Accurate abdominal trauma prediction method and system based on digital twinning

Abstract

The invention discloses an abdominal trauma accurate prediction method and system based on digital twinning, and relates to the field of auxiliary medical treatment. Extracting medical structured data of a to-be-predicted patient from the wound treatment information, and screening whether abnormal data exist or not by using the medical data screening model; if the abnormal data, the reference coherent light with different wavelengths and the reference coherent light with different wavelengths in different polarization states exist, the reference coherent light simultaneously enters the abdomen of the patient to be predicted, and reflection images of the abdomen of the patient to be predicted are collected within preset time to serve as abdomen image information of the patient to be predicted; performing multi-modal fusion analysis on the medical structural data and the abdominal image information, and constructing a digital twin abdominal trauma prediction model based on a fusion analysis result; and monitoring and analyzing the digital twin abdominal trauma prediction model in real time to generate a trauma prediction report. According to the method, information related to the abdominal trauma is more comprehensively and deeply mined, and a more accurate abdominal trauma prediction model is constructed.

Description

Technical Field

[0001] The present invention relates to the field of auxiliary medicine, and more specifically to a method and system for accurately predicting abdominal trauma based on digital twins. Background Art

[0002] In modern society, various accidents occur frequently. Abdominal trauma, as a common and serious type of trauma, poses a great threat to the life and health of patients. The abdomen contains many important organs, such as the liver, spleen, gastrointestinal tract, etc. Once these organs are traumatized, the condition often deteriorates rapidly. In addition, due to the complex anatomical structure of the abdomen, it is difficult to judge the injury.

[0003] Traditional methods of diagnosing abdominal trauma have many limitations. On the one hand, when relying on doctors' experience for diagnosis, the doctors' levels vary, the diagnostic results may be greatly biased, and there is a lack of objective and quantitative evaluation standards. On the other hand, conventional examination methods, such as ultrasound examinations, are susceptible to gas interference and have poor accuracy in detecting certain deep organ injuries. Although CT can provide more detailed image information, it has radiation risks and is not suitable for frequent examinations. Before the trauma occurs, it is impossible to effectively mine potential risk factors from routine examination data, making it difficult to achieve forward-looking and accurate predictions.

[0004] With the rapid development of digital technologies such as big data, artificial intelligence, and the Internet of Things, digital twin technology has emerged. By building a virtual model of a physical entity, digital twin technology can map the state of the physical entity in real time and perform data analysis and decision-making based on the model. In the medical field, the application of digital twin technology provides a new opportunity to solve the problem of accurate prediction of abdominal trauma. Introducing digital twin technology into abdominal trauma prediction is expected to break the bottleneck of traditional diagnosis, achieve early warning and accurate assessment of abdominal trauma risks, and buy valuable time for clinical treatment. Summary of the invention

[0005] In view of this, the present invention provides an accurate prediction method and system for abdominal trauma based on digital twins, which more comprehensively and deeply mines information related to abdominal trauma and constructs a more accurate abdominal trauma prediction model.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A digital twin-based accurate prediction method for abdominal trauma includes the following steps:

[0008] Extract the medical structured data of the patient to be predicted from the trauma treatment information, and use the medical data screening model to screen whether there is abnormal data;

[0009] If there is abnormal data, reference coherent lights of different wavelengths and different polarization states of different wavelengths are incident on the abdomen of the patient to be predicted at the same time, and reflected images of the abdomen of the patient to be predicted are collected respectively within a preset time as the abdominal image information of the patient to be predicted;

[0010] Perform multimodal fusion analysis on medical structured data and abdominal image information, and build a digital twin abdominal trauma prediction model based on the fusion analysis results;

[0011] The digital twin abdominal trauma prediction model is monitored and analyzed in real time, and the attention reasoning mechanism is used to identify abnormal abdominal trauma. When abnormal conditions exist, a multi-layer trauma diagnosis mechanism is used to perform a comprehensive hierarchical diagnosis and generate a trauma prediction report.

[0012] Optionally, the medical structured data includes: patient personal data, medical file data, medical diagnosis data, examination and test data, treatment data and medical advice data; the patient personal data includes living habit information, occupation and sports information; the medical file data includes past medical history information, underlying disease information, trauma-related information; the examination and test data includes vital signs and abdominal signs.

[0013] Optionally, the construction process of the medical data screening model is as follows:

[0014] Obtain historical medical data from the medical database and use the historical medical data to build a medical data knowledge graph;

[0015] Using the Markov decision process framework, potential false correlations are retrieved in the medical data knowledge graph. If false correlations exist, the medical data attribute information in the medical data knowledge graph triples is modified, and the corresponding entities are selected and modified to reduce the false correlations.

[0016] The entity data of the replaced medical data knowledge graph is updated, and the medical data knowledge graph is updated each time the medical data knowledge graph is used to obtain a further optimized medical data screening model.

[0017] Optionally, the medical structured data and abdominal image information are subjected to multimodal fusion analysis, specifically: extracting the time series features and image features of the medical structured data and abdominal image information; generating a mapping relationship between the time series features and the image features to achieve multimodal feature mapping, and using a spectral clustering algorithm to construct a weighted similarity matrix and a Laplace matrix; using a multimodal data encoder to generate a coding matrix with a mapping relationship for the weighted similarity matrix and the Laplace matrix.

[0018] Optionally, a medical digital twin model is obtained, and the coding matrix corresponding to the medical structured data and abdominal image information is input into the medical digital twin model to obtain a digital twin abdominal trauma prediction model.

[0019] Optionally, the attention reasoning mechanism specifically includes the following steps:

[0020] Multi-layer graph attention network, each layer consists of multiple independent attention heads. In each layer, each attention head will weight and aggregate the features of neighboring nodes, and then the outputs of multiple attention heads are spliced ​​to obtain the embedded representation of the node;

[0021] Based on the obtained node embedding representation, a similarity matrix is ​​constructed, and the similarity between nodes is calculated through the Gaussian kernel function. Based on the similarity matrix, a standardized Laplace matrix is ​​further calculated to obtain the eigenvector of the matrix;

[0022] These eigenvectors are normalized to finally obtain a normalized eigenvector matrix. Each row of the normalized eigenvector matrix is ​​clustered using a clustering algorithm to obtain the inference result of the abnormal situation.

[0023] An accurate prediction system for abdominal trauma based on digital twins, including:

[0024] Abnormal data initial screening module: used to extract the medical structured data of the patient to be predicted from the trauma treatment information, and use the medical data screening model to screen whether there is abnormal data;

[0025] Abdominal image information generation module: used for, if there is abnormal data, simultaneously incident on the abdomen of the patient to be predicted with reference coherent lights of different wavelengths and different polarization states of different wavelengths, and collecting reflected images of the abdomen of the patient to be predicted within a preset time as the abdominal image information of the patient to be predicted;

[0026] Digital twin abdominal trauma prediction model construction module: used to perform multimodal fusion analysis on medical structured data and abdominal image information, and build a digital twin abdominal trauma prediction model based on the fusion analysis results;

[0027] Trauma prediction report generation module: used to monitor and analyze the digital twin abdominal trauma prediction model in real time, use the attention reasoning mechanism to identify abnormal abdominal trauma, and when abnormal conditions exist, use the multi-layer trauma diagnosis mechanism to perform comprehensive hierarchical diagnosis and generate a trauma prediction report.

[0028] It can be seen from the above technical solutions that, compared with the prior art, the present invention provides a method and system for accurately predicting abdominal trauma based on digital twins, which has the following beneficial effects:

[0029] 1. In the case of abnormal data, reference coherent light of different wavelengths and polarization states is used to simultaneously irradiate the abdomen of the patient to be predicted, and the reflected image is collected as the abdominal image information within a preset time. This multi-dimensional image acquisition method can obtain richer abdominal information and reflect the abdominal trauma status from different angles.

[0030] 2. Build a digital twin abdominal trauma prediction model based on the fusion analysis results. Digital twin technology can create a virtual model that is highly similar to the actual abdominal trauma situation. By learning and analyzing a large amount of data, it can accurately simulate the development and changes of abdominal trauma and achieve accurate prediction of abdominal trauma.

[0031] 3. When an abnormal situation is identified, a multi-layer trauma diagnosis mechanism is used for comprehensive hierarchical diagnosis. This hierarchical diagnosis method can analyze trauma from multiple levels and different angles, comprehensively assess the severity and type of trauma, avoid the one-sidedness that may occur in a single diagnostic method, and improve the accuracy and reliability of diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0033] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0034] Figure 2 It is a schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] The embodiment of the present invention discloses a method for accurately predicting abdominal trauma based on digital twins, such as Figure 1 As shown, the following steps are included:

[0037] Step 1: extract the medical structured data of the patient to be predicted from the trauma treatment information, and use the medical data screening model to screen whether there is abnormal data;

[0038] Step 2: If there is abnormal data, reference coherent lights of different wavelengths and different polarization states of different wavelengths are incident on the abdomen of the patient to be predicted at the same time, and reflected images of the abdomen of the patient to be predicted are collected respectively within a preset time as the abdominal image information of the patient to be predicted;

[0039] Step 3: Perform multimodal fusion analysis on medical structured data and abdominal image information, and build a digital twin abdominal trauma prediction model based on the fusion analysis results;

[0040] Step 4: Monitor and analyze the digital twin abdominal trauma prediction model in real time, use the attention reasoning mechanism to identify abnormal abdominal trauma, and when abnormal conditions exist, use the multi-layer trauma diagnosis mechanism to perform a comprehensive hierarchical diagnosis and generate a trauma prediction report.

[0041] Furthermore, abdominal stress refers to various forces acting on abdominal tissues and organs, including pressure, tension, shear force, etc. These forces are in a relatively balanced state under normal physiological conditions, maintaining the normal function and position of abdominal organs. However, when affected by external factors or internal physiological changes, abdominal stress may change beyond the normal range, thereby having adverse effects on abdominal tissues and organs. Therefore, in step one, the medical structured data includes: patient personal data, medical file data, medical diagnosis data, examination and testing data, treatment data, and doctor's advice data; the patient's personal data includes living habits information, occupation and sports information; the medical file data includes past medical history information, basic disease information, and trauma-related information; the examination and testing data includes vital signs and abdominal signs.

[0042] Furthermore, it also includes the pain situation: the location of the pain, such as the right upper abdomen, left upper abdomen, around the navel, etc.; the degree of pain, which can be recorded using the visual analog scale; the nature of the pain, such as tingling, dull pain, colic, etc.; the onset time and duration of the pain, and whether the pain is accompanied by radiating pain, and to which parts of the body the pain radiates, etc.

[0043] Abdominal appearance: Observe and record whether the abdomen has swelling, bruises, wounds, etc. The degree of swelling, the size and color of bruises, the shape and depth of wounds, etc. are all important in determining the severity and type of abdominal trauma.

[0044] Gastrointestinal symptoms: whether there are nausea, vomiting, diarrhea, bloody stools and other symptoms; the color, properties, and amount of vomitus and stools are also important to record. For example, bloody vomitus or black tarry stools may indicate different abdominal trauma conditions.

[0045] Vital signs: basic vital signs data such as body temperature, blood pressure, heart rate, respiratory rate, etc. These indicators can reflect the patient's overall physical condition and the severity of the disease. Abnormal changes in vital signs may indicate that abdominal trauma has caused serious complications such as shock.

[0046] The construction process of the medical data screening model is as follows:

[0047] Step 1.1, obtain historical medical data in the medical database, and use the historical medical data to build a medical data knowledge graph;

[0048] Step 1.2: Using the Markov decision process framework, retrieve potential false correlations in the medical data knowledge graph. If false correlations exist, modify the medical data attribute information in the medical data knowledge graph triples, select the corresponding entities and modify and replace them to reduce the false correlations.

[0049] Step 1.3: Update the entity data of the replaced medical data knowledge graph, and update the medical data knowledge graph each time it is used to obtain a further optimized medical data screening model.

[0050] Furthermore, in step three, the medical structured data and abdominal image information are subjected to multimodal fusion analysis, specifically: extracting the time series features and image features of the medical structured data and abdominal image information; generating a mapping relationship between the time series features and the image features to achieve multimodal feature mapping, and using a spectral clustering algorithm to construct a weighted similarity matrix and a Laplace matrix; using a multimodal data encoder to generate a coding matrix with a mapping relationship for the weighted similarity matrix and the Laplace matrix.

[0051] Among them, the method for constructing a digital twin abdominal trauma prediction model is: obtain a medical digital twin model, use the coding matrix corresponding to medical structured data and abdominal image information to input into the medical digital twin model, and obtain a digital twin abdominal trauma prediction model.

[0052] The above technical solution has the following beneficial effects: Based on the fusion analysis results, a digital twin abdominal trauma prediction model is constructed. The digital twin technology can create a virtual model that is highly similar to the actual abdominal trauma situation. By learning and analyzing a large amount of data, the development and changes of abdominal trauma can be accurately simulated, and accurate prediction of abdominal trauma can be achieved. Real-time monitoring and analysis of the digital twin abdominal trauma prediction model can timely detect abnormalities in the model prediction results. Abdominal trauma abnormalities can be quickly identified using the attention reasoning mechanism, which helps to issue early warnings in the early stages of trauma development, buy time for timely intervention and treatment, and improve the success rate of treatment.

[0053] Furthermore, in step 4, the attention reasoning mechanism specifically includes the following steps:

[0054] Step 4.1: Multi-layer graph attention network, each layer consists of multiple independent attention heads. In each layer, each attention head will weight and aggregate the features of neighboring nodes, and then concatenate the outputs of multiple attention heads to obtain the embedded representation of the node;

[0055] Step 4.2: Based on the obtained node embedding representation, a similarity matrix is ​​constructed, and the similarity between nodes is calculated through the Gaussian kernel function. Based on the similarity matrix, a standardized Laplace matrix is ​​further calculated to obtain the eigenvector of the matrix;

[0056] Step 4.3: Normalize the eigenvectors to obtain a normalized eigenvector matrix. Use a clustering algorithm to cluster each row of the normalized eigenvector matrix to obtain the inference result of the abnormal situation.

[0057] and Figure 1 Corresponding to the method shown, the present invention also discloses a digital twin-based accurate prediction system for abdominal trauma Figure 1 The implementation of the method, the specific structure is as follows Figure 2 As shown, including:

[0058] Abnormal data initial screening module: used to extract the medical structured data of the patient to be predicted from the trauma treatment information, and use the medical data screening model to screen whether there is abnormal data;

[0059] Abdominal image information generation module: used for, if there is abnormal data, simultaneously incident on the abdomen of the patient to be predicted with reference coherent lights of different wavelengths and different polarization states of different wavelengths, and collecting reflected images of the abdomen of the patient to be predicted within a preset time as the abdominal image information of the patient to be predicted;

[0060] Digital twin abdominal trauma prediction model construction module: used to perform multimodal fusion analysis on medical structured data and abdominal image information, and build a digital twin abdominal trauma prediction model based on the fusion analysis results;

[0061] Trauma prediction report generation module: used to monitor and analyze the digital twin abdominal trauma prediction model in real time, use the attention reasoning mechanism to identify abnormal abdominal trauma, and when abnormal conditions exist, use the multi-layer trauma diagnosis mechanism to perform comprehensive hierarchical diagnosis and generate a trauma prediction report.

[0062] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0063] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A digital twin-based accurate prediction method for abdominal trauma, characterized in that: The following steps are involved: Extract the medical structured data of the patient to be predicted from the trauma treatment information, and use the medical data screening model to screen whether there is abnormal data; If there is abnormal data, reference coherent lights of different wavelengths and different polarization states of different wavelengths are incident on the abdomen of the patient to be predicted at the same time, and reflected images of the abdomen of the patient to be predicted are collected respectively within a preset time as the abdominal image information of the patient to be predicted; Perform multimodal fusion analysis on medical structured data and abdominal image information, and build a digital twin abdominal trauma prediction model based on the fusion analysis results; The digital twin abdominal trauma prediction model is monitored and analyzed in real time, and the attention reasoning mechanism is used to identify abnormal abdominal trauma. When abnormal conditions exist, a multi-layer trauma diagnosis mechanism is used to perform a comprehensive hierarchical diagnosis and generate a trauma prediction report.

2. According to claim 1, a digital twin-based accurate prediction method for abdominal trauma is characterized in that: The medical structured data includes: patient personal data, medical file data, medical diagnosis data, examination and testing data, treatment data and doctor's order data; the patient personal data includes living habit information, occupation and sports information; the medical file data includes past medical history information, basic disease information, trauma-related information; the examination and testing data includes vital signs and abdominal signs.

3. According to claim 1, a digital twin-based accurate prediction method for abdominal trauma is characterized in that: The construction process of the medical data screening model is as follows: Obtain historical medical data from the medical database and use the historical medical data to build a medical data knowledge graph; Using the Markov decision process framework, potential false correlations are retrieved in the medical data knowledge graph. If false correlations exist, the medical data attribute information in the medical data knowledge graph triples is modified, and the corresponding entities are selected and modified to reduce the false correlations. The entity data of the replaced medical data knowledge graph is updated, and the medical data knowledge graph is updated each time the medical data knowledge graph is used to obtain a further optimized medical data screening model.

4. According to claim 1, a digital twin-based accurate prediction method for abdominal trauma is characterized in that: The medical structured data and abdominal image information are subjected to multimodal fusion analysis, specifically: extracting the temporal features and image features of the medical structured data and abdominal image information; generating a mapping relationship between the temporal features and the image features to realize multimodal feature mapping, and using the spectral clustering algorithm to construct a weighted similarity matrix and a Laplace matrix; using a multimodal data encoder to generate a coding matrix with a mapping relationship for the weighted similarity matrix and the Laplace matrix.

5. The method for accurately predicting abdominal trauma based on digital twin according to claim 4, characterized in that: A medical digital twin model was obtained, and the coding matrix corresponding to the medical structured data and abdominal image information was input into the medical digital twin model to obtain a digital twin abdominal trauma prediction model.

6. The method for accurately predicting abdominal trauma based on digital twin according to claim 1, characterized in that: The attention reasoning mechanism specifically includes the following steps: Multi-layer graph attention network, each layer consists of multiple independent attention heads. In each layer, each attention head will weight and aggregate the features of neighboring nodes, and then the outputs of multiple attention heads are spliced ​​to obtain the embedded representation of the node; Based on the obtained node embedding representation, a similarity matrix is ​​constructed, and the similarity between nodes is calculated through the Gaussian kernel function. Based on the similarity matrix, a standardized Laplace matrix is ​​further calculated to obtain the eigenvector of the matrix; The eigenvectors are normalized to obtain a normalized eigenvector matrix. Each row of the normalized eigenvector matrix is ​​clustered using a clustering algorithm to obtain the inference result of the abnormal situation.

7. An accurate prediction system for abdominal trauma based on digital twins, characterized in that: include: Abnormal data initial screening module: used to extract the medical structured data of the patient to be predicted from the trauma treatment information, and use the medical data screening model to screen whether there is abnormal data; Abdominal image information generation module: if there is abnormal data, reference coherent light of different wavelengths and different wavelengths in different polarization states are incident on the abdomen of the patient to be predicted at the same time, and reflected images of the abdomen of the patient to be predicted are collected respectively within a preset time as the abdominal image information of the patient to be predicted; Digital twin abdominal trauma prediction model construction module: used to perform multimodal fusion analysis on medical structured data and abdominal image information, and build a digital twin abdominal trauma prediction model based on the fusion analysis results; Trauma prediction report generation module: used to monitor and analyze the digital twin abdominal trauma prediction model in real time, use the attention reasoning mechanism to identify abnormal abdominal trauma, and when abnormal conditions exist, use the multi-layer trauma diagnosis mechanism to perform comprehensive hierarchical diagnosis and generate a trauma prediction report.

Images