CN116580282A - Neural network model-based pressure injury staged identification system and storage medium - Google Patents
Neural network model-based pressure injury staged identification system and storage medium Download PDFInfo
- Publication number
- CN116580282A CN116580282A CN202310853766.3A CN202310853766A CN116580282A CN 116580282 A CN116580282 A CN 116580282A CN 202310853766 A CN202310853766 A CN 202310853766A CN 116580282 A CN116580282 A CN 116580282A
- Authority
- CN
- China
- Prior art keywords
- network
- neural network
- pressure injury
- feature extraction
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 208000027418 Wounds and injury Diseases 0.000 title claims abstract description 40
- 230000006378 damage Effects 0.000 title claims abstract description 36
- 208000014674 injury Diseases 0.000 title claims abstract description 36
- 238000003062 neural network model Methods 0.000 title claims abstract description 22
- 238000000605 extraction Methods 0.000 claims abstract description 25
- 206010011985 Decubitus ulcer Diseases 0.000 claims abstract description 19
- 208000004210 Pressure Ulcer Diseases 0.000 claims abstract description 19
- 230000036760 body temperature Effects 0.000 claims abstract description 12
- 230000000474 nursing effect Effects 0.000 claims abstract description 12
- 238000013527 convolutional neural network Methods 0.000 claims description 21
- 238000000034 method Methods 0.000 claims description 21
- 238000013528 artificial neural network Methods 0.000 claims description 18
- 230000009466 transformation Effects 0.000 claims description 15
- 239000013598 vector Substances 0.000 claims description 12
- 230000008447 perception Effects 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 4
- 238000013501 data transformation Methods 0.000 claims description 3
- 230000000306 recurrent effect Effects 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims description 2
- 238000003745 diagnosis Methods 0.000 abstract description 2
- 238000013145 classification model Methods 0.000 description 12
- 230000008569 process Effects 0.000 description 6
- 206010052428 Wound Diseases 0.000 description 4
- 125000004122 cyclic group Chemical group 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000004393 prognosis Methods 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 210000000988 bone and bone Anatomy 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001575 pathological effect Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 210000001519 tissue Anatomy 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/03—Recognition of patterns in medical or anatomical images
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to the technical field of medical diagnosis, and particularly relates to a pressure injury stage identification system and a storage medium based on a neural network model. The system of the invention comprises: the input module is used for inputting pressure sore images, nursing records and body temperature data; the feature extraction module is used for extracting features in the pressure sore image by adopting an image feature extraction network, extracting features in the nursing record by adopting a text feature extraction network, extracting features in the body temperature data by adopting a time sequence feature extraction network, and fusing the features to obtain multi-mode features; and the multi-layer sensing network module is used for inputting the multi-mode characteristics into the multi-layer sensing network to obtain a pressure injury stage identification result. The system has good prediction performance and good application prospect.
Description
Technical Field
The invention belongs to the technical field of medical diagnosis, and particularly relates to a pressure injury stage identification system and a storage medium based on a neural network model.
Background
Pressure injury is a local injury to skin and/or affected tissues due to pressure or pressure combined with shear force, often occurs at a bone protrusion, and can also be related to medical instruments or other things, and the prevention and treatment of pressure injury reflects the quality of clinical care. Pressure injury may occur in various departments, and since the staging and identification of pressure injury is mainly performed by specialized trained wound nurses, it is often difficult for general clinical nurses to correctly identify and stage pressure injury, which can delay subsequent wound treatment and affect patient prognosis outcome. Therefore, intelligent systems are developed to assist clinical nurses in correctly staging pressure injury, facilitating subsequent wound treatment, improving patient prognosis outcome.
Currently, the prior art has a few common sense aiming at intelligent recognition of pressure injury, for example, patent application CN202011022448.5 pressure sore picture automatic analysis system, method, equipment and medium, and discloses a method and a system for analyzing pressure sore pictures by using a machine learning model. However, in the prior art, the pressure sore is identified in stages through a neural network model, and training and reasoning are often performed only through a single mode of the pressure sore picture, and the mode is influenced by quality problems of the picture such as picture pixels, contrast, white balance and the like, so that the accuracy of the model is difficult to improve. This has an adverse effect on the application of artificial intelligence in pressure injury analysis.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a pressure injury stage identification system and a storage medium based on a neural network model, which aim to: pressure sore staged identification and classification convolutional neural network model based on multi-mode fusion, and accuracy and effectiveness of the pressure sore identification model are improved by fusing wound images, nursing pathological records, body temperature and other multi-mode data.
