an important indicator of wound healing, and wound infection is often accompanied by changes in pH value. Therefore, it is of great significance to study a wound infection assessment method and system based on pH monitoring. Summary of the invention

[0003] In order to solve the above problems, the object of the present invention is to provide a wound infection assessment method and system based on a pH monitoring model.

[0004] The purpose of the present invention can be achieved by the following technical solution: A wound infection assessment method based on a pH monitoring model comprises the following steps:

[0005] Step S1: extracting a pH data set and a wound feature set based on historical wound samples, and performing data processing on the pH data set to obtain a pH feature parameter set, constructing a pH monitoring model according to the pH feature parameter set, and training the pH monitoring model through the wound feature set;

[0006] Step S2: obtaining a wound sample to be tested, and performing wound infection assessment on the wound sample to be tested by using the pH monitoring model after model training, thereby obtaining a preliminary infection indication of the wound corresponding to the current wound sample to be tested;

[0007] Step S3: Classify the wound into different infection levels according to the preliminary infection indications, and follow up and judge the infection trends of wounds of different infection levels during the set monitoring period, so as to obtain the infection evolution of the wound.

[0008] Furthermore, the process of extracting a pH data set and a wound feature set based on historical wound samples, and processing the pH data set to obtain a pH feature parameter set includes:

[0009] Obtain historical wound samples, which are composed of wound data at several historical time nodes. The wound data at each historical time node includes wound imaging images and wound biochemical indicators. Based on the wound biochemical indicators at each historical time node, corresponding pH data are extracted, and the pH data at all historical time nodes are obtained as the pH data set of the historical wound samples;

[0010] Perform image segmentation on the wound imaging image at each historical time node, and then segment each wound imaging image into several sub-region images, and perform image feature analysis on each sub-region image to obtain corresponding wound feature elements, summarize the wound feature elements of all sub-region images of the same wound imaging image as the corresponding feature element set, integrate the feature element sets of the wound imaging images at all historical time nodes, and then construct the wound feature set corresponding to the entire historical wound sample;

[0011] The data processing of the pH data set includes taking the mean of the pH data at several historical time nodes, calculating the variance, taking the extreme value, and calculating the discreteness between the pH value corresponding to the pH data at each historical time node and the preset standard pH reference value, which is then used as the corresponding pH feature parameter set.

[0012] Furthermore, the process of constructing a pH monitoring model according to the pH characteristic parameter set includes:

[0013] Construct a convolutional neural network model, and define a model input layer and a model output layer. The model input layer is used to input a pH feature parameter set. In the model input layer, the mean, variance, extreme value, and dispersion in the pH feature parameter set are used as input items that affect wound healing changes, and then the input item with the greatest impact on wound healing changes is obtained as the main influencing feature;

[0014] The model output layer is used to calculate the corresponding influence weight for each input item in the model input layer, where the influence weights of the mean, variance, extreme value and discreteness are respectively recorded as QZ1, QZ2, QZ3 and QZ4. The model output layer of the convolutional neural network model uses the main influencing features as model parameters, and converts the current convolutional neural network model into a pH monitoring model. The pH monitoring model outputs the pH value range corresponding to the model parameters, and the currently output pH value range is used as the ideal value range for the wound infection to be in a healing trend.

[0015] Furthermore, the process of training the pH monitoring model using the wound feature set includes:

[0016] A part of the wound feature set is selected as the training data of the pH monitoring model. The pH monitoring model is trained by the training data. Then, the pH monitoring model outputs the pH prediction range of each wound imaging image corresponding to the wound feature set, obtains the pH actual range corresponding to each wound imaging image, obtains the prediction success rate of the pH monitoring model, and records it as Sc. Then, Sc=M 成功 / M 总 , where M 成功 is the number of wound imaging images where the pH prediction range output by the pH monitoring model is the same as its corresponding pH actual range, M 总is the total number of pH prediction ranges output by the pH monitoring model;

[0017] The warning threshold Γ is set. When Sc≥Γ, the final pH monitoring model is successfully constructed. When Sc<Γ, the amount of training data is increased, and the pH monitoring model is continuously trained until Sc≥Γ is met and the model training is stopped.

