An ostomy peristomal skin ai evaluation method based on det evaluation scale

Abstract

The application relates to an ostomy peristomal skin AI evaluation method based on a DET evaluation scale. Based on the mode of the DET evaluation scale, ostomy peristomal skin images corresponding to D, E and T are collected and feature training is performed, so as to construct an ostomy peristomal skin AI model based on the DET evaluation scale, and the model is used for intelligent identification of ostomy skin. The DET features in the ostomy peristomal skin image can be intelligently identified and matched to output corresponding severity scores. The area score of the ostomy peristomal skin image is calculated based on contour recognition and calculus, and the state evaluation of the ostomy peristomal skin can be intelligently and automatically realized. The ostomy patient only needs to take ostomy peristomal skin images and upload the images to a server, so that objective state evaluation can be performed by using the system, and clinical manual evaluation by medical staff is not needed, thereby saving medical staff time and resources and releasing the evaluation burden of the medical staff. The evaluation personnel can be avoided from being affected by subjectivity, so that the evaluation risk caused by different evaluation levels can be avoided, and the evaluation is efficient.

Description

Technical Field

[0001] This disclosure relates to the field of medical assessment technology, and in particular to an AI assessment method, system and electronic device for periostomy skin based on the DET assessment scale. Background Technology

[0002] The DET scoring system addresses the practical problem that the management of peristaltic skin abnormalities often relies solely on clinical experience without a defined standard of practice. It scores the condition of the peristaltic skin across three dimensions: skin color changes, lesions / ulcers, and hyperplasia. It boasts advantages such as ease of use, mastery by assessors, and applicability to various stoma types. Through intuitive, direct, and standardized assessment, it provides nurses and patients with a simple method for obtaining information about the condition of the peristaltic skin.

[0003] As attached Figure 1 The DET assessment scale is shown below. The total DET score is calculated across three domains: Discoloration (D), Erosion (E), and Tissue overgrowth (T). Each domain is further considered in terms of the area of ​​the peristaltic skin affected (the skin covered by the stoma floorboard) and the severity. Area is scored from 0 to 3, and severity is scored accordingly.

[0004] (1) Area scoring. Peristomy skin is scored from 0 to 2 points. Unaffected skin is scored from 0 points; affected area <25% is scored from 1 point; affected area 25% to 50% is scored from 2 points; affected area >50% is scored from 3 points.

[0005] (2) Severity scoring. ① Discoloration (D): No discoloration of the skin around the stoma is scored as 0 points; discoloration is scored as 1 point; discoloration of the skin around the stoma accompanied by complications (pain, luster, hardening, fever, itching, or burning sensation, etc.) is scored as 2 points. ② Erosion (E): No maceration or ulceration is scored as 0 points; damage involving the epidermis is scored as 1 point; damage involving the dermis accompanied by complications (moisture, bleeding, or ulceration) is scored as 2 points. ③ Tissue hyperplasia (T): No tissue hyperplasia is scored as 0 points; raised tissue on the skin surface is scored as 1 point; raised tissue on the skin surface accompanied by complications (bleeding, pain, moisture) is scored as 2 points. After calculating the scores for each of the three areas according to the scoring criteria, the scores for the affected area and severity of the three areas are added together to obtain a composite score, namely the Discoloration, Erosion, and Tissue Hyperplasia (DET) score. This score ranges from 0 to 15. A score of 0 indicates that the skin around the stoma is healthy; as the score increases, the severity of the problem increases. A score is given at each visit, and changes in the score reflect the improvement or deterioration of the skin problem around the stoma.

[0006] The current clinical assessment of the periostomy skin's clinical development and recovery status (peristostomy skin condition) primarily relies on healthcare professionals evaluating ostomy patients using the aforementioned DET assessment scale. Therefore, this assessment method requires assessors to possess extensive clinical experience and objective assessment standards. However, assessors are often influenced by subjective factors, leading to inconsistent assessment levels and varying levels of professional competence and experience. This results in inconsistencies in the assessment results and reports for ostomy patients within the hospital.

[0007] In addition, clinical assessment requires a significant amount of time from medical staff, which disrupts their work and fails to conserve medical resources. Summary of the Invention

[0008] To address the aforementioned issues, this application proposes an AI assessment method, system, and electronic device for peristomal skin based on the DET assessment scale.

[0009] This application proposes an AI assessment method for peristomal skin based on the DET assessment scale, comprising the following steps:

[0010] The ostomy patient app collects images of the skin around the stoma of ostomy patients and uploads them to the backend server.

[0011] The backend server receives and forwards the image of the skin around the stoma to the image processing module. The image processing module performs image contour processing on the image of the skin around the stoma, evaluates and outputs the area score of the image of the skin around the stoma.

[0012] The backend server receives and forwards the peristomal skin image to a preset stoma skin AI assessment model, and identifies the DET features in the peristomal skin image through the stoma skin AI assessment model and outputs the corresponding severity score.

[0013] The area score and the severity score are written into the ostomy patient's medical record file in the HIS medical record system.

[0014] As an optional implementation of this application, the following steps may also be included:

[0015] The medical staff can log in to the HIS medical record system on the backend server via a PC and view the medical record files of the ostomy patient.

[0016] As an optional implementation of this application, the following steps may also be included:

[0017] The image of the skin around the stoma is stored in the stoma patient management system and bound to the stoma patient's medical ID.

[0018] As an optional embodiment of this application, the image processing module may optionally perform image contour processing on the peristomal skin image, evaluate and output the area score of the peristomal skin image, including:

[0019] Based on an image contour algorithm, the affected area on the skin image around the stoma is identified and marked;

[0020] The affected area is subjected to interpolation or approximation preprocessing;

[0021] The affected area S1 of the preprocessed affected region is calculated using calculus.

[0022] Calculate the area score δ of the skin image around the stoma:

[0023] δ=S1 / S0,

[0024] S0 is the original area of ​​the skin image around the stoma.

[0025] As an optional implementation of this application, the method for generating the stoma skin AI assessment model may include:

[0026] Images of stoma skin from several stoma patients were collected, wherein the stoma skin images of the stoma patients were collected separately according to the discoloration image, erosion image and tissue proliferation image of the DET assessment scale;

[0027] Image feature engineering is performed on the stoma skin image to extract image features from the stoma skin image and form a corresponding stoma image feature set, which includes:

[0028] Color-changing feature D(n),

[0029] Slow erosion characteristic E(n),

[0030] Tissue proliferation characteristics T(n);

[0031] The stoma image feature set is divided into a training set and a validation set according to a preset ratio;

[0032] The CNN model is trained using the training set, and the initial model generated by the training is validated using the validation set.

[0033] If the verification is successful, the stoma skin AI evaluation model is obtained, and the stoma skin AI evaluation model is deployed on the backend server and put into application.

[0034] In another aspect, this application proposes a system for implementing the aforementioned AI assessment method for peristoma skin based on the DET assessment scale, comprising:

[0035] The stoma patient app is used to collect images of the skin around the stoma of stoma patients and upload them to the backend server.

[0036] The backend server is configured to receive the peristomal skin image and store it in the stoma patient management system, binding it to the stoma patient's visit ID; forward the peristomal skin image to an image processing module, which performs image contour processing on the peristomal skin image, evaluates it, and outputs an area score; forward the peristomal skin image to a preset stoma skin AI assessment model, which identifies DET features in the peristomal skin image and matches and outputs a corresponding severity score; and write the area score and the severity score into the stoma patient's medical record file in the HIS medical record system.

[0037] The medical staff PC is used to log in to the HIS medical record system on the backend server and view the medical record files of the ostomy patient.

[0038] The ostomy patient app and the medical staff PC are respectively connected to the backend server.

[0039] In another aspect, this application also proposes an electronic device comprising:

[0040] processor;

[0041] Memory used to store processor-executable instructions;

[0042] The processor is configured to implement the AI ​​assessment method for peristoma skin based on the DET assessment scale when executing the executable instructions.

[0043] Technical effects of the present invention:

[0044] This invention, based on the DET assessment scale, collects images of the peristaltic skin corresponding to D, E, and T levels and performs feature training to construct an AI model of the peristaltic skin based on the DET assessment scale. This model is used for intelligent recognition of the peristaltic skin, intelligently identifying DET features in the images and matching and outputting corresponding severity scores. Furthermore, this invention uses contour recognition and calculus to calculate the area integral of the peristaltic skin images, thus enabling intelligent and automated assessment of the peristaltic skin condition. Ostomy patients only need to upload images of their peristaltic skin to the server to receive an objective condition assessment using this system, eliminating the need for manual clinical assessment by medical staff, saving time and resources, and reducing their assessment burden.

[0045] This invention can avoid the risk of inconsistent evaluation levels due to subjective influence on evaluators, thus making the evaluation more efficient.

[0046] Other features and aspects of this disclosure will become clear from the following detailed description of exemplary embodiments with reference to the accompanying drawings. Attached Figure Description

[0047] The accompanying drawings, which are included in and form part of this specification, illustrate exemplary embodiments, features, and aspects of this disclosure together with the specification and serve to explain the principles of this disclosure.

[0048] Figure 1 The DET assessment scale is shown below;

[0049] Figure 2 The diagram shown is a schematic representation of the implementation process of the present invention;

[0050] Figure 3 The diagram shown illustrates the application composition of the system of the present invention.

[0051] Figure 4 The diagram shown is a schematic representation of the model training process of this invention.

[0052] Figure 5 The diagram shows an application schematic of the electronic device of the present invention. Detailed Implementation

[0053] Various exemplary embodiments, features, and aspects of this disclosure will now be described in detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of the embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0054] The term “exemplary” as used herein means “serving as an example, embodiment, or illustration.” Any embodiment illustrated herein as “exemplary” is not necessarily to be construed as superior to or better than other embodiments.

[0055] Furthermore, to better illustrate this disclosure, numerous specific details are set forth in the following detailed description. Those skilled in the art will understand that this disclosure can be practiced without certain specific details. In some instances, means, components, and circuits well known to those skilled in the art have not been described in detail in order to highlight the main points of this disclosure.

[0056] Example 1

[0057] like Figure 2 As shown, this application proposes an AI assessment method for peristomal skin based on the DET assessment scale, comprising the following steps:

[0058] S1. Collect images of the skin around the stoma of stoma patients through the stoma patient APP and upload them to the backend server;

[0059] S2. The backend server receives and forwards the image of the skin around the stoma to the image processing module. The image processing module performs image contour processing on the image of the skin around the stoma, evaluates and outputs the area score of the image of the skin around the stoma.

[0060] S3. The backend server receives and forwards the peristomal skin image to the preset stoma skin AI assessment model, and identifies the DET features in the peristomal skin image through the stoma skin AI assessment model and outputs the corresponding severity score.

[0061] S4. Write the area score and the severity score into the ostomy patient's medical record file in the HIS medical record system.

[0062] Combined with appendix Figure 3 The application system shown is an AI assessment system for the skin around the stoma, which can intelligently and automatically assess the stoma status of stoma patients.

[0063] Patients can log in to the backend server via an app, upload images of their stoma, and the backend server will process and recognize the stoma images. Image contour processing algorithms can be used to identify the affected area in the skin image around the stoma and mark its contour. Combined with calculation tools such as calculus, the area of ​​the affected area can be further evaluated and scored.

[0064] After receiving the image, the backend server stores the image of the skin around the stoma in the stoma patient management system and binds it to the stoma patient's medical ID. This allows doctors to log in to the backend and view the stoma images of each patient, facilitating remote viewing and management of patients.

[0065] For assessing the severity of the condition, a pre-trained and deployed AI assessment model for stoma skin can be used to intelligently identify image features in images taken of the patient's stoma. Based on the DET image features learned in the pre-trained model, the model identifies DET features in the captured images and outputs a corresponding score indicating the severity of the stoma.

[0066] Therefore, the area and severity can be scored using the two modes described above, and the scoring results can then be written into the patient's medical record in the HIS system. Subsequently, medical staff can log in to the backend server via a doctor's PC (or PDA terminal, etc.) to view the stoma status assessment results for each stoma patient. This achieves intelligent and automated identification and assessment, avoiding subjective influence from assessors and objectively reducing the workload of medical staff.

[0067] In this embodiment, the communication configuration and connection between the APP or OC client and the backend server are managed and configured by the administrator, and this embodiment does not impose any limitations.

[0068] The scoring in step S2 is based on the principle of the image processing module as follows:

[0069] As an optional embodiment of this application, the image processing module may optionally perform image contour processing on the peristomal skin image, evaluate and output the area score of the peristomal skin image, including:

[0070] Based on an image contour algorithm, the affected area on the skin image around the stoma is identified and marked;

[0071] The affected area is subjected to interpolation or approximation preprocessing;

[0072] The affected area S1 of the preprocessed affected region is calculated using calculus.

[0073] Calculate the area score δ of the skin image around the stoma:

[0074] δ=S1 / S0,

[0075] S0 is the original area of ​​the skin image around the stoma.

[0076] The image processing module here primarily uses a contour recognition algorithm to process image contours. It intelligently identifies and marks the contours of affected areas in the skin image around the stoma. Then, the system uses a program to calculate the contour coordinates of the marked areas and performs preprocessing. Finally, it uses calculus code to integrate the marked contours to calculate the area of ​​the contour region.

[0077] The main steps of the procedure for performing calculus on a contour include the following:

[0078] I. Define the problem and objectives

[0079] First, it is necessary to clarify the purpose of the calculation, such as whether it is to find the length (perimeter), area, or other integral quantities related to the shape of the contour (such as the curvature of the contour, the integral of a certain physical quantity inside the contour, etc.).

[0080] II. Establishing a Mathematical Model

[0081] 1. Determining the mathematical description of the profile: Typically, a profile can be described by a series of point sets or parametric equations. If the profile consists of discrete points, it can be transformed into a continuous curve using interpolation methods (such as linear interpolation, spline interpolation, etc.).

[0082] 2. Choose an appropriate calculus method: Select a suitable calculus method based on the calculation objective. For example, if calculating the length (perimeter) of a contour, a definite integral can be used to calculate the length of the curve segment; if calculating the area of ​​a contour, a double integral can be used to calculate the area of ​​the region enclosed by the curve.

[0083] III. Writing program code

[0084] 1. Data Input: Write code to input the contour data, which may be point set coordinates, parameter values ​​of parametric equations, etc.

[0085] 2. Interpolation and approximation (if needed): If the profile is composed of discrete points, interpolation or approximation is required to transform it into a continuous function suitable for calculus calculations.

[0086] 3. Calculus Calculation: Based on the selected calculus method, write the corresponding code to calculate the required quantities. For example, use numerical integration methods (such as the trapezoidal rule, Simpson's method, etc.) to calculate the value of definite integrals.

[0087] 4. Output Results: Output the calculation results in an appropriate format, such as printing to the console, saving to a file, or displaying them on a graphical interface.

[0088] IV. Verification and Optimization

[0089] 1. Verification Results: Verify the correctness of the program by comparing the results with those of other methods (such as manual calculation, using mathematical software, etc.).

[0090] 2. Optimize the program: Based on the test results and actual needs, optimize the program code to improve computing efficiency or add new functions.

[0091] The original area can be pre-configured in the system. After the image processing module identifies the contour area, the original area can be substituted into the formula to calculate the area score of the skin image around the stoma. Here, area score refers to area percentage, i.e.:

[0092] Peristostomy skin score: 0-2. Unaffected skin: 0 points; Affected area <25%: 1 point; Affected area 25%-50%: 2 points; Affected area >50%: 3 points.

[0093] Combined with appendix Figure 4 As shown, as an optional embodiment of this application, the method for generating the stoma skin AI assessment model may include:

[0094] Images of stoma skin from several stoma patients were collected, wherein the stoma skin images of the stoma patients were collected separately according to the discoloration image, erosion image and tissue proliferation image of the DET assessment scale;

[0095] Image feature engineering is performed on the stoma skin image to extract image features from the stoma skin image and form a corresponding stoma image feature set, which includes:

[0096] Color-changing feature D(n),

[0097] Slow erosion characteristic E(n),

[0098] Tissue proliferation characteristics T(n);

[0099] The stoma image feature set is divided into a training set and a validation set according to a preset ratio;

[0100] The CNN model is trained using the training set, and the initial model generated by the training is validated using the validation set.

[0101] If the verification is successful, the stoma skin AI evaluation model is obtained, and the stoma skin AI evaluation model is deployed on the backend server and put into application.

[0102] The process of training image features using a CNN model to build a corresponding application model can be mainly divided into the following steps:

[0103] 1. Data Preparation

[0104] Loading the dataset: The first step is to load and preprocess the image dataset. The image dataset should contain enough labeled samples so that the network can learn effective features. This includes reading the image files and assigning labels.

[0105] Data preprocessing: To improve model performance and generalization ability, images are typically preprocessed, such as scaling, cropping, and normalization. In addition, data augmentation (such as rotation, flipping, and adding noise) is also a common technique to increase the diversity of training samples.

[0106] 2. Design the network structure

[0107] Selecting layer types: Design the CNN network structure according to task requirements, including the number, order, and parameter settings of convolutional layers, pooling layers, and fully connected layers.

[0108] Activation function: After the convolutional and fully connected layers, choose an appropriate activation function, such as ReLU or Sigmoid, to introduce non-linearity and enhance the expressive power of the model.

[0109] Normalization layer (optional): To accelerate the convergence speed of the model and improve its generalization ability, a normalization layer, such as batch normalization, can be added to the network.

[0110] 3. Configure training parameters

[0111] Optimizer: Select an appropriate optimizer (such as Adam, SGD, etc.) to update the network weights in order to minimize the loss function.

[0112] Loss function: Choose an appropriate loss function based on the task type (such as classification, regression, etc.), such as cross-entropy loss function, mean squared error loss function, etc.

[0113] Learning rate: Set an appropriate learning rate to control the step size of parameter updates. A learning rate that is too high may cause the model to fail to converge, while a learning rate that is too low may cause the training process to be too slow.

[0114] 4. Training the model

[0115] Forward propagation: The input image is fed into the CNN model and passed through convolutional layers, pooling layers, and fully connected layers one by one to finally obtain the output result.

[0116] Backpropagation: Calculate the loss value based on the output and the true label, and update the network parameters through the backpropagation algorithm to reduce the loss value.

[0117] Iterative training: Repeat the forward and backward propagation process until the preset number of training rounds is reached or the loss value no longer decreases significantly.

[0118] 5. Model Evaluation and Optimization

[0119] Model evaluation: Evaluate the performance of the trained CNN model using a validation or test set, including metrics such as accuracy, recall, and F1 score. Administrators are responsible for optimizing the training and validation.

[0120] Model tuning: Based on the evaluation results, the model is tuned, such as by adjusting the network structure, optimizer parameters, learning rate, etc., to improve the model's performance.

[0121] 6. Deployment and Application

[0122] The trained CNN model can be deployed to real-world applications, such as image classification, object detection, and face recognition.

[0123] The model will be further optimized and adjusted according to the actual application requirements to improve its stability and practicality.

[0124] By following the steps above, a CNN-based image feature learning model can be constructed and applied to various image recognition tasks.

[0125] In this embodiment, the stoma skin AI evaluation model is trained and constructed using a set of peristoma skin images corresponding to D, E, and T (discoloration, erosion, and tissue overgrowth) gradations. Peristaltic skin images of different types and degrees of D, E, and T gradations are all labeled with corresponding scores.

[0126] Therefore, during the training process, after collecting images of the peristoma skin with different D, E, and T (Discoloration, Erosion, and Tissueovergrowth) features, it is necessary to label the scores so that the model can learn the scores corresponding to different discoloration features D(n), erosion features E(n), and tissueovergrowth features T(n).

[0127] Therefore, the image feature engineering of this application extracts image features from the stoma skin image, including: discoloration feature D(n), slow erosion feature E(n), and tissue proliferation feature T(n). Image feature engineering can also be performed using convolutional neural networks (CNNs), which extract image features to obtain the corresponding image features. Specifically, this can be combined with the image feature extraction principles of CNNs.

[0128] Therefore, by using the above method, a model can be trained that can identify different discoloration features D(n), slow erosion features E(n), and tissue proliferation features T(n). After identifying the features, the corresponding scores can be output, which is convenient for system statistics.

[0129] Obviously, those skilled in the art should understand that implementing all or part of the processes in the above embodiments can be accomplished by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the control embodiments described above. Those skilled in the art will understand that implementing all or part of the processes in the above embodiments can be accomplished by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the control embodiments described above. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD), etc.; the storage medium can also include combinations of the above types of memory.

[0130] Example 2

[0131] Based on the implementation principle of Embodiment 1, this application, in another aspect, proposes a system for implementing the above-described AI assessment method for peristoma skin based on the DET assessment scale, comprising:

[0132] The stoma patient app is used to collect images of the skin around the stoma of stoma patients and upload them to the backend server.

[0133] The backend server is configured to receive the peristomal skin image and store it in the stoma patient management system, binding it to the stoma patient's visit ID; forward the peristomal skin image to an image processing module, which performs image contour processing on the peristomal skin image, evaluates it, and outputs an area score; forward the peristomal skin image to a preset stoma skin AI assessment model, which identifies DET features in the peristomal skin image and matches and outputs a corresponding severity score; and write the area score and the severity score into the stoma patient's medical record file in the HIS medical record system.

[0134] The medical staff PC is used to log in to the HIS medical record system on the backend server and view the medical record files of the ostomy patient.

[0135] The ostomy patient app and the medical staff PC are respectively connected to the backend server.

[0136] Please refer to Example 1 for a detailed understanding of the specific interactions of the above system.

[0137] The modules or steps of the present invention described above can be implemented using a general-purpose computing system. They can be centralized on a single computing system or distributed across a network of multiple computing systems. Optionally, they can be implemented using program code executable by a computing system, thereby storing them in a storage system for execution by the computing system, or fabricating them separately as individual integrated circuit modules, or fabricating multiple modules or steps into a single integrated circuit module. Thus, the present invention is not limited to any specific hardware and software combination.

[0138] Example 3

[0139] Combined with appendix Figure 5 As shown, further, in another aspect, this application also proposes an electronic device, comprising:

[0140] processor;

[0141] Memory used to store processor-executable instructions;

[0142] The processor is configured to implement the AI ​​assessment method for peristomal skin based on the DET assessment scale described in Embodiment 2 when executing the executable instructions.

[0143] The electronic device of this disclosure includes a processor and a memory for storing processor-executable instructions. The processor is configured to implement, when executing the executable instructions, a peristomal skin AI assessment method based on the DET assessment scale described in Embodiment 2 above.

[0144] It should be noted here that the number of processors can be one or more. Furthermore, the electronic device in this embodiment may also include an input system and an output system. The processor, memory, input system, and output system can be connected via a bus or other means, without specific limitations herein.

[0145] As a computer-readable storage medium, the memory can be used to store software programs, computer-executable programs, and various modules, such as the program or module corresponding to the AI ​​assessment method for peristaltic skin based on the DET assessment scale in this disclosure. The processor executes various functional applications and data processing of the electronic device by running the software programs or modules stored in the memory.

[0146] The input system can be used to receive input digital numbers or signals. These signals can be key signals related to user settings and function control of the device / terminal / server. The output system can include display devices such as screens.

[0147] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or technical improvements to the embodiments in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. An AI assessment method for peristomal skin based on the DET assessment scale, characterized in that, The process includes the following steps: Using a stoma patient app, images of the skin around the stoma are collected and uploaded to a backend server; the backend server receives and forwards these images to an image processing module; the image processing module performs image contour processing on the images, evaluates them, and outputs an area score; wherein the image contour processing and area scoring completely follow the original application's workflow: based on an image contour algorithm, the affected areas on the stoma skin images are identified and marked; the affected areas are preprocessed using interpolation or approximation; and calculus is used to calculate the affected area S1 of the preprocessed affected areas. The area score δ of the peristomal skin image is calculated as follows: δ = S1 / S0, where S0 is the original area of ​​the peristomal skin image. The entire process is automated, without manual touch selection or region segmentation, automatically completing region recognition and area quantification. It directly matches the DET scale's 0-3 point area grading standard to output the area score. The backend server receives and forwards the peristomal skin image to a preset stoma skin AI assessment model. This model identifies DET features in the peristomal skin image and outputs the corresponding severity score. The stoma skin AI assessment model is a dedicated DET scale CNN model, and the model generation method strictly follows the original application: collecting stoma skin images from several stoma patients, wherein the stoma skin images of the stoma patients are processed according to the DET scale. Images of discoloration, erosion, and tissue proliferation in the assessment scale were collected. Image feature engineering was performed on the stoma skin images to extract image features and form a corresponding stoma image feature set, which includes: discoloration feature D(n), erosion feature E(n), and tissue proliferation feature T(n). The stoma image feature set was divided into a training set and a validation set according to a preset ratio. The CNN model was trained using the training set, and the initial model generated by the training set was validated using the validation set. If the validation was successful, the stoma skin AI assessment model was obtained. The model takes skin images as end-to-end input and directly outputs three types of features (DET) with severity scores of 0-2 points each, without requiring secondary manual interpretation of feature levels by medical staff. The area score and the severity score were written into the stoma patient's medical record file in the HIS medical record system. The original application supporting process was executed simultaneously: the skin image around the stoma was stored in the stoma patient management system and bound to the stoma patient's visit ID. Medical staff PC You can log in to the backend server HIS medical record system to retrieve the complete patient ostomy medical record file with scores.

2. The AI ​​assessment method for peristomal skin based on the DET assessment scale according to claim 1, characterized in that, It also includes the following steps: The medical staff can log in to the HIS medical record system on the backend server via a PC and view the medical record files of the ostomy patient.

3. The AI ​​assessment method for peristomal skin based on the DET assessment scale according to claim 1, characterized in that, It also includes the following steps: The image of the skin around the stoma is stored in the stoma patient management system and bound to the stoma patient's medical ID.

4. A system for implementing the AI ​​assessment method for peristoma skin based on the DET assessment scale as described in any one of claims 1-3, characterized in that, include: The stoma patient app is used to collect images of the skin around the stoma of stoma patients and upload them to the backend server. The backend server is used to receive the images of the skin around the stoma and store them in the stoma patient management system, and bind them to the stoma patient's medical ID; The system forwards the image of the skin around the stoma to an image processing module, which performs image contour processing on the image, evaluates it, and outputs an area score for the image; it also forwards the image to a preset stoma skin AI assessment model, which identifies DET features in the image and matches and outputs a corresponding severity score; and it writes the area score and the severity score into the stoma patient's medical record file in the HIS medical record system. The medical staff PC is used to log in to the HIS medical record system on the backend server and view the medical record files of the ostomy patient. The ostomy patient app and the medical staff PC are respectively connected to the backend server.

5. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to implement, when executing the executable instructions, an AI assessment method for peristomal skin based on the DET assessment scale according to any one of claims 1-3.

Images