A neural network model-based pressure injury staging recognition system, comprising:
the input module is used for inputting pressure sore images, nursing records and body temperature data;
the feature extraction module is used for extracting features in the pressure sore image by adopting an image feature extraction network, extracting features in the nursing record by adopting a text feature extraction network, extracting features in the body temperature data by adopting a time sequence feature extraction network, and fusing the features to obtain multi-mode features;
and the multi-layer sensing network module is used for inputting the multi-mode characteristics into the multi-layer sensing network to obtain a pressure injury stage identification result.
Preferably, the image feature extraction network is selected from convolutional neural networks.
Preferably, the text feature extraction network is selected from NLP models.
Preferably, the timing feature extraction network is selected from a recurrent neural network.
Preferably, the multi-layer perception network comprises a convolutional neural network, a residual network, a pooling network, a multi-layer perceptron, a cyclic neural network and an attention network.
Preferably, the multi-layer sensing network comprises a plurality of classified convolutional neural network models and a multi-outcome voting module; and each classification convolutional neural network model obtains a prediction result of the pressure injury stage, and the multi-result voting module determines the final prediction by adopting a voting method.
Preferably, the number of network layers and/or the network type of the plurality of classified convolutional neural network models have differences, and the differences are at least one of the following network configuration parameters: the number of layers of the neural network, the number of network elements of each layer, the type of pooling network, the Dropout probability, the convolution kernel size, the convolution step size and whether residual connection is performed or not.
Preferably, the plurality of classified convolutional neural network models are respectively obtained by training multiple data sets constructed by adopting one original data set, and the construction method of the multiple data sets comprises the following steps:
step 1, for each data set to be generated, firstly generating a random vector #p 1, p 2, p 3, p 4, p 5, v 1, v 2, v 3 )Wherein each parameter in the random vector is derived from uniformly distributed random samples,，，/>，/>，/>，/>，，/>；
step 2, carrying out data transformation on the original pictures in the original data set according to the random vectors generated in the step 1 to obtain randomly transformed pictures, wherein the random vectors arep i For controlling whether data enhancement is performed, each original picture transformation step being performed under the condition thatp i >0.5；v i For controlling the degree of transformation, wherev 1 For controlling the contrast of the contrast transformation,v 2 for controlling the angle of the random rotation,v 3 a scaling factor for controlling the scaling transformation;p i is thatp 1 、p 2 、p 3 、p 4 Or (b)p 5 ，v i Is thatv 1 、v 2 Or (b)v 3 ；
Step 3, repeating the step 1 and the step 2N times to obtain a multi-data set consisting of N data sets.
Preferably, the model training module is further included for model training of the multi-layer perception network.
The present invention also provides a computer-readable storage medium having stored thereon: a computer program for implementing the neural network model-based pressure injury stage identification system.
Aiming at the task of pressure injury stage identification, the invention provides a prediction system, and the system takes data such as pressure sore images, nursing records, body temperature data and the like as input to construct a system for extracting characteristics of multi-mode data and predicting results by adopting a multi-mode model and a multi-result voting method. The application of the multi-modal model and the multi-outcome voting method enables the prediction result of the invention to be more accurate.
In a preferred scheme, the method changes the original data set in the model training stage to obtain training data of multiple data sets, and can train to obtain a multi-mode model formed by more convolution neural network models, so that the pluralism of data characteristics is improved on the premise of reducing the requirement on the marked data quantity, and the prediction performance of the model is further improved.
Therefore, the invention has good application prospect.
It should be apparent that, in light of the foregoing, various modifications, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.
The above-described aspects of the present invention will be described in further detail below with reference to specific embodiments in the form of examples. It should not be understood that the scope of the above subject matter of the present invention is limited to the following examples only. All techniques implemented based on the above description of the invention are within the scope of the invention.
Drawings
FIG. 1 is a schematic flow chart of embodiment 1 of the present invention;
FIG. 2 is a flowchart of the multi-dataset generation step of embodiment 1 of the present invention;
FIG. 3 is a schematic structural diagram of a multi-modal model in embodiment 1 of the present invention;
FIG. 4 is a flowchart of the multi-model prediction step in embodiment 1 of the present invention;
fig. 5 is a schematic flow chart of the multi-outcome voting in example 1 of the present invention.
Detailed Description
It should be noted that, in the embodiments, algorithms of steps such as data acquisition, transmission, storage, and processing, which are not specifically described, and hardware structures, circuit connections, and the like, which are not specifically described may be implemented through the disclosure of the prior art.
