Intelligent decision-making and priority distribution system for trauma rescue

Abstract

The invention relates to the technical field of trauma rescue, in particular to a trauma rescue intelligent decision-making and priority distribution system which comprises the steps of collecting original rescue data; preprocessing the original rescue data to obtain preprocessed rescue data; processing the pre-processed rescue data through a deep learning model to obtain shock probability, injury type and environmental physiology associated data; calculating ISS total score priority by using shock probability, injury type and environmental physiology associated data association; and generating a patient priority list according to the ISS total score priority, and preferentially carrying out resource allocation on the patient with the highest score in the patient priority list. According to the method, the injury condition scores of the patients are dynamically generated through the deep learning model, the medical resources are automatically allocated according to the high-low sequence of the scores, the optimal ambulance route is generated in combination with the GIS map, it is ensured that the critical patients can be treated preferentially, and the fairness and timeliness of rescue are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of trauma rescue, and particularly to an intelligent decision-making and priority allocation system for trauma rescue. Background Art

[0002] When major disasters occur, such as earthquakes, floods, fires or traffic accidents, a large number of casualties often emerge, while the medical resources at the scene are extremely limited.

[0003] In the scenario of emergency medical rescue, the survival rate of patients highly depends on rapid triage, scientific classification and resource scheduling. However, the traditional methods are limited by factors such as timeliness, resource limitations, chaotic on-site information and complex injury condition judgment, and there is an urgent need for technological empowerment.

[0004] Traditional trauma rescue decisions often highly rely on on-site manual evaluation by medical staff and judgment based on past experience. This method has significant subjectivity. Medical staff with different experience levels may have differences in the evaluation of the same injury condition, and the manual evaluation efficiency is low, which easily leads to uneven resource allocation. Especially in emergency situations, resources may not be allocated to the most needy injured in a timely and reasonable manner due to human factors. Therefore, more advanced technologies are needed to optimize the trauma rescue process and improve the rescue efficiency and accuracy. Summary of the Invention

[0005] The present invention provides an intelligent decision-making and priority allocation system for trauma rescue to solve the defect of long time consumption in traditional manual triage in the prior art.

[0006] The present invention provides an intelligent decision-making and priority allocation system for trauma rescue, including:

[0007] A data acquisition module, configured to acquire original rescue data.

[0008] A data preprocessing module, configured to preprocess the original rescue data to obtain preprocessed rescue data.

[0009] A multi-modal deep learning module, which processes the preprocessed rescue data through a deep learning model to obtain shock probability, injury type, and environmental-physiological correlation data.

[0010] An ISS priority calculation module, configured to calculate the total ISS score priority through correlation calculation of shock probability, injury type, environmental risk data and physiological signal data.

[0011] A priority ranking module, configured to generate a patient priority list according to the total ISS score priority, and preferentially allocate resources to the patient with the highest score in the patient priority list.

[0012] A trauma rescue intelligent decision-making and priority allocation system provided by the present invention, the original rescue data includes: physiological signal data, injury images, and environmental risk data.

[0013] The preprocessing of the data preprocessing module includes:

[0014] Data denoising: Perform denoising processing on the physiological signal data.

[0015] Data normalization: Map the original rescue data to a preset interval range.

[0016] Image cropping: Crop the irrelevant background area in the injury image.

[0017] Image enhancement: Enhance the contrast and clarity of the injury image.

[0018] Data standardization: Standardize the environmental risk data.

[0019] A trauma rescue intelligent decision-making and priority allocation system provided by the present invention, the multi-modal deep learning module includes:

[0020] A physiological signal analysis unit for processing time series data using an LSTM network model to predict the probability of shock.

[0021] An imaging injury recognition unit for detecting the type and severity of injuries within the segmented area using a CNN model.

[0022] An environmental risk quantification unit for associating environmental risk data with physiological signal data using a fully connected network model to obtain environmental-physiological association data.

[0023] A trauma rescue intelligent decision-making and priority allocation system provided by the present invention, the specific steps for processing time series data using an LSTM network model to predict the probability of shock are:

[0024] Extract feature vectors from the physiological signal data.

[0025] Set the length and step size of the sliding window, and process the feature vectors into time series feature vectors.

[0026] Construct an LSTM network model, and determine the number of LSTM layers and the number of neurons in each layer.

[0027] Use loss function one and optimizer one to train the LSTM network model.

[0028] Input the time series feature vectors into the trained LSTM network model to obtain the probability of shock.

[0029] According to a trauma rescue intelligent decision-making and priority allocation system provided by the present invention, the specific steps of using a CNN model to detect the injury type and severity in the segmentation area are as follows:

[0030] Separate the injury area in the injury image from the background.

[0031] Use the CNN model to extract injury features from the injury area.

[0032] Identify the injury type based on the injury features and evaluate the severity of the injury.

[0033] According to a trauma rescue intelligent decision-making and priority allocation system provided by the present invention, the specific steps of using a fully connected network model to associate environmental risk data and physiological signal data are as follows:

[0034] Determine the number of neurons in the input layer, hidden layer, and output layer in the fully connected network structure.

[0035] Use loss function two and optimizer two for model training.

[0036] Analyze the prediction results of the fully connected network model on the validation set, obtain the deviation between the prediction results of the fully connected network model on the validation set and the true values, adjust the feature weights, and increase the feature interaction terms to correct the deviation between the prediction results on the validation set and the true values, so as to obtain the association between environmental risk data and physiological signal data.

[0037] According to a trauma rescue intelligent decision-making and priority allocation system provided by the present invention, the IISS priority calculation module includes:

[0038] ISS basic score calculation unit, calculate the sum of squares according to the highest scores of the injury type and severity to obtain the basic ISS score.

[0039] Shock risk dynamic correction unit, use the shock probability and shock risk weight coefficient for calculation to dynamically correct the ISS basic score to obtain the corrected total ISS score.

[0040] Environmental priority compensation unit, use the environmental risk coefficient obtained after associating the environmental risk data and physiological signal data to perform environmental priority compensation on the corrected total ISS score to obtain the compensated total ISS score priority.

[0041] According to a trauma rescue intelligent decision-making and priority allocation system provided by the present invention, the steps of calculating the corrected total ISS score are as follows:

[0042] Calculate the variable ISS score according to the basic ISS score, shock probability, and shock risk weight coefficient.

[0043] Calculate the corrected total ISS score based on the changing ISS score and the baseline ISS score.

[0044] According to a trauma rescue intelligent decision-making and priority allocation system provided by the present invention, the priority ranking module generates a priority list including:

[0045] Calculate the comprehensive priority of the patient based on the ISS priority for the severity of the patient's trauma.

[0046] Rank the patients from high to low according to the comprehensive priority of the patients to generate a patient priority list.

[0047] According to the patient priority list, allocate corresponding medical resources to patients with different priorities and formulate a resource allocation plan.

[0048] Plan the optimal transportation route for patients with different priorities according to the location and traffic conditions of the hospital.

[0049] According to a trauma rescue intelligent decision-making and priority allocation system provided by the present invention, the comprehensive priority of the patient is:

[0050] Obtain the index stability factor S from the physiological signal data.

[0051] Obtain the transportation time T of the patient from the injury location to the medical institution from the historical data.

[0052] Obtain the environmental hazard degree factor E from the environmental risk data.

[0053] Allocate weights according to the importance of the stability factor, transportation time, environmental hazard degree factor, and the priority of the compensated total ISS score in the comprehensive ranking.

[0054] Calculate the comprehensive priority of the patient through the stability factor, transportation time, environmental hazard degree factor, and the priority of the compensated total ISS score.

[0055] A trauma rescue intelligent decision-making and priority allocation system provided by the present invention dynamically generates a patient injury score through a deep learning model, automatically allocates medical resources according to the score ranking, and combines with a GIS map to generate the optimal route for ambulances, solving the problem of long time-consuming for traditional manual triage. The beneficial effects obtained are:

[0056] The present invention constructs a multi-dimensional data fusion platform by integrating physiological signals, imaging data, and environmental sensor information. Using a deep learning model to deeply mine and analyze massive data, it can dynamically generate an accurate patient injury score, which not only covers the severity of the injury, but also considers the physiological state of the injured and the possible impact of environmental factors on the rescue, thus realizing the precision and efficiency of medical resource allocation.

[0057] The present invention utilizes a deep learning model to deeply mine and analyze a large amount of data. The deep learning model has powerful feature extraction and pattern recognition capabilities, can learn valuable patterns and rules from complex data, and then dynamically generate accurate patient injury scores, greatly improving the efficiency and accuracy of injury judgment, and can quickly identify patients in urgent need of treatment. This score not only covers the severity of the injury, but also comprehensively considers the physiological state of the injured and the possible impact of environmental factors on rescue, providing a scientific basis for the allocation of medical resources.

[0058] The present invention quickly processes and analyzes this multi-source data, objectively quantifies the severity of the injury through algorithms, and automatically allocates medical resources according to the score ranking. It automatically allocates medical resources according to the score ranking to ensure that critically ill patients can receive treatment first, greatly improving the fairness and timeliness of rescue, changing the unbalanced and untimely situation that may occur when manually allocating resources in the past, ensuring that resources can be accurately delivered to the places where they are most needed, and improving the overall rescue efficiency. Combined with GIS, the system can generate the optimal route for ambulances and push it to the emergency command center in real time. This function not only shortens the response time, but also optimizes the rescue route through algorithms, avoiding the influence of unforeseen factors such as traffic congestion, and significantly improving the rescue efficiency.

[0059] The present invention reduces human errors and improves the efficiency of resource utilization through data-driven decision support, making the trauma rescue process more scientific and intelligent. The intelligent system can continuously learn and optimize its decision-making model. With the increase in data volume and the iteration of algorithms, the accuracy and efficiency of its decisions will continue to improve, ultimately achieving faster response, better resource allocation, and higher treatment success rates, bringing a revolutionary change to the trauma rescue work. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0061] Figure 1 It is a module diagram of a trauma rescue intelligent decision-making and priority allocation system provided by an embodiment of the present invention;

[0062] Figure 2 It is a flow schematic diagram of a physiological signal analysis unit provided by an embodiment of the present invention;

[0063] Figure 3 It is a flow schematic diagram of an environmental risk quantification unit provided by an embodiment of the present invention. Detailed implementation manners

[0064] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without creative efforts shall fall within the protection scope of the present invention.

[0065] The following combines Figures 1 - 3 to describe an intelligent decision-making and priority assignment system for trauma rescue of the present invention.

[0066] As Figure 1 shown, an intelligent decision-making and priority assignment system for trauma rescue provided by an embodiment of the present invention includes: a data acquisition module, a data preprocessing module, a multimodal deep learning module, an ISS priority calculation module, and a priority sorting module.

[0067] The data acquisition module is used to acquire data during the trauma rescue process to obtain the original rescue data. The original rescue data includes physiological signal data, injury images, and environmental risk data.

[0068] The physiological parameters such as the patient's heart rate, blood pressure, and blood oxygen saturation are collected in real time through wearable devices, and the sampling frequency is greater than 100 Hz to ensure data accuracy. The wounded are quickly scanned using a portable ultrasound device or a vehicle-mounted CT scanner to generate CT / X-ray images, which are transmitted to the edge computing node through a 5G / satellite network. The environmental sensors deployed at the scene synchronously collect data such as the temperature, air pressure, and pollutant concentration at the disaster site, and mark the spatio-temporal coordinates for subsequent analysis.

[0069] The data preprocessing module is used to preprocess the original data. The preprocessing includes:

[0070] Denoising the physiological signal data to remove abnormal fluctuations caused by factors such as equipment interference, and at the same time performing a normalization operation to map the data to a specific interval range for subsequent model processing.

[0071] Performing operations such as cropping and enhancement on the image data, cropping off irrelevant background areas, and enhancing key features such as the contrast and clarity of the image to highlight the image details helpful for injury judgment.

[0072] Standardizing the environmental sensor information so that its data form and magnitude are adapted to other data.

[0073] According to the patient's basic information and the existing diagnosis and classification criteria, the preprocessed data is classified and labeled, for example, initially classified and marked according to different trauma types, severity ranges, etc.

[0074] A multi-modal deep learning module processes the preprocessed rescue data through a deep learning model to obtain the shock probability, injury type, and environmental-physiological correlation data. The multi-modal deep learning module includes: a physiological signal analysis unit, an image injury recognition unit, and an environmental risk quantification unit.

[0075] As Figure 2 shown, the physiological signal analysis unit is used to construct an LSTM network, process time series data using the LSTM network, and predict the shock probability.

[0076] Extract features related to shock risk from the preprocessed rescue data to obtain a feature vector, set the sliding window length and step size, generate overlapping time windows, construct an LSTM network model, design a network structure including LSTM layers, determine the number of LSTM layers and the number of neurons in each layer. A Dropout layer can be added after the LSTM layer to prevent overfitting, train the model, divide the dataset into a training set, a validation set, and a test set, and use loss function one and optimizer one for model training. Here, loss function one includes: mean squared error, cross-entropy loss, mean absolute error, and optimizer one includes: stochastic gradient descent, Adam, Adagrad. Input the time series feature vector into the trained LSTM network model to obtain the shock probability.

[0077] The image injury recognition unit is used to detect the injury type and severity within the segmented area using a CNN model.

[0078] Use image segmentation technology to separate the injury area in the image data from the background, use a CNN model to extract features from the segmented injury area to obtain injury features, evaluate the severity of the injury based on the injury features, and classify the injury degree into different injury types according to the injury features.

[0079] As Figure 3 shown, the environmental risk quantification unit is used to correlate environmental risk data with physiological signal data using a fully connected network model.

[0080] Determine the number of neurons in the input layer, hidden layer, and output layer in the fully connected network structure, use loss function two and optimizer two for model training. Here, loss function two can be: mean squared error, cross-entropy loss, mean absolute error, and optimizer two includes: stochastic gradient descent, Adam, Adagrad. Analyze the prediction results of the fully connected network model on the validation set, obtain the deviation between the prediction results of the fully connected network model on the validation set and the true values, adjust the feature weights and add feature interaction terms to correct the deviation between the prediction results on the validation set and the true values, and obtain the correlation between environmental risk data and physiological signal data.

[0081] The ISS priority calculation module is used to calculate the dynamic ISS score using the shock probability.

[0082] The ISS basic score calculation unit: calculates the sum of squares based on the highest scores of the injury type and severity to obtain the basic ISS score.

[0083] The human body is divided into six regions, including the head, neck, face, chest, abdomen and pelvis, limbs and skin. Identify the three most severely traumatized sites and give the corresponding AIS (Abbreviated Injury Scale) index. The AIS index is a means of quantifying the injury of organs and tissues according to the severity of the injury, rated from 1 to 6 points according to the degree of injury, where 1 point is for mild injury, 2 points is for moderate injury, 3 points is for relatively severe injury, 4 points is for severe injury, 5 points is for critical injury, and 6 points is for the most severe injury. Calculate the sum of squares of the highest AIS scores of these three most severely injured regions, that is, ISS basic score = AIS1² + AIS2² + AIS3² 2 If the AIS of any injured region reaches 6 points, that is, the maximum injury, the injury severity score of this region is automatically determined to be 75 points. At this time, regardless of other injury conditions, the ISS basic score is also automatically determined to be 75 points. At the same time, the total score of the ISS basic score shall not exceed 75 points.

[0084] The shock risk dynamic correction unit: uses the shock probability and the shock risk weight coefficient to calculate and dynamically correct the ISS basic score to obtain the corrected total ISS score.

[0085] The formula for the corrected total ISS score is expressed as:

[0086] ISS t = ISS b + ISS b × P s × ω s

[0087] In the formula, ISS t is the corrected total ISS score, ISS b is the basic ISS score, P s is the shock probability, ω s is the shock risk weight coefficient.

[0088] The environmental priority compensation unit: uses the environmental risk coefficient obtained after correlating the environmental risk data and the physiological signal data to perform environmental priority compensation on the corrected total ISS score to obtain the priority of the compensated total ISS score. If the environmental risk coefficient ≥ 1.2, the total ISS score is increased by an additional 10%.

[0089] The priority ranking module is used to perform resource allocation according to the patient priority list calculated from the dynamic ISS score.

[0090] Calculate the comprehensive priority of the patient based on the ISS priority to obtain the patient's overall priority.

[0091] The formula for calculating the comprehensive priority of the patient is expressed as:

[0092] P = ω1×ISS + ω2×S + ω3×T + ω4×E

[0093] In the formula, ISS is the ISS score, ω1 is the weight coefficient of the ISS score, S is the physiological index stability factor, ω2 is the weight coefficient of the physiological index stability factor, T is the transportation time factor, ω2 is the weight coefficient of the transportation time factor, E is the environmental hazard degree factor, and ω4 is the weight coefficient of the environmental hazard degree factor.

[0094] Generate a patient priority list: Sort the patients from high to low according to the patient's comprehensive priority to generate a patient priority list. In this list, patients with high priorities (such as patients with acute diseases, patients who need special attention in the short term after surgery, patients with unstable conditions, etc.) will be ranked at the front to ensure that they can receive timely treatment.

[0095] Formulate a resource allocation plan: According to the patient priority list, allocate corresponding medical resources to patients with different priorities. This includes deciding which patients can obtain medical resources such as beds, doctors, and drugs first to ensure that the patients in the greatest need can receive timely and effective treatment. The resource allocation plan needs to follow the principle of making efficient use of limited medical resources to meet the medical needs of patients.

[0096] Plan the transportation route: For patients who need to be transported, consider factors such as the location of the hospital and traffic conditions to plan the optimal transportation route. This includes selecting appropriate means of transportation and determining the best driving route, etc., to ensure that the patient can reach the destination hospital as soon as possible and safely. When planning the transportation route, the issue of safety risk control during the patient's transportation also needs to be considered.

[0097] Example 1: The heart rate of casualty A is 80 beats per minute, blood pressure is 120 / 80 mmHg, and blood oxygen saturation is 98%, etc. The heart rate of casualty B is 100 beats per minute, blood pressure is 140 / 90 mmHg, and blood oxygen saturation is 95%, etc. The CT image of casualty A shows a suspected rib fracture in the chest, and the CT image of casualty B indicates a soft tissue contusion in the abdomen, etc. Transmit these imaging data to the edge computing node through the 5G network. The environmental sensors deployed at the accident site work synchronously, and the on-site temperature is collected as 30°C, the air pressure is normal, but due to situations such as vehicle fuel leakage, the pollutant concentration is relatively high, and the spatio-temporal coordinate information at this time and place is marked.

[0098] Extract features related to shock risk from the preprocessed rescue data, set the sliding window length to 10 minutes and the step size to 2 minutes to generate overlapping time windows, construct a network structure with 2 LSTM layers, each LSTM layer having 64 neurons, add a Dropout layer later to prevent overfitting, and finally make the final prediction through a fully connected layer. Divide the dataset into a training set, a validation set, and a test set in the ratio of 7:2:1, use the cross-entropy loss function and the Adam optimizer for model training, monitor the performance of the validation set and adjust the hyperparameters. After analysis, the shock probability of casualty A is predicted to be 0.2, and the shock probability of casualty B is predicted to be 0.8.

[0099] Use image segmentation technology to separate the damaged areas in the CT images of casualties A and B from the background, use a CNN model to extract features from the segmented damaged areas, construct a classification model to classify the damages into different types such as fractures and soft tissue contusions, and construct a regression model to evaluate the severity of the damages. Through the analysis of the trained model, it is determined that the rib fracture in the chest of casualty A is a moderate injury (AIS index is 2 points), and the soft tissue contusion in the abdomen of casualty B is a relatively serious injury (AIS index is 3 points).

[0100] Design a fully connected network structure. The input layer has 3 neurons (corresponding to the 3 input features of temperature, air pressure, and pollutant concentration), the hidden layer has 5 neurons, and the output layer outputs the environmental risk coefficient. Divide the training set and the validation set, select the mean squared error loss function and the SGD optimizer for model training, analyze the prediction results of the validation set, and adjust the feature weights, etc. for correction after identifying the deviation. Finally, the environmental risk coefficient of the accident site is obtained as 1.3 because of the existence of high temperature and high pollutant concentration.

[0101] ISS basic score calculation unit: For casualty A, the three most severely damaged parts of his body are chest abrasion with an AIS index of 2 points, abrasions on the limbs with an AIS index of 1 point, and skin abrasions with an AIS index of 1 point. Calculate the ISS basic score = 2 2 +1 2 +1 2 = 6 points. For casualty B, the three most severely damaged parts of the body are abdominal abrasion with an AIS index of 3 points, chest contusion with an AIS index of 1 point, and limb abrasions with an AIS index of 1 point. The ISS basic score = 3 2 +1 2 +1 2 = 11 points.

[0102] Shock risk dynamic correction unit: For casualty A, the shock probability is 0.2. The ISS basic score is dynamically corrected by using the shock probability and the weight coefficient of 0.3. The corrected total ISS score = 6 + 0.2×0.3 = 6.06 points. For casualty B, the corrected total ISS score = 11 + 0.8×0.3 = 11.24 points.

[0103] Environmental priority compensation unit: Since the environmental risk coefficient at the accident scene is 1.3 ≥ 1.2, environmental priority compensation is carried out for the total ISS scores of casualties A and B. The priority of the total ISS score after compensation for casualty A = 6.06×1.1 ≈ 6.67 points. The priority of the total ISS score after compensation for casualty B = 11.24×1.1 ≈ 12.36 points.

[0104] Generate a patient priority list: According to the calculated dynamic ISS scores, casualty B is ranked in front and casualty A is ranked behind to generate a patient priority list.

[0105] Formulate a resource allocation plan: Based on the priority list, arrange a bed for casualty B that can perform further abdominal examinations and treatments first, deploy more experienced surgeons for diagnosis, and prepare relevant drugs that may be used. For casualty A, also arrange subsequent examination and basic treatment resources accordingly to ensure that their injuries can be properly treated.

[0106] Plan the transportation route: Combining the GIS map information, considering factors such as the distance to surrounding hospitals and traffic congestion, plan the best transportation route for casualty B. For example, select a hospital with a suitable distance and smooth traffic at the moment, and arrange the ambulance to drive according to the planned route to ensure that casualty B can be delivered as soon as possible. Similarly, plan a suitable transportation route for casualty A to ensure their safe transportation to the corresponding hospital for treatment.

[0107] A trauma rescue intelligent decision-making and priority allocation system provided by the present invention dynamically generates patient injury scores through a deep learning model, automatically allocates medical resources according to the ranking of the scores, and generates the optimal route for the ambulance in combination with the GIS map, solving the problem of long time-consuming traditional manual triage. The beneficial effects obtained are:

[0108] The present invention integrates physiological signals, imaging data, and environmental sensor information to construct a multi-dimensional data fusion platform. Using a deep learning model to deeply mine and analyze massive data can dynamically generate accurate patient injury scores.

[0109] The present invention utilizes a deep learning model to deeply mine and analyze massive data. The deep learning model has powerful feature extraction and pattern recognition capabilities, can learn valuable patterns and rules from complex data, and then dynamically generate accurate patient injury scores, greatly improving the efficiency and accuracy of injury judgment and quickly identifying patients in urgent need of treatment. This score not only covers the severity of the injury but also comprehensively considers the physiological state of the injured and the possible impact of environmental factors on rescue, providing a scientific basis for the allocation of medical resources.

[0110] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0111] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods of each embodiment or some parts of the embodiments.

[0112] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments or perform equivalent replacements for some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of each embodiment of the present invention.

Claims

1. An intelligent decision-making and priority allocation system for trauma rescue, characterized in that, Including: A data acquisition module for acquiring original rescue data; A data preprocessing module for preprocessing the original rescue data to obtain preprocessed rescue data; A multi-modal deep learning module that processes the preprocessed rescue data through a deep learning model to obtain shock probability, injury type, and environmental-physiological correlation data; An ISS priority calculation module for calculating the total ISS score priority using the shock probability, injury type, and environmental-physiological correlation data; A priority ranking module for generating a patient priority list based on the total ISS score priority and preferentially allocating resources to the patient with the highest score in the patient priority list.

2. The intelligent decision-making and priority allocation system for trauma rescue according to claim 1, wherein, The original rescue data includes: physiological signal data, injury images, and environmental risk data; The preprocessing of the data preprocessing module includes: Data denoising: Denoising the physiological signal data; Data normalization: Mapping the original rescue data to a preset interval range; Image cropping: Cropping the irrelevant background area in the injury image; Image enhancement: Enhancing the contrast and clarity of the injury image; Data standardization: Standardizing the environmental risk data.

3. The intelligent decision-making and priority allocation system for trauma rescue according to claim 2, characterized in that, The multi-modal deep learning module includes: A physiological signal analysis unit for processing time series data using an LSTM network model to predict shock probability; An imaging injury recognition unit for detecting the injury type and severity within the segmentation area using a CNN model; An environmental risk quantification unit for correlating the environmental risk data with the physiological signal data using a fully connected network model to obtain environmental-physiological correlation data.

4. The intelligent decision-making and priority allocation system for trauma rescue according to claim 3, wherein The specific steps for processing time series data using an LSTM network model to predict shock probability are: Extracting feature vectors from the physiological signal data; Setting the length and step size of the sliding window and processing the feature vectors into time series feature vectors; Constructing an LSTM network model and determining the number of LSTM layers and the number of neurons in each layer; Training the LSTM network model using loss function one and optimizer one; Inputting the time series feature vectors into the trained LSTM network model to obtain the shock probability.

5. The intelligent decision-making and priority allocation system for trauma rescue according to claim 3, characterized in that, The specific steps for detecting the injury type and severity within the segmentation area using a CNN model are: Separating the injury area in the injury image from the background; Using a CNN model to extract injury features from the injury area; Identifying the injury type based on the injury features and evaluating the severity of the injury.

6. The intelligent decision-making and priority allocation system for trauma rescue according to claim 3, wherein The specific steps for correlating environmental risk data with physiological signal data using a fully connected network model are: Determining the number of neurons in the input layer, hidden layer, and output layer in the fully connected network structure; Training the fully connected network model using loss function two and optimizer two; Analyzing the prediction results of the fully connected network model on the validation set, obtaining the deviation between the prediction results of the fully connected network model on the validation set and the true values, and adjusting the feature weights and adding feature interaction terms to correct the deviation between the prediction results on the validation set and the true values to obtain the correlation between environmental risk data and physiological signal data.

7. The intelligent decision-making and priority allocation system for trauma rescue according to claim 2, characterized in that, The ISS priority calculation module includes: The ISS basic score calculation unit calculates the sum of squares based on the highest scores of the injury type and severity to obtain the basic ISS score; The shock risk dynamic correction unit uses the shock probability and the shock risk weight coefficient to calculate and dynamically correct the ISS basic score to obtain the corrected total ISS score; The environmental priority compensation unit uses the environmental risk coefficient obtained after correlating the environmental risk data and the physiological signal data to perform environmental priority compensation on the corrected total ISS score to obtain the compensated total ISS score priority.

8. An intelligent decision-making and priority allocation system for trauma rescue according to claim 7, characterized in that, The steps for calculating the corrected total ISS score are as follows: Calculate the variable ISS score based on the basic ISS score, the shock probability, and the shock risk weight coefficient; Calculate the corrected total ISS score based on the variable ISS score and the basic ISS score.

9. The intelligent decision-making and priority allocation system for trauma rescue according to claim 7, characterized in that, The priority ranking module generates a priority list including: Calculate the comprehensive priority of the patient's trauma severity according to the ISS priority; Sort the patients from high to low according to the patient's comprehensive priority to generate a patient priority list; According to the patient priority list, allocate corresponding medical resources to patients with different priorities to formulate a resource allocation plan; According to the location and traffic conditions of the hospital, plan the optimal transportation route for patients with different priorities.

10. A trauma rescue intelligent decision-making and priority allocation system according to claim 9, characterized in that, The patient's comprehensive priority is calculated as: Obtain the index stability factor S from the physiological signal data; Obtain the transportation time T for the patient to reach the medical institution from the injury location from historical data; Obtain the environmental hazard factor E from the environmental risk data; Allocate weights according to the importance of the stability factor, the transportation time, the environmental hazard factor, and the compensated total ISS score priority in the comprehensive ranking; Calculate the comprehensive priority of the patient through the stability factor, the transportation time, the environmental hazard factor, and the compensated total ISS score priority.

Images