Intelligent assessment decision system for trauma patient emergency

Abstract

The application discloses an intelligent evaluation and decision system for first aid of trauma patients, and particularly relates to the field of medical first aid data processing, and comprises a multi-source fusion module, a type inference module, a path generation module, a model iteration module, and a first aid monitoring device and an image terminal applied in a patient first aid process; the multi-source fusion module is used for acquiring a patient first aid information set through the first aid monitoring device and the image terminal, performing data standardization processing on the patient first aid information set, extracting a feature code set, and constructing an individualized injury atlas; the type inference module is used for solving spatial morphological features and dynamic evolution paths of the individualized injury atlas, performing trauma type inference and pathological factor modeling, and generating a trauma-pathology risk image. Based on dynamic modeling of the patient first aid information set and the trauma-pathology risk image, in combination with resource constraints and physiological state changes, a first aid path is derived and adaptive iteration is performed to construct an intelligent decision system conforming to heterogeneous trauma characteristics.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical emergency data processing, more particularly, to an intelligent assessment and decision system for trauma patient emergency. BACKGROUND

[0002] In the practice of emergency medicine, the injury types of trauma patients are highly heterogeneous, covering blunt trauma, penetrating trauma, burn combined injury and blast combined injury, etc. Different trauma types have significant differences in tissue damage mechanism, blood loss rate, physiological compensation response and immune inflammatory pathway. The existing emergency decision system is generally based on unified parameter setting, which fails to combine patient-specific physiological data and injury characteristics for personalized modeling. In addition, during the accident site and transportation process, there is a lack of information and communication technology (ICT) support based on medical care data processing, which leads to the inability of rescue strategies to adapt to specific trauma patterns in real time, easily causing serious complications such as excessive fluid load, low perfusion delay, airway management failure, etc., ultimately increasing the pre-hospital mortality rate and in-hospital complication rate. Therefore, the current emergency decision mode has the core problem of lacking ICT capability for dynamic processing and decision support based on patient-specific data in the management of heterogeneous trauma patients. SUMMARY

[0003] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide an intelligent assessment and decision system for trauma patient emergency, which dynamically models based on a patient emergency information set and a trauma-pathological risk image, combines resource constraints and physiological state changes to derive an emergency path and perform adaptive iteration, and constructs an intelligent decision system that conforms to the characteristics of heterogeneous trauma, to solve the problems raised in the background art.

[0004] To achieve the above-mentioned purpose, the present application provides the following technical scheme: an intelligent assessment and decision system for trauma patient emergency, comprising a multi-source fusion module, a type inference module, a path generation module, a model iteration module, and an emergency monitoring device and an image terminal applied in the patient emergency process;

[0005] The multi-source fusion module is used to obtain a patient emergency information set through the emergency monitoring device and the image terminal, perform data standardization processing on the patient emergency information set, extract a feature code set, and construct an individualized injury atlas;

[0006] The type inference module is used to solve the spatial morphological characteristics and dynamic evolution path of the individualized injury atlas, perform trauma type inference and pathological factor modeling, and generate a trauma-pathological risk image;

[0007] The path generation module is used to derive a preliminary emergency path graph based on the trauma-pathological risk image, combine the patient emergency information set and the emergency drug inventory information, and perform path adjustment;

[0008] The model iteration module is configured to generate an emergency process log based on the updated emergency path graph and the dynamic change record of the patient emergency information set, extract a model training factor set, and train an emergency strategy model.

[0009] In a preferred embodiment, the multi-source fusion module is configured to acquire data collected by the emergency monitoring device and the image terminal, and form a patient emergency information set including a vital sign sequence, a bioelectric activity curve, and an image segment. The patient emergency information set is subjected to data standardization processing to generate a standardized data matrix.

[0010] The standardized data matrix is subjected to spatio-temporal joint feature extraction, spatial encoding, and time series convolution processing to form a feature encoding set.

[0011] The feature encoding set is subjected to distribution fitting and spatial inference to construct an individualized injury atlas containing a damage region boundary and a type probability distribution.

[0012] In a preferred embodiment, the type inference module is configured to solve spatial morphological features and dynamic evolution paths from the individualized injury atlas, perform specific classification inference, and generate a trauma label set. The trauma label set and a patient historical pathological data set are subjected to interactive modeling to extract a pathological correlation factor and form a risk factor vector.