Example 1 pressure injury stage identification System based on neural network model
The system of the present embodiment includes:
the input module is used for inputting pressure sore images, nursing records and body temperature data;
the feature extraction module is used for extracting features in the pressure sore image by adopting an image feature extraction network, extracting features in the nursing record by adopting a text feature extraction network, extracting features in the body temperature data by adopting a time sequence feature extraction network, and fusing the features to obtain multi-mode features;
the multi-layer sensing network module is used for inputting the multi-mode characteristics into a multi-layer sensing network to obtain a pressure injury stage identification result;
and the model training module is used for carrying out model training on the multi-layer perception network.
The method for identifying the pressure injury stage by adopting the system of the embodiment is shown in figure 1, and comprises two stages which are respectively
Training phase: the training stage is used for establishing a plurality of neural network classification models based on multi-mode input, generating a plurality of data sets by using the original data marked by experts in a data enhancement mode, and training each neural network classification model. The main process of the training stage comprises multi-data set generation, multi-model construction and multi-model training.
Reasoning: the reasoning stage is to input the original data to be predicted (pressure sore image, nursing record and body temperature data) into a plurality of neural network classification models obtained in the training stage for reasoning, and synthesize the prediction results of the models through a voting method to obtain a final prediction result. The main process of the reasoning stage comprises multi-model prediction and multi-result voting to obtain the final classification result.
The method comprises the following steps:
1. multi-dataset generation
The purposes of multi-dataset generation include two: firstly, expert annotation data are enhanced, and the scale of a training data set is expanded; secondly, a plurality of training data sets are formed to train a plurality of classification models by controlling different data enhancement parameters. The main process is as shown in fig. 2, including:
step 1, for each data set to be generated, firstly generating a random vector #p 1, p 2, p 3, p 4, p 5, v 1, v 2, v 3 )Wherein each parameter in the random vector is derived from uniformly distributed random samples,，，/>，/>，/>，/>，，/>；
step 2, carrying out data transformation on the original pictures in the original data set according to the random vectors generated in the step 1 to obtain randomly transformed pictures, wherein the random vectors arep i For controlling whether data enhancement is performed, each original picture transformation step being performed under the condition thatp i >0.5（p i Is a scalar. For the firstiThe number of the transformations is chosen,p i greater than 0.5, the change is performed ifp i 0.5 or less, the transformation is not performed);v i for controlling the degree of transformation, wherev i For controlling the degree of the transformation,v 1 for controlling the contrast of the contrast transformation,v 2 for controlling the angle of the random rotation,v 3 a scaling factor for controlling the scaling transformation;p i is thatp 1 、p 2 、p 3 、p 4 Or (b)p 5 ，v i Is thatv 1 、v 2 Or (b)v 3 ；
Step 3, repeating the step 1 and the step 2N times to obtain a multi-data set consisting of N data sets.
2. Multimodal model construction
As shown in fig. 3, a multi-classification neural network model based on multi-modal data input such as pressure sore images, care records, and body temperature data is constructed. For different types of data, different feature extraction methods are used, a convolutional neural network is used for extracting image features, an NLP model is used for extracting text features, a cyclic neural network is used for extracting time sequence features, and the different types of features are fused to construct a comprehensive multi-modal feature representation. And finally, inputting the multi-mode features into a multi-layer sensing network, adding an attention network into the image features, the text features and the time sequence features, and finally obtaining a classification result.
3. Multi-model training
Initializing N convolutional neural network classification models for N data sets, wherein the training process for the N convolutional neural network classification models is as follows:
step a, generating original data marked by expert according to the method of the section of generating multiple data setsNA plurality of training data sets;
step b, constructingNAnd a convolutional neural network classification model. For the firstiIndividual training data setsD i Constructing a convolutional neural network classification modelNet i . On the selection of network structure, the difference of network layer number and network type is ensured to be the firstiThe difference between the convolutional neural network and other convolutional neural networks is obvious. Different convolutional neural network models can be obtained by setting network configuration parameters such as the layer number of different neural networks, the number of network elements of each layer, the pooling network type, the Dropout probability, the convolution kernel size, the convolution step length, whether residual connection is or not and the like.
Step c, training each neural network classification model to finally obtainNAnd (3) training a convolutional neural network.
Furthermore, optionally: the care record is text data, and as one of task inputs, key text features are extracted through a text feature extraction network (such as a recurrent neural network or a transducer network); the data can be used as optional data, so that the accuracy of the image recognition network can be improved;
the temperature data is time sequence data, which is used as one of task inputs, and the key time sequence characteristics are extracted through a time sequence characteristic extraction network (a cyclic neural network or a transducer network is generally used for extracting the key time sequence characteristics, and the data can be used as optional data, so that the accuracy of the image recognition network can be improved.