[0018] Furthermore, the process of obtaining a wound sample to be tested, evaluating the wound infection of the wound sample to be tested by using the pH monitoring model after model training, and then obtaining a preliminary indication of infection of the wound corresponding to the current wound sample to be tested includes:

[0019] The wound sample to be tested is an analysis object that needs to be subsequently subjected to wound infection analysis;

[0020] The pH monitoring model after model training is used to analyze the pH value of the current wound sample to be tested, set a simulated environment for pH analysis, perform sample statistical analysis of the wound sample to be tested in the simulated environment, and then obtain the main sample characteristics of the current wound sample to be tested and input them into the pH monitoring model to obtain the pH value corresponding to the wound sample to be tested;

[0021] The preliminary infection indications include first-class indications and second-class indications. The first-class indications are wound assessment indications indicating that the wound corresponding to the wound sample to be tested is infected, and the second-class indications are wound assessment indications indicating that the wound corresponding to the wound sample to be tested is not infected.

[0022] The pH ranges corresponding to the first-class indications and the second-class indications are set and denoted as Ω1 and Ω2 respectively, and the pH value of the wound sample to be tested is denoted as pH-D; when pH-D∈Ω1, the preliminary indication of infection of the wound corresponding to the current wound sample to be tested is a first-class indication; when pH-D∈Ω2, the preliminary indication of infection of the wound corresponding to the current wound sample to be tested is a second-class indication.

[0023] Further, the process of classifying wounds into different infection grades based on initial signs of infection includes:

[0024] The infection level is divided into safe level, mild infection level, moderate infection level and severe infection level;

[0025] When the preliminary infection indication is a Class II indication, the infection level corresponding to the wound is classified as a safe level; when the preliminary infection indication is a Class I indication, a level judgment value 1 and a level judgment value 2 are set and recorded as Pd1 and Pd2 respectively, wherein 0<Pd1<Pd2, and the level judgment value 1 and the level judgment value 2 are used to classify the infection level of the wound as any one of a mild infection level, a moderate infection level and a severe infection level;

[0026] When 0<pH-D<Pd1, the infection level corresponding to the wound is classified as a mild infection level;

[0027] When Pd1≤pH-D<Pd2, the infection level corresponding to the wound is classified as a moderate infection level;

[0028] When pH-D ≥ Pd2, the infection level corresponding to the wound is classified as severe infection level.

[0029] Furthermore, the process of following up and judging the infection trend of wounds of different infection levels in a set monitoring period, and then obtaining the infection evolution of the wounds includes:

[0030] A monitoring period is set, where the monitoring period includes a number of monitoring time nodes, and the number of monitoring time nodes are numbered and recorded as j, where j=1, 2, 3, ..., m, where m is a natural number greater than 0;

[0031] The pH value of the wound at each monitoring time node is obtained through the pH monitoring model, and the wound pH value corresponding to the monitoring time node numbered j is recorded as pH-D[j], a two-dimensional rectangular coordinate system is established, the time values ​​of several monitoring time nodes are used as the horizontal coordinate, and the wound pH value corresponding to each monitoring time node is used as the vertical coordinate, and then the infection trend curve corresponding to the current wound is drawn. The infection trend curve is used to follow up and judge the infection trend of the current wound;

[0032] By following up the infection trend, the infection evolution of the current wound can be determined, including infection worsening, infection termination and infection improvement.

[0033] Furthermore, a wound infection assessment system based on a pH monitoring model comprises:

[0034] The pH monitoring model construction module extracts the pH data set and the wound feature set based on the historical wound samples, processes the pH data set, and then obtains the pH feature parameter set. The pH monitoring model is constructed according to the pH feature parameter set, and the pH monitoring model is trained through the wound feature set.