[0013] The risk factor vector and the trauma label set are subjected to risk weighted mapping to generate a trauma-pathology risk portrait.

[0014] In a preferred embodiment, the path generation module acquires patient injury features and pathology risk parameters from the trauma-pathology risk portrait, combines the patient emergency information set and emergency medicine inventory information, performs reasoning based on resource constraints and physiological urgency, and generates a preliminary emergency path graph.

[0015] The preliminary emergency path graph is combined with the patient emergency information set that changes in real time, and deviation detection and adaptability analysis are performed to solve a path adjustment signal.

[0016] The path adjustment signal is fed back to the preliminary emergency path graph, and dynamic evolution reconstruction is performed to form an updated emergency path graph.

[0017] In a preferred embodiment, the path generation module acquires an emergency operation sequence and a change record of the patient emergency information set from the updated emergency path graph, performs data synchronization acquisition and encryption notarization, and generates an emergency process log.

[0018] The emergency process log is executed to model a causal chain and extract feature factors, a model training factor set for strategy training is solved, the model training factor set is executed to perform adaptive training, and an updated emergency strategy model is constructed for generation and adjustment of an updated emergency path graph.

[0019] In a preferred embodiment, in the multi-source fusion module, definitions are made For an individualized injury atlas, the individualized injury atlas represents an injury distribution function at a spatial position and a time ;

[0020] ;

[0021] wherein:

[0022] ;

[0023] wherein is a vital sign sequence matrix, the vital sign sequence matrix representing a vital sign data set at a position and a time ; is a bioelectric activity curve matrix, the bioelectric activity curve matrix representing a bioelectric signal set at a position and a time ; is an image segment matrix, the image segment matrix representing an image feature set at a position and a time ; is a data energy density map of a modality , represents a modality index set, respectively representing vital signs , bioelectric activities , and image segments three different data sources; represents an integral of vital sign sequence data feature density on , represents a spatial region of vital sign sequence data; represents an integral of bioelectric activity curve data feature density on , represents a spatial region of bioelectric activity curve data; represents an integral of image segment data feature density on , represents a spatial region of image segment data; is an energy density of a modality data; represents a position and a time the joint gradient operator.

[0024] In a preferred embodiment, in the type inference module, define for the trauma-pathology risk profile, the trauma-pathology risk profile characterizes the patient's injury risk intensity at position , time ;

[0025] ;

[0026] ;

[0027] where is the second order partial derivative of the individualized injury map in position direction; is the second order partial derivative of the individualized injury map in time direction; represents the feature vector of the historical pathology data of the th class; is the pathology correlation factor, which is used to measure the contribution of the pathology feature of the historical pathology data of the th class to the trauma risk; is the sum of the historical pathology data; is an exponential function; represents the integral of the energy density of the pathology feature on ; is the spatial region of the historical pathology data of the th class; is the squared two-norm of the historical pathology data; the symbol in the formula is an element-level product.

[0028] In a preferred embodiment, in the path generation module, define as the preliminary first-aid path map:

[0029] ;

[0030] define as the first-aid path adjustment quantity:

[0031] ;

[0032] define as the updated first-aid path map:

[0033] ;

[0034] define as the logarithmic mapping function of the drug inventory information:

[0035] ;

[0036] wherein is a patient emergency information set; is an emergency medicine inventory information matrix, the emergency medicine inventory information matrix representing the medicine inventory and type distribution at location , time ; represents a norm; represents a norm; is a weight coefficient of the risk profile and patient state matching item; is a weight coefficient of the medicine resource constraint item; represents all emergency path graph sets;

[0037] wherein represents a spatial gradient of the path graph time derivative; is a partial derivative of the preliminary emergency path graph with respect to time ; is a time variation gradient of the patient emergency information set; represents the overall spatial definition domain of the patient emergency information set at location .

[0038] In a preferred embodiment, in the path generation module, define as an updated emergency strategy model; define as the emergency strategy model parameter set of the th iteration;

[0039] ;

[0040] wherein is a strategy model training step parameter; represents the gradient operation of the current ; is an emergency process log data definition domain; represents the path update decision set at spatial location , time ; is a time derivative of the patient emergency information set; represents the integral of , accumulating the deviation error of all spatial locations.

[0041] Technical effects and advantages of the present application:

[0042] 1. Based on the dynamic inference of patient emergency information set and trauma-pathology risk image, combined with the characteristics of injury heterogeneity, the individualized emergency path is derived, which solves the problem that the existing decision cannot adapt to different trauma modes, reduces the risk of complications such as abnormal fluid load and low perfusion delay;

[0043] 2. Through standardized processing of data collected by emergency monitoring equipment and image terminal and spatiotemporal feature extraction, a high-dimensional feature code set under a unified feature scale is constructed, and the stability of multi-modal data fusion and the accuracy of injury state representation are improved;

[0044] 3. Through the construction of resource constraint optimization model based on emergency medicine inventory information and patient state change, the available resource state is considered when deriving the emergency path, which improves the executability of the decision and reduces the treatment delay caused by unbalanced resource scheduling;

[0045] 4. Through the collection of emergency path graph and patient emergency information set change record, the causal chain modeling and feature factor extraction are executed, the emergency strategy model is trained, the model adaptive update based on emergency process data is realized, and the dynamic adjustment ability of the strategy is enhanced;

[0046] 5. Based on the injury atlas inference method of spatial and temporal evolution, the key change characteristics in the trauma process are captured, the dynamic representation ability of injury evolution law is improved, and the real-time judgment and intervention decision of emergency personnel are assisted. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 The figure is a schematic diagram of the system module of the present application. DETAILED DESCRIPTION

[0048] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0049] With reference to the drawings in the description, Figure 1 An intelligent assessment and decision system for trauma patient emergency in an embodiment of the present application includes a multi-source fusion module, a type inference module, a path generation module, a model iteration module, and an emergency monitoring device and image terminal applied in the patient emergency process.

[0050] The multi-source fusion module is used to acquire the patient emergency information set through the emergency monitoring device and the image terminal, perform data standardization processing on the patient emergency information set, extract a high-dimensional feature code set, and construct an individualized injury atlas.

[0051] The type inference module is configured to solve the spatial morphological features and dynamic evolution path of the individualized injury atlas, perform trauma type inference and pathological factor modeling, and generate a trauma-pathology risk image.

[0052] The path generation module is configured to infer a preliminary first-aid path based on the trauma-pathology risk image, in combination with the patient first-aid information set and the first-aid drug inventory information, and perform path adjustment.

[0053] The model iteration module is configured to generate a first-aid process log based on the updated first-aid path and the dynamic change record of the patient first-aid information set, extract a model training factor set, and train a first-aid strategy model.

[0054] The multi-source fusion module is configured to acquire data collected by first-aid monitoring equipment and image terminals, and form a patient first-aid information set, which includes a vital sign sequence, a bioelectric activity curve, and an image segment. The patient first-aid information set is subjected to data standardization processing to generate a standardized data matrix.

[0055] The standardized data matrix is subjected to spatio-temporal joint feature extraction, spatial encoding and time series convolution processing to form a high-dimensional feature encoding set.

[0056] The high-dimensional feature encoding set is subjected to distribution fitting and spatial inference to construct an individualized injury atlas containing injury region boundaries and type probability distribution.

[0057] The type inference module is configured to solve the spatial morphological features and dynamic evolution path of the individualized injury atlas, perform specific classification inference, and generate a trauma label set. The trauma label set and the patient historical pathology data set are subjected to interactive modeling to extract pathological correlation factors and form a risk factor vector.

[0058] The risk factor vector and the trauma label set are subjected to risk weighted mapping to generate a trauma-pathology risk image.

[0059] The path generation module acquires patient injury features and pathology risk parameters from the trauma-pathology risk image, combines the on-site patient first-aid information set and the first-aid drug inventory information, performs reasoning based on resource constraints and physiological urgency, and generates a preliminary first-aid path.

[0060] The preliminary first-aid path is combined with the patient first-aid information set that changes in real time to perform deviation detection and adaptability analysis, and the path adjustment signal is solved.

[0061] The path adjustment signal is fed back to the preliminary first-aid path to perform dynamic evolution reconstruction and form an updated first-aid path.

[0062] The path generation module obtains the change record of the first aid operation sequence and the patient first aid information set by updating the first aid path graph, performs data synchronization collection and encryption storage, and generates a first aid process log;

[0063] The first aid process log is executed to model the causal chain and extract the characteristic factor, and the model training factor set for strategy training is solved. The model training factor set is executed to adaptively train an updated first aid strategy model, which is used to generate and adjust the first aid path graph.

[0064] It should be noted that for the formula structure involved in the present scheme, the dimensionless term can be used as a proportional or structural adjustment factor. When combined with quantities with units, it only plays a value scaling role and does not introduce new physical dimensions, so it will not change or confuse the unit system of the whole expression; such combination of "dimensionless term and quantity unit term" can be understood as a complex structure expression form commonly used in mathematical and physical modeling, which conforms to the principle of dimensional consistency and has a clear physical interpretation basis;

[0065] Secondly, in the formula structure of the present scheme, if it involves multiple variable terms with different physical units, including but not limited to time, mass or energy variables, their joint occurrence is to express the cooperative modeling relationship of multiple physical mechanisms. Each variable can be mapped by a function, combined by a ratio, or normalized to form a unified structure. The unit is clear and the meaning is clear. The whole expression conforms to the principle of dimensional consistency and the common norm of engineering modeling;

[0066] In the present scheme, if a constant, weight, adjustment factor, threshold parameter, proportion coefficient, etc. are designed, they are all adjustable control parameters for different application environments. Their values depend on the target device configuration, data input characteristics and performance optimization goals. In the implementation phase, they are set within a reasonable range through model verification, performance constraints or engineering calibration. Although such parameters do not have a unique value, they have a clear adjustment logic and calculation path, and belong to the deterministic setting process in engineering implementation. The purpose of such setting is to ensure that the scheme has both general adaptability and reproducibility and operability, without affecting its technical clarity and implementability;

[0067] In the multi-source fusion module, define as an individualized injury atlas, the individualized injury atlas represents the injury distribution function in the spatial position and time ;

[0068] ;

[0069] Among them:

[0070] ;

[0071] wherein is a vital sign sequence matrix, the vital sign sequence matrix representing a set of vital sign data at position , time ; is an electro-bioactivity curve matrix, the electro-bioactivity curve matrix representing a set of electro-bioactivity data at position , time ; is a video clip matrix, the video clip matrix representing a set of video features at position , time ; is a data energy density map of modality , represents a set of modality indexes, respectively representing three different data sources of vital sign , electro-bioactivity , and video clip ; represents an integral of vital sign sequence data feature density over , represents a spatial region of vital sign sequence data; represents an integral of electro-bioactivity curve data feature density over , represents a spatial region of electro-bioactivity curve data; represents an integral of video clip data feature density over , represents a spatial region of video clip data; represents a natural logarithm function, the natural logarithm function being used for nonlinear compression of feature values in the above equation to avoid model instability caused by too large values; is an energy density of modality data, the energy density of modality data being used to represent a sum of squares of feature components, reflecting signal strength; represents a square of two-norm of modality data plus one, used to ensure that the domain of the logarithm function is positive; represents a joint gradient operator of position , time , the joint gradient operator being used to measure the rate of change of the set of patient emergency information in spatial distribution and time evolution simultaneously; position represents a patient spatial position variable; time represents a patient emergency time variable, the patient emergency time variable being used to define the time evolution dimension in the emergency process.

[0072] In the type inference module, define is a trauma-pathology risk portrait, the trauma-pathology risk portrait characterizing a patient in position , time of injury risk intensity;

[0073] ;

[0074] ;

[0075] wherein is the second-order partial derivative of the individualized injury map in position direction, the second-order partial derivative of the individualized injury map in position direction represents the local change rate of injury distribution along the spatial axis, which is used to depict the spatial diffusion trend of trauma; is the second-order partial derivative of the individualized injury map in time direction, the second-order partial derivative of the individualized injury map in time direction represents the acceleration of injury distribution change with time, which is used to depict the dynamic change characteristics of trauma evolution process; represents the feature vector of the th historical pathological data, the feature vector of the historical pathological data represents a subset of pathological information at position , including but not limited to historical pathological data such as hemorrhage, organ dysfunction, etc.; is a pathological correlation factor, the pathological correlation factor is used to measure the contribution of the pathological characteristics of the th historical pathological data to trauma risk; is the sum of historical pathological data; is an exponential function, the exponential function in the above formula is used to construct an attenuation model of risk factor density, to enhance the specificity of local pathological abnormal area; represents the integral of the energy density of the pathological characteristics at , reflecting the aggregation degree of the pathological factor in the spatial distribution; is the spatial region of the th historical pathological data; is the two-norm square of historical pathological data, representing the sum of squares of each component of historical pathological data, which is used to quantify the pathological intensity; the symbol in the formula is an element-level product, which represents point-by-point multiplication between corresponding elements, i.e. for each position and time , the second-order derivative change rate is multiplied by the pathological risk factor, combining local dynamic change and pathological characteristics.

[0076] In the path generation module, define as the preliminary first-aid path map:

[0077] ;

[0078] Definition Adjustment of first-aid path:

[0079] ;

[0080] Definition Update first-aid path map:

[0081] ;

[0082] Definition Logarithmic mapping function of drug inventory information:

[0083] ;

[0084] Wherein is a patient first-aid information set, the patient first-aid information set including multi-source data such as vital signs, bioelectric activity, and image information; is a first-aid drug inventory information matrix, the first-aid drug inventory information matrix representing the drug inventory and type distribution at a position and a time ; is a preliminary first-aid path map representing an initial path of the first-aid strategy based on risk and resource deduction; represents a one-norm; represents a two-norm; is a weight coefficient of risk portrait and patient state matching item, in actual application the normalized value can be calculated through the real-time gradient difference amplitude of the vital sign sequence in the patient first-aid information set and the trauma-pathological risk portrait; is a weight coefficient of drug resource constraint item, in actual application the resource tightness can be calculated through the spatial distribution change rate of the first-aid drug inventory information, and the normalized value is calculated accordingly; represents a set of all possible first-aid path maps; represents an optimal path map solving operation for minimizing the objective function in the set of all possible first-aid path maps;

[0085] Wherein represents a spatial gradient of path map time derivative, which is used to describe the path change trend; is a partial derivative of the preliminary first-aid path map with respect to the time , the partial derivative of the preliminary first-aid path map with respect to the time representing the change rate of the path evolution with time; is a time change gradient of the patient first-aid information set; represents the patient first-aid information set at a position The overall space domain where the patient is located; a logarithmic mapping function of the drug inventory information For taking logarithmic transformation of the first-aid drug inventory matrix, preventing abnormal interference of extreme inventory values.

[0086] In the path generation module, define For the updated first-aid strategy model, the updated first-aid strategy model is the limit convergence result as the number of iterations tends to infinity; define as the first iteration of the first-aid strategy model parameter set, the first iteration of the first-aid strategy model parameter set represents the strategy parameter after iterations of updates;

[0087] ;

[0088] Wherein is a strategy model training step parameter, the strategy model training step parameter is used to control the amplitude of each iteration update; represents the gradient operation of the current , the gradient operation of the current represents the directional derivative based on the model parameter; is a first-aid process log data domain, the first-aid process log data domain is a time-space range recording the changes of the first-aid path and patient information; represents the path update decision set at the spatial position , the time In the above formula is the time derivative of the patient first-aid information set, the time derivative of the patient first-aid information set represents the patient information change rate at the position ; represents the square error of the path graph and the patient state change, which is used to measure the deviation between the first-aid path and the dynamic physiological state change of the patient; represents the integral of , which accumulates the deviation error of all spatial positions.

[0089] It needs to be overall explained that the intelligent evaluation and decision system for trauma patient first-aid proposed in the application, around the trauma heterogeneity management demand under the complex environment of first-aid scene, through the multi-module hierarchical processing mode, constructs an information communication technology (ICT) system conforming to the medical health data processing logic; the system formation process follows the complete link from multi-source first-aid information collection, to damage inference analysis, to path generation decision and adaptive strategy iteration;

[0090] Specifically, by the multi-source fusion module, the patient emergency information set obtained by the emergency monitoring device and the image terminal is uniformly processed; the information set includes a vital sign sequence, a bioelectric activity curve and an image segment, and covers original information of the patient in physiological, pathological and image aspects; in order to make the data of different sources and modalities comparable and unified, the system performs a data standardization processing step after obtaining the information, normalizes each index to a unified scale, and avoids information deviation caused by different units between modalities; then, by spatio-temporal joint feature extraction, high-dimensional and multi-level feature code sets are extracted by using spatial coding and time series convolution means, so that the dynamic change characteristics of the current physiological state of the patient are maximally retained; based on the feature code set, the system further constructs an individualized injury atlas to accurately depict the distribution and development trend of the injury region in the spatial position and time evolution process of the patient in the emergency scene;

[0091] On the basis of the individualized injury atlas, the system introduces a type inference module to specially process the trauma type identification and pathological risk modeling tasks; by solving the spatial morphological features and dynamic evolution path of the injury atlas, the system infers the specific type of trauma, and performs interactive modeling based on the trauma label set and the patient historical pathological data set; the interactive modeling stage not only focuses on the current injury condition, but also integrates the patient's past pathological feature information such as bleeding condition and organ dysfunction record to establish a risk factor vector, which truly reflects the trauma development possibility and deterioration risk; then the system generates a trauma-pathological risk portrait by performing risk weighted mapping of the risk factor vector and the trauma label set, so that the decision basis changes from a single perspective to a multi-factor comprehensive evaluation, enhancing the adaptability of the emergency decision under different trauma modes;

[0092] Combined with the trauma-pathological risk portrait, the system introduces a path generation module to form the core support of the emergency decision; path generation first constructs an emergency path optimization model based on the risk portrait, combines the patient emergency information set and the emergency medicine inventory information, and deduces a preliminary emergency path graph; this path graph not only considers the current physiological urgency state of the patient, but also takes into account the constraint factors of the available medical resources on site, ensuring that the generated decision path has actual operability and resource availability; after the preliminary path generation, the system further combines the real-time changes of the patient emergency information set to continuously perform deviation detection and adaptability analysis; specifically, by comparing the time change trend of the path graph with the patient state change rate, the path adjustment signal is solved in real time to realize the adaptive evolution of the emergency path in the dynamic environment; finally, the adjustment signal is fed back to the preliminary path graph to complete the dynamic evolution reconstruction of the path, form an updated emergency path graph, and ensure that the emergency process responds in time to the changes in the patient's condition, improving safety;

[0093] In order to support system continuous optimization and self iteration, a model iteration module is designed; the module generates emergency process logs by synchronously collecting dynamic change records of the updated emergency path graph and the patient emergency information set; the log data is solved by the cause-effect chain modeling and the characteristic factor extraction, and a model training factor set for strategy training is obtained; through training, the system can continuously optimize the decision logic based on new data, and construct an updated emergency strategy model; the strategy model has self-adaptive learning ability, can continuously adjust internal parameters with the data accumulated in the emergency process, improve the accuracy and adaptability of the emergency path decision, and realize the continuous evolution of the system.

[0094] The above merely describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An intelligent assessment and decision system for trauma patient emergency treatment, comprising a multi-source fusion module, a type inference module, a path generation module, a model iteration module, and an emergency monitoring device and an image terminal applied in a patient emergency process, characterized in that: the multi-source fusion module is configured to acquire a patient emergency information set through the emergency monitoring device and the image terminal, perform data standardization processing on the patient emergency information set, extract a feature code set, and construct an individualized injury atlas; the type inference module is configured to solve the spatial morphological features and dynamic evolution path of the individualized injury atlas, perform trauma type inference and pathological factor modeling, and generate a trauma-pathological risk portrait; the path generation module is configured to derive a preliminary emergency path graph based on the trauma-pathological risk portrait, in combination with the patient emergency information set and emergency drug inventory information, and perform path adjustment; the model iteration module is configured to generate an emergency process log based on the dynamic change record of the updated emergency path graph and the patient emergency information set, extract a model training factor set, and train an emergency strategy model; In the multi-source fusion module, define For individualized injury atlas, individualized injury atlas Indicates the spatial position And the injury distribution function of time : ； wherein: ； wherein is a vital sign sequence matrix, the vital sign sequence matrix representing a set of vital sign data at position , time ; is a bioelectric activity curve matrix, the bioelectric activity curve matrix representing a set of bioelectric signal at position , time ; is a video clip matrix, the video clip matrix representing a set of video features at position , time ; is a data energy density map of modality , represents a set of modality indices, respectively representing vital sign , bioelectric activity , video clip three different data sources; represents an integral of vital sign sequence data feature density over , represents a spatial region of vital sign sequence data; represents an integral of bioelectric activity curve data feature density over , represents a spatial region of bioelectric activity curve data; represents an integral of video clip data feature density over , represents a spatial region of video clip data; is an energy density of modality data; represents a joint gradient operator at position , time ; the type inference module is configured to solve the spatial morphological features and dynamic evolution path of the individualized injury atlas, and perform specific classification inference to generate a trauma label set; the trauma label set and a patient historical pathological data set are subjected to interactive modeling to extract a pathological correlation factor and form a risk factor vector; the risk factor vector is combined with the trauma label set to perform risk weighted mapping to generate a trauma-pathological risk portrait; In the type inference module, one defines a trauma-pathology risk profile, a trauma-pathology risk profile characterizing the patient's injury risk intensity at a location , at a time ; ； ； wherein is the second partial derivative of the individualized injury map in position direction; is the second partial derivative of the individualized injury map in time direction; denotes the feature vector of the class of historical pathology data; is a pathology correlation factor, the pathology correlation factor being used to measure the contribution of the pathology feature of the class of historical pathology data to the trauma risk; is the sum of the historical pathology data; is an exponential function; denotes the integration of the energy density over the pathology feature; is the spatial region of the class of historical pathology data; is the two-norm square of the historical pathology data; the sign is an element-wise product. 2.The intelligent assessment and decision system for trauma patient emergency treatment according to claim 1, characterized in that: the multi-source fusion module is configured to acquire data collected by the emergency monitoring device and the image terminal, and form a patient emergency information set, the patient emergency information set including a vital sign sequence, a bioelectric activity curve, and an image segment, perform data standardization processing on the patient emergency information set, and generate a standardized data matrix; spatial and temporal joint feature extraction is performed on the standardized data matrix to perform spatial coding and time series convolution processing to form a feature code set; the feature code set is subjected to distribution fitting and spatial inference to construct an individualized injury atlas containing injury region boundaries and type probability distribution. 3.The intelligent assessment and decision system for trauma patient emergency treatment according to claim 2, characterized in that: the path generation module acquires patient injury features and pathological risk parameters through the trauma-pathological risk portrait, combines the on-site patient emergency information set and emergency drug inventory information, performs reasoning based on resource constraints and physiological urgency, and generates a preliminary emergency path graph; the preliminary emergency path graph is combined with the patient emergency information set that changes in real time to perform deviation detection and adaptive analysis to solve a path adjustment signal; the path adjustment signal is fed back to the preliminary emergency path graph to perform dynamic evolution reconstruction to form an updated emergency path graph. 4.The intelligent assessment and decision system for trauma patient emergency treatment according to claim 3, characterized in that: the path generation module acquires an emergency operation sequence and a change record of the patient emergency information set through the updated emergency path graph, performs data synchronous acquisition and encrypted storage, and generates an emergency process log. The emergency process log is executed to model a causal chain and extract feature factors, a model training factor set for strategy training is solved, the model training factor set is executed to perform adaptive training, and an updated emergency strategy model is constructed for generation and adjustment of an updated emergency path graph.

5. The intelligent assessment and decision system for trauma patient emergency according to claim 4, wherein the intelligent assessment and decision system for trauma patient emergency further comprises: In the path generation module, define for the preliminary first-aid path map: ； Definitions Adjustment to the emergency path: ； Definitions To update the first aid path map: ； Definitions Logarithmic mapping function for drug inventory information: ； wherein is a patient emergency information set; is an emergency medicine inventory information matrix, the emergency medicine inventory information matrix representing a distribution of medicine inventory and types at a location , time ; represents a norm; represents a norm; is a weight coefficient of a risk profile and patient state matching item; is a weight coefficient of a medicine resource constraint item; represents a set of all emergency path graphs; wherein denotes the spatial gradient of the path time derivative; is the partial derivative of the preliminary emergency path with respect to time at the point is the time variation gradient of the patient emergency information set; denotes the overall spatial domain in which the patient emergency information set is located at the position .

6. The intelligent assessment and decision system for trauma patient emergency according to claim 5, wherein the intelligent assessment and decision system for trauma patient emergency further comprises: In the path generation module, define an updated first aid strategy model; define a set of first aid strategy model parameters for the first iteration; ； in Train the step size parameters for the policy model; Indicates the current Gradient calculation; Define domains for emergency medical procedure log data; Indicates spatial location ,time The path update decision set; The time derivative of the patient emergency information set; Indicates to Integrate within the space to accumulate the deviation error of all spatial positions.

Images