4. Multi-model prediction
The multi-model predictions are obtained using a model training phaseNThe individual neural network classification models respectively predict pictures to be identified (nursing records and temperature data are input as optional data sources) to obtainNAnd predicting the result. The process is shown in fig. 4.
5. Multiple outcome voting
In order to improve the accuracy of the classification prediction result, the final classification result is determined by a voting method. The process is shown in fig. 5. Wherein the firstiPersonal modelNet i The prediction result of (2) isWherein m is the number of categories of pressure sore image classification, < >>Represents the firstiPersonal modelNet i Predicting pressure sore images as categoriesjProbability size of (2), soThen (1)iThe category of the individual model predictions isMaxIDX(Pi)I.e. the lower index value with the highest probability in the predicted result.
And determining the final prediction type by utilizing a voting method according to the predicted categories of the N neural network classification models. And counting the prediction types of the neural network classification models, wherein the type with the most prediction types is the final prediction type, and the final prediction type is determined according to a random selection mode under the condition that the most types are the same.
According to the embodiment, the method and the device construct the pressure injury stage identification based on the multi-mode fusion neural network model, have good prediction performance and have good application prospect.
Claims (10)
1. A neural network model-based pressure injury staging recognition system, comprising:
the input module is used for inputting pressure sore images, nursing records and body temperature data;
the feature extraction module is used for extracting features in the pressure sore image by adopting an image feature extraction network, extracting features in the nursing record by adopting a text feature extraction network, extracting features in the body temperature data by adopting a time sequence feature extraction network, and fusing the features to obtain multi-mode features;
and the multi-layer sensing network module is used for inputting the multi-mode characteristics into the multi-layer sensing network to obtain a pressure injury stage identification result.
2. The neural network model-based pressure injury staging recognition system of claim 1, wherein:
the image feature extraction network is selected from convolutional neural networks.
3. The neural network model-based pressure injury staging recognition system of claim 1, wherein:
the text feature extraction network is selected from an NLP model.
4. The neural network model-based pressure injury staging recognition system of claim 1, wherein:
the timing feature extraction network is selected from a recurrent neural network.
5. The neural network model-based pressure injury staging recognition system of claim 1, wherein: the multi-layer perception network comprises a convolution neural network, a residual network, a pooling network, a multi-layer perceptron, a circulation neural network and an attention network.
6. The neural network model-based pressure injury staging recognition system of claim 1, wherein: the multi-layer perception network comprises a plurality of classification convolutional neural network models and a multi-result voting module; and each classification convolutional neural network model obtains a prediction result of the pressure injury stage, and the multi-result voting module determines the final prediction by adopting a voting method.
7. The neural network model-based pressure injury staging recognition system of claim 6 wherein: the network layers and/or network types of the plurality of classified convolutional neural network models have differences, and the differences are at least one of the following network configuration parameters: the number of layers of the neural network, the number of network elements of each layer, the type of pooling network, the Dropout probability, the convolution kernel size, the convolution step size and whether residual connection is performed or not.
8. The neural network model-based pressure injury staging recognition system of claim 6 wherein: the multiple classification convolutional neural network models are respectively obtained by training multiple data sets constructed by adopting one original data set, and the construction method of the multiple data sets comprises the following steps:
step 1, for each data set to be generated, firstly generating a random vector #p 1, p 2, p 3, p 4, p 5, v 1, v 2, v 3 )Wherein each parameter in the random vector is derived from uniformly distributed random samples,，，/>，/>，/>，/>，，/>；
step 2, carrying out data transformation on the original pictures in the original data set according to the random vectors generated in the step 1 to obtain randomly transformed pictures, wherein the random vectors arep i For controlling whether data enhancement is performed, each original picture transformation step being performed under the condition thatp i >0.5；v i For controlling the degree of transformation, wherev 1 For controlling the contrast of the contrast transformation,v 2 for controlling the angle of the random rotation,v 3 a scaling factor for controlling the scaling transformation;p i is thatp 1 、p 2 、p 3 、p 4 Or (b)p 5 ，v i Is thatv 1 、v 2 Or (b)v 3 ；
Step 3, repeating the step 1 and the step 2N times to obtain a multi-data set consisting of N data sets.
9. The neural network model-based pressure injury staging recognition system of claim 1, wherein: the model training module is used for carrying out model training on the multi-layer perception network.