[0035] The wound infection preliminary assessment module is used to obtain the wound sample to be tested, and conduct wound infection assessment on the wound sample to be tested through the pH monitoring model after model training, and then obtain the preliminary infection indication of the wound corresponding to the current wound sample to be tested;

[0036] The wound infection detail follow-up module is used to classify wounds into different infection levels according to the initial signs of infection, and to follow up and judge the infection trends of wounds of different infection levels within the set monitoring period, so as to obtain the infection evolution of the wounds.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows: a pH data set and a wound feature set are extracted based on historical wound samples, a pH feature parameter set is obtained after data processing of the pH data set, a pH monitoring model is constructed according to the pH feature parameter set, and the pH monitoring model is trained through the wound feature set to obtain wound samples to be detected, and wound infection assessment of the wound samples to be detected is performed through the pH monitoring model after the model training is completed, so as to obtain preliminary infection indications of the wound corresponding to the current wound sample to be detected, and the wounds are divided into different infection levels according to the preliminary infection indications, and the infection trends of wounds of different infection levels are followed up and judged to obtain the infection evolution of the wound, which greatly reduces the subjectivity of human judgment on the severity of wound infection, speeds up the processing efficiency of wound infection assessment, and improves the accuracy of wound infection assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0039] like Figure 1 As shown, a wound infection assessment method based on a pH monitoring model comprises the following steps:

[0040] Step S1: extracting a pH data set and a wound feature set based on historical wound samples, and performing data processing on the pH data set to obtain a pH feature parameter set, constructing a pH monitoring model according to the pH feature parameter set, and training the pH monitoring model through the wound feature set;

[0041] Step S2: obtaining a wound sample to be tested, and performing wound infection assessment on the wound sample to be tested by using the pH monitoring model after model training, thereby obtaining a preliminary infection indication of the wound corresponding to the current wound sample to be tested;

[0042] Step S3: Classify the wound into different infection levels according to the preliminary infection indications, and follow up and judge the infection trends of wounds of different infection levels during the set monitoring period, so as to obtain the infection evolution of the wound.

[0043] It should be further explained that, in the specific implementation process, the process of extracting the pH data set and the wound feature set based on the historical wound samples, and processing the pH data set to obtain the pH feature parameter set includes:

[0044] Obtaining a historical wound sample, wherein the historical wound sample is composed of wound data at a plurality of historical time nodes, wherein the wound data at each historical time node includes a wound imaging diagram and a wound biochemical index, extracting corresponding pH data based on the wound biochemical index at each historical time node, and obtaining the pH data at all historical time nodes as a pH data set of the historical wound sample;

[0045] Perform image segmentation on the wound imaging image at each historical time node, and then segment each wound imaging image into several sub-region images of equal area, perform image feature analysis on each sub-region image, and then obtain the wound feature elements included in each sub-region image, summarize the wound feature elements of all sub-region images of the same wound imaging image, and then use them as the feature element set of the wound imaging image at the current historical time node, integrate the feature element sets of the wound imaging images at all historical time nodes, and then construct the wound feature set corresponding to the entire historical wound sample;

[0046] The data processing of the pH data set includes taking the mean of the pH data at several historical time nodes, calculating the variance, taking the extreme value, and calculating the discreteness between the pH value corresponding to the pH data at each historical time node and the preset standard pH reference value, and then using the obtained mean, variance, extreme value and discreteness as the pH feature parameter set.

[0047] It should be further explained that, in the specific implementation process, the process of constructing a pH monitoring model according to the pH characteristic parameter set and training the pH monitoring model through the wound characteristic set includes:

[0048] Constructing a convolutional neural network model, and defining a model input layer and a model output layer corresponding to the convolutional neural network model, wherein the model input layer is used to input a pH characteristic parameter set, and in the model input layer, the mean, variance, extreme value, and dispersion in the pH characteristic parameter set are respectively used as input items affecting wound healing changes, thereby obtaining the input item with the greatest degree of influence on wound healing changes, and using the input item as the main influencing feature, and the degree of influence is characterized by the influence weight output in the model output layer;

[0049] The model output layer is used to calculate the corresponding influence weight of each input item in the model input layer, wherein the influence weights of the mean, variance, extreme value and dispersion are recorded as QZ1, QZ2, QZ3 and QZ4 respectively;

[0050] The calculation formulas of QZ1, QZ2, QZ3 and QZ4 are as follows:

[0051] ;

[0052] ;

[0053] ;

[0054] ;

[0055] Among them, n1 represents the number of times the input item is the mean in the model input layer, n2 represents the number of times the input item is the variance in the model input layer, n3 represents the number of times the input item is the extreme value in the model input layer, and n4 represents the number of times the input item is the discreteness in the model input layer;

[0056] The model output layer of the convolutional neural network model uses the main influencing features as model parameters, and converts the current convolutional neural network model into a pH monitoring model. The pH monitoring model outputs the pH value range corresponding to the model parameters, and uses the currently output pH value range as the ideal value range for the wound infection to be in a healing trend.

[0057] A part of the wound feature set is selected as the training data of the pH monitoring model, and the amount of the training data is recorded as D 训练 The pH monitoring model is trained by using the training data, and then the pH monitoring model outputs the pH prediction range of each wound imaging image corresponding to the wound feature set, and obtains the pH actual range corresponding to each wound imaging image;

[0058] Obtain the prediction success rate of the pH monitoring model and record the prediction success rate as Sc, then Sc=M 成功 / M 总 , where M 成功 is the number of wound imaging images where the pH prediction range output by the pH monitoring model is the same as its corresponding pH actual range, M 总 is the total number of pH prediction ranges output by the pH monitoring model;

[0059] An early warning threshold Γ is set, and the early warning threshold is used to determine whether the current pH monitoring model is in the expected prediction state, that is, the prediction success rate of the pH monitoring model is at a qualified minimum limit value. When Sc≥Γ, the final pH monitoring model is successfully constructed. When Sc<Γ, the amount of training data is increased, and the model training of the pH monitoring model is continued until Sc≥Γ is satisfied, and the model training is stopped.

[0060] It should be further explained that, in the specific implementation process, the process of obtaining the wound sample to be tested, evaluating the wound infection of the wound sample to be tested by the pH monitoring model after the model training, and then obtaining the preliminary infection indication of the wound corresponding to the current wound sample to be tested includes:

[0061] The wound sample to be tested is an analysis object that needs to be subsequently subjected to wound infection analysis;

[0062] The pH monitoring model after the model training is used to analyze the pH value of the current wound sample to be detected, set a simulated environment for pH analysis, perform sample statistical analysis of the wound sample to be detected in the simulated environment, and then obtain the main sample characteristics of the current wound sample to be detected, and input the main sample characteristics into the pH monitoring model to obtain the pH value corresponding to the wound sample to be detected;

[0063] It should be noted that the main sample characteristics are the index items with the highest proportion among several biological indexes of the wound sample to be tested under the simulated environment;

[0064] The preliminary infection indications corresponding to the wound include a first-class indication and a second-class indication, wherein the first-class indication is an indication of wound assessment that the wound corresponding to the wound sample to be tested is infected, and the second-class indication is an indication of wound assessment that the wound corresponding to the wound sample to be tested is not infected;

[0065] A pH interval corresponding to the first type of indication and a pH interval corresponding to the second type of indication are set, and are respectively recorded as Ω1 and Ω2, and the pH value corresponding to the wound sample to be tested is recorded as pH-D, and the subordinate relationship between pH-D and Ω1 and Ω2 is determined, thereby obtaining a preliminary infection indication of the wound corresponding to the wound sample to be tested;

[0066] When pH-D∈Ω1, the preliminary infection indication of the wound corresponding to the current wound sample to be tested is a Class I indication;

[0067] When pH-D∈Ω2, the preliminary infection indication of the wound corresponding to the current wound sample to be tested is a Class II indication;

[0068] It should be noted that there are only two situations for defining the preliminary infection indications of the wound corresponding to the wound sample to be tested, namely, one is that the wound corresponding to the first type of indication is infected, and the other is that the wound corresponding to the second type of indication is not infected.

[0069] It should be further explained that, in the specific implementation process, the process of classifying wounds into different infection levels according to the initial signs of infection includes:

[0070] The infection level is divided into safe level, mild infection level, moderate infection level and severe infection level;

[0071] When the preliminary infection indication is a Class II indication, the infection level corresponding to the wound is classified as a safe level, which means that the current wound is not infected for the time being;

[0072] When the preliminary infection indication is a Class I indication, a level determination value 1 and a level determination value 2 are set, and the level determination value 1 and the level determination value 2 are respectively recorded as Pd1 and Pd2, wherein the numerical value is: 0<Pd1<Pd2, and the level determination value 1 and the level determination value 2 are used to specifically classify the infection level of the wound as any one of a mild infection level, a moderate infection level, and a severe infection level;

[0073] According to the relationship between pH-D, Pd1 and Pd2, the following classification is made:

[0074] When 0<pH-D<Pd1, the infection level corresponding to the wound is classified as a mild infection level;

[0075] When Pd1≤pH-D<Pd2, the infection level corresponding to the wound is classified as a moderate infection level;

[0076] When pH-D ≥ Pd2, the infection level corresponding to the wound is classified as severe infection level.

[0077] It should be further explained that, in the specific implementation process, the process of following up and judging the infection trend of wounds of different infection levels in the set monitoring period, and then obtaining the infection evolution of the wounds includes:

[0078] A monitoring period is set, wherein the monitoring period includes a plurality of monitoring time nodes, and the plurality of monitoring time nodes are numbered and the number is recorded as j, then j=1, 2, 3, ..., m, where m is a natural number greater than 0;

[0079] The pH value of the wound at each monitoring time node is obtained through the pH monitoring model, and the wound pH value corresponding to the monitoring time node numbered j is recorded as pH-D[j], a two-dimensional rectangular coordinate system is established, the time values ​​of several monitoring time nodes are used as the horizontal coordinate, and the wound pH value corresponding to each monitoring time node is used as the vertical coordinate, and then the infection trend curve corresponding to the current wound is drawn;

[0080] The infection trend curve is used to follow up and judge the infection trend of the current wound;

[0081] Obtain the infection evolution of the current wound by following up the infection trend, wherein the infection evolution includes infection deterioration, infection termination and infection improvement;

[0082] The specific infection trend follow-up judgment is as follows:

[0083] The curve portion corresponding to every two adjacent monitoring time nodes on the infection trend curve is taken as a sub-trend curve segment, and the middle point of the curve corresponding to each sub-trend curve segment in the horizontal coordinate direction is taken as the tangent point, the slope of the tangent line corresponding to the tangent point is obtained, and the slope of the tangent line is taken as the slope of the segmented curve corresponding to the current sub-trend curve segment;

[0084] Repeat the above operation to obtain the segmented curve slopes corresponding to all other sub-trend curve segments on the infection trend curve, accumulate the segmented curve slopes of all sub-trend curve segments included in the infection trend curve, and use the accumulated value as the final infection evolution judgment value of the current infection trend curve, and record the infection evolution judgment value as K;

[0085] When K < 0, the infection evolution of the current wound is marked as infection improvement;

[0086] When K = 0, the infection evolution of the current wound is marked as infection termination;

[0087] When K>0, the infection evolution of the current wound is marked as infection worsening;

[0088] It should be noted that the value of the segmented curve slope corresponding to each sub-trend curve segment includes positive values, negative values ​​and 0. When it is a positive value, it indicates that the corresponding sub-trend curve segment is in an upward trend on the infection trend curve. When it is a negative value, it indicates that the corresponding sub-trend curve segment is in a downward trend on the infection trend curve. When it is 0, it indicates that the sub-trend curve segment is in a stable trend on the infection trend curve. The segmented curve slopes of several sub-trend curve segments are accumulated to generate the final infection evolution judgment value. When the infection evolution judgment value is a positive number, it indicates that the current infection trend curve is in an upward trend as a whole, corresponding to the continued expansion of the wound infection. When it is a negative number, it indicates that the infection trend curve is in a downward trend as a whole, corresponding to the gradual reduction of the infection range of the wound and the improvement of the infection situation. When it is 0, the infection situation maintains the initial infection level of the wound.

[0089] The present invention also provides a wound infection assessment system based on a pH monitoring model, the system comprising:

[0090] The pH monitoring model construction module extracts the pH data set and the wound feature set based on the historical wound samples, processes the pH data set, and then obtains the pH feature parameter set. The pH monitoring model is constructed according to the pH feature parameter set, and the pH monitoring model is trained through the wound feature set.

[0091] The wound infection preliminary assessment module is used to obtain the wound sample to be tested, and conduct wound infection assessment on the wound sample to be tested through the pH monitoring model after model training, and then obtain the preliminary infection indication of the wound corresponding to the current wound sample to be tested;

[0092] The wound infection detail follow-up module is used to classify wounds into different infection levels according to the initial signs of infection, and to follow up and judge the infection trends of wounds of different infection levels within the set monitoring period, so as to obtain the infection evolution of the wounds.

[0093] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A wound infection assessment method based on a pH monitoring model, characterized in that: The following steps are involved: Step S1: extracting a pH data set and a wound feature set based on historical wound samples, and performing data processing on the pH data set to obtain a pH feature parameter set, constructing a pH monitoring model according to the pH feature parameter set, and training the pH monitoring model through the wound feature set; Step S2: obtaining a wound sample to be tested, and performing wound infection assessment on the wound sample to be tested by using the pH monitoring model after model training, thereby obtaining a preliminary infection indication of the wound corresponding to the current wound sample to be tested; Step S3: Classify the wound into different infection levels according to the preliminary infection indications, and follow up and judge the infection trends of wounds of different infection levels during the set monitoring period, so as to obtain the infection evolution of the wound.

2. A wound infection assessment method based on a pH monitoring model according to claim 1, characterized in that: The process of extracting a pH data set and a wound feature set based on historical wound samples, and processing the pH data set to obtain a pH feature parameter set includes: Obtain historical wound samples, which are composed of wound data at several historical time nodes. The wound data at each historical time node includes wound imaging images and wound biochemical indicators. Based on the wound biochemical indicators at each historical time node, corresponding pH data are extracted, and the pH data at all historical time nodes are obtained as the pH data set of the historical wound samples; Perform image segmentation on the wound imaging image at each historical time node, and then segment each wound imaging image into several sub-region images, and perform image feature analysis on each sub-region image to obtain corresponding wound feature elements, summarize the wound feature elements of all sub-region images of the same wound imaging image as the corresponding feature element set, integrate the feature element sets of the wound imaging images at all historical time nodes, and then construct the wound feature set corresponding to the entire historical wound sample; The data processing of the pH data set includes taking the mean of the pH data at several historical time nodes, calculating the variance, taking the extreme value, and calculating the discreteness between the pH value corresponding to the pH data at each historical time node and the preset standard pH reference value, which is then used as the corresponding pH feature parameter set.

3. A wound infection assessment method based on a pH monitoring model according to claim 2, characterized in that: The process of constructing a pH monitoring model based on the pH characteristic parameter set includes: Construct a convolutional neural network model, and define a model input layer and a model output layer. The model input layer is used to input a pH feature parameter set. In the model input layer, the mean, variance, extreme value, and dispersion in the pH feature parameter set are used as input items that affect wound healing changes, and then the input item with the greatest impact on wound healing changes is obtained as the main influencing feature; The model output layer is used to calculate the corresponding influence weight for each input item in the model input layer, where the influence weights of the mean, variance, extreme value and discreteness are respectively recorded as QZ1, QZ2, QZ3 and QZ4. The model output layer of the convolutional neural network model uses the main influencing features as model parameters, and converts the current convolutional neural network model into a pH monitoring model. The pH monitoring model outputs the pH value range corresponding to the model parameters, and the currently output pH value range is used as the ideal value range for the wound infection to be in a healing trend.

4. A wound infection assessment method based on a pH monitoring model according to claim 3, characterized in that: The process of model training the pH monitoring model using the wound feature set includes: A part of the wound feature set is selected as the training data of the pH monitoring model. The pH monitoring model is trained by the training data. Then, the pH monitoring model outputs the pH prediction range of each wound imaging image corresponding to the wound feature set, obtains the pH actual range corresponding to each wound imaging image, obtains the prediction success rate of the pH monitoring model, and records it as Sc. Then, Sc=M 成功 / M 总 , where M 成功 is the number of wound images in which the pH prediction range output by the pH monitoring model is the same as its corresponding pH actual range, M 总 is the total number of pH prediction ranges output by the pH monitoring model; The warning threshold Γ is set. When Sc≥Γ, the final pH monitoring model is successfully constructed. When Sc<Γ, the amount of training data is increased, and the pH monitoring model is continuously trained until Sc≥Γ is met and the model training is stopped.

5. A wound infection assessment method based on a pH monitoring model according to claim 4, characterized in that: The process of obtaining a wound sample to be tested, evaluating the wound infection of the wound sample to be tested by using the pH monitoring model after model training, and then obtaining the preliminary infection indication of the wound corresponding to the current wound sample to be tested includes: The wound sample to be tested is an analysis object that needs to be subsequently subjected to wound infection analysis; The pH monitoring model after model training is used to analyze the pH value of the current wound sample to be tested, set a simulated environment for pH analysis, perform sample statistical analysis of the wound sample to be tested in the simulated environment, and then obtain the main sample characteristics of the current wound sample to be tested and input them into the pH monitoring model to obtain the pH value corresponding to the wound sample to be tested; The preliminary infection indications include first-class indications and second-class indications. The first-class indications are wound assessment indications indicating that the wound corresponding to the wound sample to be tested is infected, and the second-class indications are wound assessment indications indicating that the wound corresponding to the wound sample to be tested is not infected. The pH ranges corresponding to the first-class indications and the second-class indications are set and denoted as Ω1 and Ω2 respectively, and the pH value of the wound sample to be tested is denoted as pH-D; when pH-D∈Ω1, the preliminary indication of infection of the wound corresponding to the current wound sample to be tested is a first-class indication; when pH-D∈Ω2, the preliminary indication of infection of the wound corresponding to the current wound sample to be tested is a second-class indication.

6. A wound infection assessment method based on a pH monitoring model according to claim 5, characterized in that: The process of classifying wounds into different infection grades based on initial signs of infection includes: The infection level is divided into safe level, mild infection level, moderate infection level and severe infection level; When the preliminary infection indication is a Class II indication, the infection level corresponding to the wound is classified as a safe level; when the preliminary infection indication is a Class I indication, a level judgment value 1 and a level judgment value 2 are set and recorded as Pd1 and Pd2 respectively, wherein 0<Pd1<Pd2, and the level judgment value 1 and the level judgment value 2 are used to classify the infection level of the wound as any one of a mild infection level, a moderate infection level and a severe infection level; When 0<pH-D<Pd1, the infection level corresponding to the wound is classified as a mild infection level; When Pd1≤pH-D<Pd2, the infection level corresponding to the wound is classified as a moderate infection level; When pH-D ≥ Pd2, the infection level corresponding to the wound is classified as severe infection level.

7. A wound infection assessment method based on a pH monitoring model according to claim 6, characterized in that: The process of following up and judging the infection trend of wounds of different infection levels within the set monitoring period and obtaining the infection evolution of the wounds includes: A monitoring period is set, where the monitoring period includes a number of monitoring time nodes, and the numbered monitoring time nodes are numbered and recorded as j, where j=1, 2, 3, ..., m, where m is a natural number greater than 0; The pH value of the wound at each monitoring time node is obtained through the pH monitoring model, and the wound pH value corresponding to the monitoring time node numbered j is recorded as pH-D[j], a two-dimensional rectangular coordinate system is established, the time values ​​of several monitoring time nodes are used as the horizontal coordinate, and the wound pH value corresponding to each monitoring time node is used as the vertical coordinate, and then the infection trend curve corresponding to the current wound is drawn. The infection trend curve is used to follow up and judge the infection trend of the current wound; By following up the infection trend, the infection evolution of the current wound can be determined, including infection worsening, infection termination and infection improvement.

8. A wound infection assessment system based on a pH monitoring model, used to implement the wound infection assessment method according to any one of claims 1 to 7, characterized in that: The system includes: The pH monitoring model construction module extracts the pH data set and the wound feature set based on the historical wound samples, processes the pH data set, and then obtains the pH feature parameter set. The pH monitoring model is constructed according to the pH feature parameter set, and the pH monitoring model is trained through the wound feature set. The wound infection preliminary assessment module is used to obtain the wound sample to be tested, and conduct wound infection assessment on the wound sample to be tested through the pH monitoring model after model training, and then obtain the preliminary infection indication of the wound corresponding to the current wound sample to be tested; The wound infection detail follow-up module is used to classify wounds into different infection levels according to the initial signs of infection, and to follow up and judge the infection trends of wounds of different infection levels within the set monitoring period, so as to obtain the infection evolution of the wounds.

Images