10. A computer-readable storage medium having stored thereon: computer program for implementing a neural network model-based pressure injury staging recognition system according to any one of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310853766.3A CN116580282A (en) | 2023-07-12 | 2023-07-12 | Neural network model-based pressure injury staged identification system and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310853766.3A CN116580282A (en) | 2023-07-12 | 2023-07-12 | Neural network model-based pressure injury staged identification system and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116580282A true CN116580282A (en) | 2023-08-11 |
Family
ID=87534451
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310853766.3A Pending CN116580282A (en) | 2023-07-12 | 2023-07-12 | Neural network model-based pressure injury staged identification system and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116580282A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111444960A (en) * | 2020-03-26 | 2020-07-24 | 上海交通大学 | Skin disease image classification system based on multi-mode data input |
| CN112530584A (en) * | 2020-12-15 | 2021-03-19 | 贵州小宝健康科技有限公司 | Medical diagnosis assisting method and system |
| CN113261945A (en) * | 2020-02-14 | 2021-08-17 | 台北医学大学 | Monitoring system and device for monitoring pressure sores and computer implementation method |
| CN113962329A (en) * | 2021-11-15 | 2022-01-21 | 长沙理工大学 | Novel image recognition algorithm based on deep ensemble learning |
| TWI777521B (en) * | 2021-04-26 | 2022-09-11 | 智齡科技股份有限公司 | Pressure injury identification and analysis system and method |
| CN115937153A (en) * | 2022-12-13 | 2023-04-07 | 北京瑞医博科技有限公司 | Model training method and device, electronic equipment and computer storage medium |
-
2023
- 2023-07-12 CN CN202310853766.3A patent/CN116580282A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113261945A (en) * | 2020-02-14 | 2021-08-17 | 台北医学大学 | Monitoring system and device for monitoring pressure sores and computer implementation method |
| CN111444960A (en) * | 2020-03-26 | 2020-07-24 | 上海交通大学 | Skin disease image classification system based on multi-mode data input |
| CN112530584A (en) * | 2020-12-15 | 2021-03-19 | 贵州小宝健康科技有限公司 | Medical diagnosis assisting method and system |
| TWI777521B (en) * | 2021-04-26 | 2022-09-11 | 智齡科技股份有限公司 | Pressure injury identification and analysis system and method |
| CN113962329A (en) * | 2021-11-15 | 2022-01-21 | 长沙理工大学 | Novel image recognition algorithm based on deep ensemble learning |
| CN115937153A (en) * | 2022-12-13 | 2023-04-07 | 北京瑞医博科技有限公司 | Model training method and device, electronic equipment and computer storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107766894B (en) | Remote sensing image natural language generation method based on attention mechanism and deep learning | |
| CN108416065B (en) | Hierarchical neural network-based image-sentence description generation system and method | |
| KR102153920B1 (en) | System and method for interpreting medical images through the generation of refined artificial intelligence reinforcement learning data | |
| CN109783666B (en) | Image scene graph generation method based on iterative refinement | |
| JP2021166046A (en) | Method for training convolutional neural network for image recognition using image conditional mask language modeling | |
| CN111126574A (en) | Method and device for training machine learning model based on endoscopic image and storage medium | |
| CN109753571B (en) | Scene map low-dimensional space embedding method based on secondary theme space projection | |
| WO2003030085A1 (en) | System and method of face recognition using portions of learned model | |
| JP7111088B2 (en) | Image retrieval device, learning method and program | |
| CN111400548B (en) | Recommendation method and device based on deep learning and Markov chain | |
| CN111126155B (en) | Pedestrian re-identification method for generating countermeasure network based on semantic constraint | |
| CN117151112A (en) | Multi-mode key phrase generation method | |
| Zhao et al. | Deeply supervised active learning for finger bones segmentation | |
| Lauren et al. | A low-dimensional vector representation for words using an extreme learning machine | |
| WO2023231753A1 (en) | Neural network training method, data processing method, and device | |
| CN117079017A (en) | Credible small sample image identification and classification method | |
| US20240161035A1 (en) | Multi-model medical scan analysis system and methods for use therewith | |
| CN116797817A (en) | Autism disease prediction technology based on self-supervision graph convolution model | |
| CN116232699A (en) | Training method of fine-grained network intrusion detection model and network intrusion detection method | |
| CN116580282A (en) | Neural network model-based pressure injury staged identification system and storage medium | |
| CN116994695A (en) | Training method, device, equipment and storage medium of report generation model | |
| CN114168780A (en) | Multimodal data processing method, electronic device, and storage medium | |
| Shekerbek et al. | APPLICATION OF MATHEMATICAL METHODS AND MACHINE LEARNING ALGORITHMS FOR CLASSIFICATION OF X-RAY IMAGES. | |
| CN117688974B (en) | Knowledge graph-based generation type large model modeling method, system and equipment | |
| CN115831356B (en) | Auxiliary prediction diagnosis method based on artificial intelligence algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |