Informatization process processing method and system for chest pain stroke trauma treatment mode

Abstract

The invention discloses an informatization process processing method for a thoracic pain stroke trauma treatment mode, and the method comprises the following steps: 1, information collection and preprocessing: collecting the physiological sign data of a patient through a mobile terminal, and analyzing the physiological sign data based on a deep learning model, so as to obtain a preliminary diagnosis result of an illness state; classifying and marking the patient according to the preliminary diagnosis result of the illness state to obtain a classification and marking result, and uploading the classification and marking result, the physiological sign data and the geographic position information of the patient to a regional medical collaborative server; step 2, multi-center cooperative distribution: after receiving the classification marking result, the physiological sign data and the geographic position information of the patient, a regional medical cooperative server performs calculation processing based on a preset medical resource scheduling model; and step 3, performing a treatment process: sending the optimal treatment path scheme to a medical information system of a target medical center by the regional medical cooperation server, triggering the target medical center to start an emergency plan, and deploying medical resources.

Description

Technical Field

[0001] This application relates to the technical field of medical informatization process processing, and particularly to an informatization process processing method and system for the treatment mode of chest pain, stroke and trauma. Background Art

[0002] At present, with the acceleration of the urbanization process and the deepening of the aging of the population, the incidence of acute and critical diseases such as chest pain, stroke, and trauma has been increasing year by year. These diseases are characterized by sudden onset, severe illness, rapid progression, and high mortality, and require timely and effective diagnosis and treatment within the golden treatment time. And multi-center linkage first aid networks have been established in large and medium-sized cities, and through the establishment of chest pain centers, stroke centers, and trauma centers, the standardized treatment of acute and critical diseases has been realized.

[0003] In the multi-center treatment mode, it is necessary to integrate multiple medical resources such as pre-hospital first aid, regional medical centers, and specialized treatment centers, and coordinate all parties to complete the timely treatment of patients. The medical information system plays a key role in this process, mainly reflected in: first, realizing the rapid assessment and grading of the patient's condition through informatization means; second, intelligently selecting the optimal treatment center based on real-time data; third, using the information platform to complete the coordinated allocation of medical resources; fourth, sharing and utilizing expert resources through the remote consultation system.

[0004] However, the existing technologies have the following main defects in practical applications:

[0005] 1. The treatment resource allocation mechanism is not scientific enough: The existing technologies mainly allocate treatments based on the geographical location and basic medical capabilities of medical institutions, and fail to fully consider dynamic factors such as the real-time treatment capabilities of medical institutions, the occupancy of beds, and the on-duty status of experts. This leads to problems such as unreasonable allocation of treatment resources and transfer of patients between multiple hospitals in the actual treatment process, delaying the best treatment opportunity.

[0006] 2. The multi-center collaborative treatment process is fragmented: At present, the information systems of each medical institution are relatively independent, lacking a unified information exchange standard and data sharing mechanism. During the multi-center collaborative treatment process, the patient's condition information, examination results, treatment records, etc. cannot be shared in real time, resulting in fragmentation of the treatment process and affecting the treatment effect. At the same time, expert consultations and remote guidance are also inefficient due to information asymmetry.

[0007] 3. The ability to optimize dynamic treatment plans is insufficient: The formulation of existing treatment plans often adopts a preset standardized process, lacking the ability to dynamically optimize according to the real-time changes in the patient's condition and the status of treatment resources. Especially during the patient transfer process, it is impossible to adjust the treatment strategy and transfer route based on the real-time changes in the patient's vital signs, increasing the treatment risk.

[0008] The above-mentioned technical deficiencies seriously affect the treatment effects of critical illnesses such as chest pain, stroke, and trauma. There is an urgent need to develop new technical solutions to solve these problems. Summary of the Invention

[0009] The purpose of this application is to provide an information-based process processing method and system for the treatment mode of chest pain, stroke, and trauma, so as to solve the following main deficiencies existing in the prior art in practical applications: 1. The treatment resource allocation mechanism is not scientific enough; 2. The multi-center collaborative treatment process is fragmented; 3. The ability to optimize dynamic treatment plans is insufficient.

[0010] On the one hand, an information-based process processing method for the treatment mode of chest pain, stroke, and trauma provided by this application adopts the following technical solutions:

[0011] The information-based process processing method for the treatment mode of chest pain, stroke, and trauma includes the following steps:

[0012] Step 1: Information collection and preprocessing: Collect patient physiological sign data through a mobile terminal. The physiological sign data includes electrocardiogram data, blood pressure data, blood oxygen saturation data, and respiratory rate data; analyze the physiological sign data based on a deep learning model to obtain a preliminary diagnosis result of the condition; classify and label the patient according to the preliminary diagnosis result of the condition to obtain a classification and labeling result, and upload the classification and labeling result, the physiological sign data, and the patient's geographical location information to the regional medical collaboration server;

[0013] Step 2: Multi-center collaborative allocation: After receiving the classification and labeling result, the physiological sign data, and the patient's geographical location information, the regional medical collaboration server performs calculation and processing based on a preset medical resource scheduling model. The medical resource scheduling model comprehensively considers the bed capacity, the number of specialist doctors, the status of first-aid equipment, the patient's geographical location information, and the real-time road condition information of each medical center to generate an optimal treatment path plan. The optimal treatment path plan includes the target medical center selection result and the transport route planning result;

[0014] Step 3: Execution of the treatment process: The regional medical collaboration server sends the optimal treatment path plan to the medical information system of the target medical center to trigger the target medical center to start an emergency plan and allocate medical resources; at the same time, send the transport route planning result in the optimal treatment path plan to the ambulance on-board terminal to guide the ambulance to complete the transport according to the planned route; during the transport process, the ambulance on-board terminal real-time collects and uploads the patient's physiological sign data to the medical information system of the target medical center, so that the expert team of the target medical center can real-time master the changes in the patient's condition and conduct remote guidance.

[0015] Furthermore, the information collection and preprocessing also includes:

[0016] 1) Preprocessing of physiological sign data. First, preprocess the physiological sign data such as electrocardiogram, blood pressure, blood oxygen saturation, and respiratory rate collected; ensure the data quality of the input deep learning model through preprocessing;

[0017] 2) Multimodal feature fusion. By designing a multimodal feature fusion network structure, the multimodal feature fusion network structure includes four parallel feature extraction branches: an electrocardiogram branch, a blood pressure data branch, a blood oxygen saturation branch, and a respiratory rate branch; the features extracted by each branch are adaptively weighted and fused through an attention mechanism to generate a unified feature representation;

[0018] 3) Disease diagnosis model. By constructing a three - level diagnosis model system; the first level determines whether it is a critical illness, that is, the critical illness judgment layer, which is implemented using a binary classifier; the second level determines the specific disease category, that is, the disease classification layer, which is completed using a multi - classifier; the third level evaluates the severity of the disease, that is, the severity evaluation layer, which gives a quantitative score through a regression model; the model is trained using real case data from multiple tertiary hospitals, and the generalization ability of the model in different medical scenarios is improved through transfer learning methods;

[0019] 4) Classification label generation. Based on the output results of the diagnosis model, a set of standardized classification label rules is designed; the label information includes three dimensions: urgency level (P1 - P4), disease category code (based on the ICD - 10 coding system), and severity score (0 - 100 points); this multi - dimensional labeling method is not only convenient for the precise allocation of medical resources but also conducive to the formulation of subsequent treatment plans;

[0020] 5) Data encryption and secure transmission. Adopt a hierarchical encryption strategy to protect patients' privacy data; the original physiological sign data is encrypted using the AES - 256 algorithm; the classification label results and geographical location information are encrypted using RSA asymmetric encryption; the SSL / TLS protocol is used during the transmission process to ensure communication security; at the same time, a two - way authentication mechanism is established to ensure that the data can only be accessed by authorized medical institutions;

[0021] 6) Data upload and status synchronization. By designing a breakpoint resumption mechanism, ensure the reliable transmission of data in the case of unstable mobile networks; adopt a chunked transmission strategy to upload large physiological data (such as electrocardiograms) in slices, and each data slice carries a check code; the server side provides real - time feedback on the data reception status to ensure data integrity; at the same time, a data version control mechanism is established to synchronize data to each medical collaboration unit in a timely manner when data is updated.

[0022] Furthermore, the critical illness judgment layer constructs a binary classifier using an improved ResNet-50 network structure to mainly complete the rapid identification of critical illnesses;

[0023] The disease classification layer adopts an ensemble learning method to fuse multiple specialty discriminant models, and the outputs of each specialty model are integrated through a soft voting mechanism to obtain the final disease classification result;

[0024] The severity assessment layer constructs a regression model based on Transformer, quantifies the disease severity into a score value ranging from 0 to 100, and the model adopts a multi-head self-attention mechanism to respectively focus on the abnormal degrees of different physiological indicators.

[0025] Furthermore, the multi-center collaborative allocation also includes:

[0026] 1) Data reception and parsing mechanism. The collaborative server adopts a distributed message queue architecture to establish three independent data reception channels, namely for processing classification marker results, physiological sign data, and geographical location information; through data sharding and merging technology, ensure the complete reception of large-scale physiological sign data; adopt a timestamp synchronization mechanism to achieve the temporal alignment of multi-source data; after data reception, perform integrity verification, and verify the integrity of the data packet through the SHA-256 algorithm to ensure that the information is lossless;

[0027] 2) Data preprocessing and standardization. Uniformly preprocess the received data; first, perform data format standardization to convert data from different sources into a unified JSON format; then perform time series resampling on the physiological sign data with a unified sampling frequency of 100Hz; perform coordinate system conversion and accuracy calibration on the geographical location information, and adopt the WGS84 coordinate system to ensure positioning accuracy;

[0028] 3) Obtain the real-time status of medical resources. By constructing a hierarchical medical resource monitoring network, collect the resource status information of each medical center in real time. The resource status information includes:

[0029] (1) Bed resource layer: Real-time obtain the usage status of general beds, ICU beds, and operating rooms through the hospital information system interface;

[0030] (2) Medical staff layer: Use an intelligent scheduling system to track the on-duty status and workload of specialist doctors and nurses;

[0031] (3) Equipment resource layer: Monitor the operating status and usage of first-aid equipment through the Internet of Things sensing system;

[0032] (4) Drug and consumable layer: Real-time monitor the inventory levels of important drugs and medical consumables;

[0033] 4) Dynamic weight calculation mechanism, which designs an adaptive weight calculation model to dynamically adjust the weight coefficients of various resources according to different disease types:

[0034] (1) For myocardial infarction patients, the weight of the catheterization laboratory equipment status is increased to 0.3, and the weight of the specialist doctor is 0.25;

[0035] (2) For stroke patients, the weight of the CT / MRI equipment status is 0.28, and the weight of the neurospecialist doctor is 0.27;

[0036] (3) For trauma patients, the weight of the operating room status reaches 0.35, and the weight of the emergency doctor is 0.25;

[0037] The weight coefficients are dynamically optimized based on the historical treatment effects through machine learning algorithms, and the weight parameters are updated once a week;

[0038] 5) Multi-objective optimization scheduling algorithm, which uses an improved genetic algorithm to achieve multi-objective optimization scheduling. The algorithm designs three key objective functions: minimizing the patient transfer time, maximizing the medical resource matching degree, and balancing the load of each medical center; by introducing the Pareto optimal solution set, the NSGA-III algorithm is used to solve the multi-objective optimization problem; during the algorithm iteration process, adaptive crossover operators and mutation operators are used to improve the diversity and convergence speed of the solutions;

[0039] 6) Resource reservation strategy, which realizes the dynamic resource reservation mechanism based on the time series prediction model:

[0040] (1) Short-term prediction (0 - 2 hours): The ARIMA model is used to predict the first aid demand;

[0041] (2) Medium-term prediction (2 - 24 hours): Combining the LSTM network to analyze the historical data trend;

[0042] (3) Long-term prediction (1 - 7 days): Considering seasonal factors through the Prophet model;

[0043] According to the prediction results, appropriate resources are reserved for different levels of first aid demands at different time periods to ensure the system's ability to respond to emergencies.

[0044] Furthermore, the multi-objective optimization scheduling algorithm also includes:

[0045] (1) Objective function construction, which includes: the objective function for minimizing transfer time, the objective function for maximizing medical resource matching degree, and the objective function for load balancing;

[0046] (5) Improved NSGA-III algorithm design, the improved NSGA-III algorithm design includes: chromosome encoding scheme, adaptive crossover operator, dynamic mutation operator;

[0047] (6) Reference point generation and fitness evaluation, the reference point generation and fitness evaluation include reference point distribution optimization, fitness evaluation mechanism;

[0048] (7) Elite retention strategy, the elite retention strategy includes archive set update mechanism, environmental selection strategy.

[0049] Furthermore, the execution of the treatment process also includes:

[0050] 1) Medical resource pre-allocation mechanism, including emergency plan trigger process and resource allocation order issuance; the emergency plan trigger process is based on a hierarchical response mechanism, and the system automatically matches the plan level; the resource allocation order issuance adopts a parallel allocation mechanism to synchronously activate multiple resource nodes;

[0051] 2) Transfer route planning and navigation, including dynamic route planning algorithm and navigation information push mechanism; the dynamic route planning algorithm constructs a real-time road network model by combining multi-source data, and at the same time adopts an improved A* algorithm to calculate the optimal transfer route in real time; the navigation information push mechanism designs a hierarchical navigation information display system.

[0052] Furthermore, the plan levels include:

[0053] (1) P1 level: Vital signs are unstable and immediate rescue is required;

[0054] (2) P2 level: The condition is critical and urgent treatment is required;

[0055] (3) P3 level: The condition is stable and timely treatment is required;

[0056] The above system automatically calculates the plan level according to the patient classification marking result and combines with the physiological sign data, and pushes warning information to relevant departments through the medical information system;

[0057] The resource nodes include:

[0058] (4) Medical staff allocation: Push emergency assembly notice through mobile terminal;

[0059] (5) Equipment preparation: Trigger the preheating program of relevant inspection equipment;

[0060] (6) Bed reservation: Lock eligible bed resources;

[0061] (4) Operating room preparation: Start the operating room disinfection procedure and equipment inspection; The above system ensures the orderly execution of various deployment instructions according to the predetermined priority through the workflow engine.

[0062] Furthermore, the multi-source data includes:

[0063] (1) Basic road network data: Road grade, traffic conditions;

[0064] (2) Real-time traffic data: Section congestion index, accident information;

[0065] (3) Weather environment data: Precipitation, visibility influencing factors;

[0066] The evaluation indicators of the transfer route include:

[0067] (3) Time cost: Estimated travel time;

[0068] (4) Safety factor: Road condition score;

[0069] (3) Treatment demand: Adaptability to special road conditions;

[0070] The hierarchical navigation information display system includes:

[0071] (3) Macro level: Overall route overview;

[0072] (4) Middle level: Key section prompts;

[0073] (3) Micro level: Real-time steering guidance.

[0074] On the other hand, an information-based process processing system for the treatment mode of chest pain, stroke and trauma provided by the present application adopts the following technical solutions:

[0075] The information-based process processing system for the treatment mode of chest pain, stroke and trauma includes three core modules, and the three core modules are respectively:

[0076] 1) Pre-hospital first aid module: Equipped with multi-parameter monitoring equipment, adopts a modified chest pain scoring scale, and realizes real-time transmission of vital signs;

[0077] 2) Resource scheduling module: Establish a medical institution database, record the status of beds and equipment, and maintain expert scheduling information;

[0078] 3) Remote consultation module: Provide video conferencing functions, support sharing of examination results and remote consultation.

[0079] Furthermore, the information-based process processing system adopts a three-level architecture design, which includes:

[0080] 1) Data acquisition layer, which acquires patients' vital signs and 12-lead electrocardiogram data through pre-hospital emergency terminals;

[0081] 2) Business processing layer, which sets up a resource scheduling engine and allocates medical resources according to preset rules;

[0082] 3) Application service layer, which provides a mobile emergency platform, a hospital management platform and an expert consultation platform.

[0083] Compared with the prior art, the beneficial effects of this application are as follows:

[0084] 1. Improve the accuracy and scientificity of treatment decisions: Through intelligent analysis of physiological sign data by a deep learning model, accurate preliminary diagnosis of the condition is achieved; based on comprehensive analysis of multi-dimensional data, human judgment bias is reduced; a standardized classification and marking mechanism is adopted to ensure the objectivity of treatment grading.

[0085] 2. Optimize the efficiency of medical resource allocation: Through a medical resource scheduling model, optimal matching of resources is achieved; multi-dimensional resource status such as beds, doctors, and equipment is comprehensively considered to avoid resource misallocation; based on real-time data for dynamic adjustment, resource utilization rate is improved; the number of patient transfers is reduced, and medical resource waste is decreased.

[0086] 3. Shorten the treatment response time: Achieve rapid pre-hospital triage and intelligent allocation; based on the optimal path planning of real-time road conditions, transportation time is reduced; the target medical center starts the emergency plan in advance to shorten the in-hospital preparation time; real-time sharing of information throughout the process improves the treatment efficiency.

[0087] 4. Improve the treatment quality: Achieve full-process dynamic monitoring of patients' vital signs; support remote real-time guidance by expert teams; ensure the continuity and standardization of the treatment process; reduce medical risks during transportation.

[0088] 5. Enhance the multi-center collaboration ability: Establish a unified information sharing platform; achieve collaborative allocation of medical resources; support cross-institutional expert consultations; improve the overall efficiency of the regional emergency network. Description of the Drawings

[0089] Figure 1 is a flowchart of the information-based process processing method for the chest pain, stroke and trauma treatment mode in the embodiment of this application.

[0090] Figure 2 is a flowchart of the information-based process processing system for the chest pain, stroke and trauma treatment mode in the embodiment of this application.

[0091] Figure 3 is a flowchart of the three-level architecture design in the embodiment of this application. Detailed Embodiments

[0092] The following will further elaborate on this application in conjunction with the attached Figures 1-3 drawings.

[0093] The embodiment of this application discloses an information - based process - handling method for the treatment of chest pain, stroke, and trauma. Referring to Figure 1 , in this embodiment, the information - based process - handling method includes the following steps:

[0094] Step 1: Information collection and pre - processing: Collect patients' physiological sign data through a mobile terminal. The physiological sign data includes electrocardiogram data, blood pressure data, blood oxygen saturation data, and respiratory rate data; Analyze the physiological sign data based on a deep - learning model to obtain a preliminary diagnosis result of the condition; Classify and label the patients according to the preliminary diagnosis result of the condition to obtain a classification and labeling result, and upload the classification and labeling result, physiological sign data, and patients' geographical location information to the regional medical collaboration server.

[0095] Specifically, referring to Figure 1 , in this embodiment, the information collection and pre - processing further includes:

[0096] 1) Pre - processing of physiological sign data: First, pre - process the collected physiological sign data such as electrocardiogram, blood pressure, blood oxygen saturation, and respiratory rate; Use wavelet transform to remove baseline drift and high - frequency noise from the electrocardiogram data, and extract features such as QRS complex and ST segment; Perform smoothing filtering on the blood pressure data to obtain stable values of systolic blood pressure and diastolic blood pressure; Perform moving average noise reduction on the blood oxygen saturation and respiratory rate data; Ensure the data quality of the input deep - learning model through pre - processing.

[0097] 2) Multi - modal feature fusion: Design a multi - modal feature fusion network structure. The multi - modal feature fusion network structure includes four parallel feature extraction branches, namely an electrocardiogram branch, a blood pressure data branch, a blood oxygen saturation branch, and a respiratory rate branch; The electrocardiogram branch uses a one - dimensional convolutional neural network to extract temporal features; The blood pressure data branch uses a recurrent neural network to capture dynamic change features; The blood oxygen saturation and respiratory rate branches use a fully - connected layer to extract static features; The features extracted by each branch are adaptively weighted and fused through an attention mechanism to generate a unified feature representation.

[0098] 3) Disease diagnosis model, by constructing a three - level diagnosis model system; the first level determines whether it is a critical illness, i.e., the critical illness judgment layer, which is implemented using a binary classifier; the second level determines the specific disease categories (such as myocardial infarction, stroke, trauma, etc.), i.e., the disease classification layer, which is completed using a multi - classifier; the third level evaluates the severity of the condition, i.e., the severity assessment layer, which gives a quantitative score through a regression model; the model training uses real - case data from multiple top - tier hospitals, and the generalization ability of the model in different medical scenarios is improved through transfer learning methods.

[0099] More specifically, referring to Figure 1 , in this embodiment, the critical illness judgment layer constructs a binary classifier using an improved ResNet - 50 network structure, mainly to complete the rapid identification of critical illnesses. Based on the original ResNet structure, a temporal attention module is added to capture the dynamic change characteristics of physiological indicators. This module calculates the correlation weights of features at different time points and focuses on the mutation of abnormal indicators. The network input is pre - processed multi - modal physiological data, and the output is the probability value of critical illness. When the probability value exceeds a preset threshold (the empirical value is set to 0.85), a critical illness warning is triggered.

[0100] At the same time, the disease classification layer adopts an ensemble learning method to fuse multiple specialist discriminant models, and the outputs of each specialist model are integrated through a soft voting mechanism to obtain the final disease classification result. Specifically, for the diagnosis of myocardial infarction, a special ST - segment analysis module is designed to extract ST - segment features through wavelet transform and combine other electrocardiogram features to achieve accurate identification; for the diagnosis of stroke, a temporal feature extraction network based on LSTM is constructed to focus on analyzing the dynamic change patterns of indicators such as blood pressure and blood oxygen; for trauma assessment, a multi - modal feature fusion network is used to comprehensively analyze the change trend of vital signs.

[0101] In addition, the severity assessment layer constructs a regression model based on Transformer to quantify the severity of the condition into a score value from 0 to 100. The model uses a multi - head self - attention mechanism to respectively focus on the abnormal degrees of different physiological indicators. And the score calculation considers the following key factors:

[0102] ① The deviation degree of each physiological indicator from the normal value;

[0103] ② The duration of abnormal indicators;

[0104] ③ The co - variation relationship between multiple indicators;

[0105] ④ The prognosis of similar patients in historical cases.

[0106] In addition, for transfer learning optimization, a hierarchical transfer learning strategy is adopted to improve the adaptability of the model in different medical scenarios. First, one million standardized case data from 15 tertiary hospitals are used for pre-training the basic model, and then the model is fine-tuned according to the specific conditions of hospitals at different levels and in different regions. In the fine-tuning process, a progressive learning rate adjustment strategy is adopted. A smaller learning rate is used for the underlying feature extraction part to maintain general features, and a larger learning rate is used for the high-level classification part to adapt to local characteristics.

[0107] 4) Classification label generation: Based on the output results of the diagnostic model, a set of standardized classification label rules is designed; the label information includes three dimensions: the degree of urgency (P1 - P4), the disease category code (based on the ICD-10 coding system), and the severity score (0 - 100 points); this multi-dimensional labeling method is not only convenient for the precise allocation of medical resources but also conducive to the formulation of subsequent treatment plans.

[0108] 5) Data encryption and secure transmission: A hierarchical encryption strategy is adopted to protect patients' privacy data; the original physiological sign data is encrypted using the AES-256 algorithm; the classification label results and geographical location information are encrypted using RSA asymmetric encryption; the SSL / TLS protocol is used during the transmission process to ensure communication security; at the same time, a two-way authentication mechanism is established to ensure that the data can only be accessed by authorized medical institutions.

[0109] 6) Data upload and status synchronization: By designing a resume breakpoint mechanism, reliable data transmission is ensured in the case of unstable mobile networks; a chunked transmission strategy is adopted to upload large physiological data (such as electrocardiograms) in pieces, and each data chunk is accompanied by a check code; the server side provides real-time feedback on the data reception status to ensure data integrity; at the same time, a data version control mechanism is established to synchronize data to each medical collaboration unit in a timely manner when data is updated.

[0110] Step 2: Multi-center collaborative allocation: After receiving the classification label results, physiological sign data, and the patient's geographical location information, the regional medical collaboration server performs calculation and processing based on a preset medical resource scheduling model. The medical resource scheduling model comprehensively considers the bed capacity, the number of specialist doctors, the status of first-aid equipment, the patient's geographical location information, and the real-time road conditions of each medical center to generate an optimal treatment path plan. The optimal treatment path plan includes the selection result of the target medical center and the planning result of the transfer route.

[0111] Specifically, referring to Figure 1 , in this embodiment, this multi-center collaborative allocation further includes:

[0112] 1) Data reception and parsing mechanism. The collaborative server adopts a distributed message queue architecture and establishes three independent data reception channels for processing classification marker results, physiological sign data, and geographical location information respectively. Through data sharding and merging technology, it ensures the complete reception of large-scale physiological sign data. By using a timestamp synchronization mechanism, it realizes the temporal alignment of multi-source data. After data reception, integrity verification is carried out, and the integrity of data packets is verified through the SHA-256 algorithm to ensure no loss of information.

[0113] 2) Data preprocessing and standardization. The received data is uniformly preprocessed. First, data format standardization is carried out to convert data from different sources into a unified JSON format. Then, time series resampling is performed on physiological sign data with a unified sampling frequency of 100Hz. For geographical location information, coordinate system conversion and accuracy calibration are carried out, and the WGS84 coordinate system is used to ensure positioning accuracy.

[0114] 3) Real-time status acquisition of medical resources. By constructing a hierarchical medical resource monitoring network, the resource status information of each medical center is collected in real time. More specifically, the resource status information includes:

[0115] (1) Bed resource layer: The usage status of general beds, ICU beds, and operating rooms is obtained in real time through the hospital information system interface.

[0116] (2) Medical staff layer: The intelligent scheduling system is used to track the on-duty status and workload of specialist doctors and nurses.

[0117] (3) Equipment resource layer: The operation status and usage of first-aid equipment are monitored through the Internet of Things perception system.

[0118] (4) Drug and consumable layer: The inventory levels of important drugs and medical consumables are monitored in real time.

[0119] 4) Dynamic weight calculation mechanism. By designing an adaptive weight calculation model, the weight coefficients of various resources are dynamically adjusted according to different disease types. The weight coefficients are dynamically optimized based on historical treatment effects through machine learning algorithms and the weight parameters are updated once a week. More specifically, the weight coefficients include:

[0120] (1) For patients with myocardial infarction, the weight of the catheterization laboratory equipment status is increased to 0.3, and the weight of specialist doctors is 0.25.

[0121] (2) For patients with stroke, the weight of the CT / MRI equipment status is 0.28, and the weight of neurologist doctors is 0.27.

[0122] (3) For trauma patients, the weight of the operating room status reaches 0.35, and the weight of emergency doctors is 0.25.

[0123] 5) The multi-objective optimization scheduling algorithm uses an improved genetic algorithm to achieve multi-objective optimization scheduling. The algorithm designs three key objective functions: minimizing the patient transfer time, maximizing the medical resource matching degree, and balancing the loads of each medical center. At the same time, by introducing the Pareto optimal solution set, the NSGA-III algorithm is used to solve the multi-objective optimization problem. During the algorithm iteration process, adaptive crossover operators and mutation operators are adopted to improve the diversity and convergence speed of the solutions.

[0124] More specifically, referring to Figure 1 , in this embodiment, the multi-objective optimization scheduling algorithm further includes:

[0125] (1) Construction of objective functions; the construction of the objective functions includes: the objective function for minimizing transfer time, the objective function for maximizing medical resource matching degree, and the objective function for load balancing. Among them,

[0126] ① The objective function for minimizing transfer time constructs an evaluation model of transfer time considering multiple factors:

[0127] T = ∑(α1D + α2C + α3J);

[0128] Where: D represents the geographical distance factor, using the actual road network distance; C represents the traffic congestion factor, estimated based on real-time traffic data; J represents the ambulance scheduling delay; α1, α2, and α3 are dynamic weight coefficients obtained through historical data regression.

[0129] ② The objective function for maximizing medical resource matching degree designs a hierarchical resource matching evaluation function:

[0130] M = ∑(β1E + β2S + β3R);

[0131] Where: E represents the equipment matching degree, considering the matching degree between equipment availability and disease requirements; S represents the specialist doctor matching degree, evaluating the compliance between the doctor's professional direction and the disease; R represents the bed resource matching degree, including the adaptability of general beds and ICU beds; β1, β2, and β3 are resource type weights that are dynamically adjusted according to the disease type.

[0132] ③ The objective function for load balancing constructs a load balancing objective using the method of minimizing variance:

[0133] L = min(σ 2 (W1, W2,..., W n ));

[0134] Where: W i represents the standardized workload of the i-th medical center; σ 2 represents the load variance, reflecting the balance degree of the load distribution among each center.

[0135] (2) Design of the improved NSGA-III algorithm; the design of the improved NSGA-III algorithm includes: chromosome coding scheme, adaptive crossover operator, and dynamic mutation operator. Among them,

[0136] ① The chromosome coding scheme designs a three-level hybrid coding structure, which are: the first layer: medical center selection, using integer coding; the second layer: resource allocation scheme, using real number coding; the third layer: time arrangement, using permutation coding.

[0137] ② The adaptive crossover operator designs an adaptive crossover operator based on the distribution characteristics of solutions, which are: when the population distribution is too concentrated, increase the crossover probability to improve diversity; when the population distribution is too dispersed, decrease the crossover probability to accelerate convergence; the adjustment range of the crossover probability Pc is [0.6, 0.9], and it is dynamically adjusted according to the population diversity index.

[0138] ③ The dynamic mutation operator introduces a dynamic mutation mechanism based on the idea of simulated annealing, which are: in the initial stage, use large-scale mutation to explore the solution space; as the iteration progresses, gradually reduce the mutation range for refined search; the mutation probability Pm decays exponentially with the iteration number t: Pm = Pm0 * exp(-λt).

[0139] (3) The reference point generation and fitness evaluation; the reference point generation and fitness evaluation include: reference point distribution optimization, fitness evaluation mechanism; among them

[0140] 1. The reference point distribution optimization includes:

[0141] ① Use the adaptive grid method to generate reference points to ensure that the reference points are evenly distributed in the objective space;

[0142] ② Dynamically adjust the reference point density according to the characteristics of the objective function;

[0143] ③ Increase the reference point density in the high-priority target area to improve the search efficiency.

[0144] 2. The fitness evaluation mechanism designs a fitness evaluation method based on hierarchical distance:

[0145] ① The first level: non-dominated sorting, dividing the Pareto layer;

[0146] ② The second level: calculate the correlation degree between the individual and the reference point;

[0147] ③ The third level: evaluate the position of the individual on the reference line to which it belongs.

[0148] (4) Elite retention strategy; the elite retention strategy includes an archive set update mechanism and an environmental selection strategy; 1. The archive set update mechanism includes;

[0149] ①Maintain an external archive set of fixed size to store non-dominated solutions;

[0150] ②Use crowding distance sorting for archive set pruning;

[0151] ③Use the adaptive grid method to maintain the diversity of solutions.

[0152] 2. The environmental selection strategy includes:

[0153] ①Adopt a tournament selection mechanism to select high-quality individuals;

[0154] ②Introduce a local search mechanism to accelerate the convergence speed;

[0155] ③Design a compensation mechanism to prevent local optimal traps.

[0156] 6) Resource reservation strategy, based on a time series prediction model, to implement a dynamic resource reservation mechanism for medical resources: (1) Short-term prediction (0 - 2 hours): Use the ARIMA model to predict emergency needs; (2) Medium-term prediction (2 - 24 hours): Analyze historical data trends by combining with the LSTM network; (3) Long-term prediction (1 - 7 days): Consider seasonal factors through the Prophet model; According to the prediction results, reserve appropriate amounts of resources for different levels of emergency needs at different time periods to ensure the system's ability to respond to emergencies.

[0157] Step 3. Execution of the treatment process: The regional medical collaboration server sends the optimal treatment path plan to the medical information system of the target medical center, triggering the target medical center to activate the emergency plan and allocate medical resources; at the same time, it sends the transfer route planning result in the optimal treatment path plan to the ambulance on-board terminal to guide the ambulance to complete the transfer according to the planned route; during the transfer process, the ambulance on-board terminal collects and uploads the patient's physiological sign data to the medical information system of the target medical center in real time, enabling the expert team of the target medical center to grasp the patient's condition changes in real time and conduct remote guidance.

[0158] Specifically, referring to Figure 1 , in this embodiment, the execution of the treatment process further includes:

[0159] 1) Medical resource pre-allocation mechanism; The medical resource pre-allocation mechanism includes: an emergency plan trigger process and the issuance of resource allocation instructions. Among them, the emergency plan trigger process is based on a hierarchical response mechanism, and the system automatically matches the plan level, and the plan level includes:

[0160] (1) P1 level: The vital signs are unstable and immediate rescue is required;

[0161] (2) P2 level: The condition is critical and urgent treatment is required;

[0162] (3) P3 level: The condition is stable and timely treatment is required.

[0163] Based on the patient classification and marking results, the above system automatically calculates the pre-plan level in combination with the physiological sign data, and pushes warning information to relevant departments through the medical information system.

[0164] Meanwhile, the allocation instruction of the resources adopts a parallel allocation mechanism to synchronously activate multiple resource nodes, which include:

[0165] (1) Allocation of medical staff: Push emergency assembly notices through mobile terminals;

[0166] (2) Equipment preparation: Trigger the preheating procedure of relevant inspection equipment;

[0167] (3) Bed reservation: Lock the eligible bed resources;

[0168] (4) Operating room preparation: Start the disinfection procedure and equipment inspection of the operating room.

[0169] The above system ensures the orderly execution of each allocation instruction according to the predetermined priority through the workflow engine.

[0170] 2) Transfer route planning and navigation; The transfer route planning and navigation include: dynamic route planning algorithm and navigation information push mechanism. Among them, the dynamic route planning algorithm constructs a real-time road network model by combining multi-source data, and the multi-source data includes:

[0171] (1) Basic road network data: road grade, traffic conditions;

[0172] (2) Real-time traffic data: section congestion index, accident information;

[0173] (3) Weather environment data: influencing factors such as precipitation and visibility.

[0174] Meanwhile, an improved A* algorithm is adopted to calculate the optimal transfer route in real time, and the evaluation indexes of the transfer route include:

[0175] (1) Time cost: estimated driving time;

[0176] (2) Safety factor: road condition score;

[0177] (3) Treatment demand: adaptability to special road conditions.

[0178] In addition, the navigation information push mechanism designs a hierarchical navigation information display system, and the hierarchical navigation information display system includes:

[0179] (1) Macro layer: overall route overview;

[0180] (2) Middle layer: key section prompts;

[0181] (3) Microscopic layer: Real-time steering guidance.

[0182] On the other hand, the embodiment of the present application also discloses an information-based process processing system for the treatment mode of chest pain stroke trauma. Referring to Figure 2 , in this embodiment, the information-based process processing system includes three core modules, which are respectively:

[0183] 1) Pre-hospital emergency module: Equipped with multi-parameter monitoring equipment, using a modified chest pain scoring scale to achieve real-time transmission of vital signs;

[0184] 2) Resource scheduling module: Establish a medical institution database, record the status of beds and equipment, and maintain expert scheduling information;

[0185] 3) Remote consultation module: Provide video conferencing functions, support sharing of examination results and remote consultation.

[0186] At the same time, referring to Figure 3 , in this embodiment, the information-based process processing system adopts a three-level architecture design, which includes:

[0187] 1) Data acquisition layer, the data acquisition layer collects patients' vital signs and 12-lead electrocardiogram data through pre-hospital emergency terminals;

[0188] 2) Business processing layer, the business processing layer sets up a resource scheduling engine and allocates medical resources according to preset rules;

[0189] 3) Application service layer, the application service layer provides a mobile first aid platform, a hospital management platform and an expert consultation platform.

[0190] In addition, specifically, in this embodiment, the data processing processes are as follows: Pre-hospital emergency personnel collect patient information → The system automatically grades → Send a warning to the hospital → Expert remote guidance.

[0191] To sum up, by providing an information-based process processing method and system for the treatment mode of chest pain stroke trauma, the present application has the following technical effects:

[0192] 1. Improve the accuracy and scientificity of treatment decisions: Through intelligent analysis of physiological sign data by a deep learning model, accurate preliminary diagnosis of the condition is achieved; Based on the comprehensive analysis of multi-dimensional data, the deviation of human judgment is reduced; A standardized classification and marking mechanism is adopted to ensure the objectivity of treatment grading.

[0193] 2. Optimize the efficiency of medical resource allocation: Achieve the optimal matching of resources through the medical resource scheduling model; comprehensively consider the multi-dimensional resource status of beds, doctors, equipment, etc. to avoid resource misallocation; dynamically adjust based on real-time data to improve resource utilization; reduce the number of patient transfers and waste of medical resources.

[0194] 3. Shorten the treatment response time: Achieve rapid pre-hospital triage and intelligent allocation; based on the optimal path planning of real-time road conditions to reduce the transfer time; the target medical center starts the emergency plan in advance to shorten the in-hospital preparation time;; real-time sharing of the whole process information to improve the treatment efficiency.

[0195] 4. Improve the treatment quality: Achieve the whole-process dynamic monitoring of the patient's vital signs; support the remote real-time guidance of the expert team; ensure the continuity and standardization of the treatment process; reduce the medical risks during the transfer process.

[0196] 5. Enhance the multi-center collaboration ability: Establish a unified information sharing platform; achieve the collaborative allocation of medical resources; support cross-institutional expert consultations; improve the overall efficiency of the regional first-aid network.

[0197] The above are only the embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structural or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be similarly included in the patent protection scope of the present invention.

Claims

1. An information-based process handling method for the treatment mode of chest pain, stroke and trauma, characterized in that It includes the following steps: Step 1, Information collection and preprocessing: Collect the patient's physiological sign data through a mobile terminal. The physiological sign data includes electrocardiogram data, blood pressure data, blood oxygen saturation data, and respiratory rate data; Analyze the physiological sign data based on a deep learning model to obtain a preliminary diagnosis result of the condition; Classify and label the patient according to the preliminary diagnosis result of the condition to obtain a classification and labeling result, and upload the classification and labeling result, the physiological sign data, and the patient's geographical location information to the regional medical collaboration server; Step 2, Multi-center collaborative allocation: After receiving the classification and labeling result, the physiological sign data, and the patient's geographical location information, the regional medical collaboration server performs calculation and processing based on a preset medical resource scheduling model. The medical resource scheduling model comprehensively considers the bed capacity, the number of specialist doctors, the status of emergency equipment, the patient's geographical location information, and the real-time road condition information of each medical center to generate an optimal treatment path plan. The optimal treatment path plan includes the target medical center selection result and the transport route planning result; Step 3, Execution of the treatment process: The regional medical collaboration server sends the optimal treatment path plan to the medical information system of the target medical center, triggers the target medical center to start an emergency plan, and allocate medical resources; at the same time, sends the transport route planning result in the optimal treatment path plan to the ambulance on-vehicle terminal to guide the ambulance to complete the transport according to the planned route; During the transport process, the ambulance on-vehicle terminal collects and uploads the patient's physiological sign data to the medical information system of the target medical center in real time, so that the expert team of the target medical center can grasp the patient's condition changes in real time and conduct remote guidance.

2. The information-based process processing method for the chest pain stroke trauma treatment mode according to claim 1, wherein, The information collection and preprocessing also include: 1) Physiological sign data preprocessing. First, preprocess the collected physiological sign data such as electrocardiogram, blood pressure, blood oxygen saturation, and respiratory rate; ensure the data quality of the input deep learning model through preprocessing; 2) Multi-modal feature fusion. By designing a multi-modal feature fusion network structure, the multi-modal feature fusion network structure includes four parallel feature extraction branches: an electrocardiogram branch, a blood pressure data branch, a blood oxygen saturation branch, and a respiratory rate branch; the features extracted by each branch are adaptively weighted and fused through an attention mechanism to generate a unified feature representation; 3) Disease diagnosis model. By constructing a three-level diagnosis model system; the first level judges whether it is a critical illness, that is, the critical illness judgment layer, which is implemented by a binary classifier; the second level determines the specific disease type, that is, the disease classification layer, which is completed by a multi-classifier; the third level evaluates the severity of the condition, that is, the severity evaluation layer, which gives a quantitative score through a regression model; the model training uses real case data from multiple top three hospitals, and improves the generalization ability of the model in different medical scenarios through transfer learning methods; 4) Classification marker generation: Based on the output results of the diagnostic model, a set of standardized classification marker rules are designed; the marker information includes three dimensions: urgency level (P1 - P4), disease category code (based on the ICD-10 coding system), and severity score (0 - 100 points); this multi-dimensional marking method is convenient for the precise allocation of medical resources and is also conducive to the formulation of subsequent treatment plans; 5) Data encryption and secure transmission: A hierarchical encryption strategy is adopted to protect patients' privacy data; the original physiological sign data is encrypted using the AES-256 algorithm; the classification marker results and geographical location information are encrypted using RSA asymmetric encryption; the SSL / TLS protocol is used during the transmission process to ensure communication security; at the same time, a two-way authentication mechanism is established to ensure that the data can only be accessed by authorized medical institutions; 6) Data upload and status synchronization: By designing a breakpoint resumption mechanism, reliable data transmission is ensured in case of unstable mobile networks; a chunked transmission strategy is adopted to upload large physiological data (such as electrocardiograms) in slices, and each data slice is accompanied by a check code; the server side provides real-time feedback on the data reception status to ensure data integrity; at the same time, a data version control mechanism is established to synchronize data to each medical collaboration unit in a timely manner when data is updated.

3. The information-based process processing method for the chest pain stroke trauma treatment mode according to claim 2, wherein, The critical illness judgment layer uses an improved ResNet-50 network structure to construct a binary classifier, mainly for the rapid identification of critical illnesses; The disease classification layer uses an ensemble learning method to fuse multiple specialty discriminant models, and the outputs of each specialty model are integrated through a soft voting mechanism to obtain the final disease classification result; The severity assessment layer constructs a regression model based on Transformer, quantifying the disease severity into a score value from 0 to 100 points. The model uses a multi-head self-attention mechanism to focus on the abnormal degrees of different physiological indicators respectively.

4. The information-based process processing method for the chest pain stroke trauma treatment mode according to claim 1, wherein, The multi-center collaborative allocation also includes: 1) Data reception and parsing mechanism: The collaborative server adopts a distributed message queue architecture, establishing three independent data reception channels for processing classification marker results, physiological sign data, and geographical location information respectively; through data slice merging technology, the complete reception of large-scale physiological sign data is ensured; a timestamp synchronization mechanism is used to achieve the temporal alignment of multi-source data; after data reception, integrity verification is performed, and the integrity of the data packet is verified through the SHA-256 algorithm to ensure that the information is lossless; 2) Data preprocessing and standardization: The received data is uniformly preprocessed; first, data format standardization is carried out to convert data from different sources into a unified JSON format; then, time series resampling is performed on the physiological sign data, with a unified sampling frequency of 100Hz; coordinate system conversion and accuracy calibration are performed on the geographical location information, and the WGS84 coordinate system is used to ensure positioning accuracy; 3) Real-time acquisition of medical resource status: By constructing a hierarchical medical resource monitoring network, the resource status information of each medical center is collected in real time. The resource status information includes: (1) Bed resource layer: The usage status of general beds, ICU beds, and operating rooms is obtained in real time through the hospital information system interface; (2) Medical staff: Use an intelligent scheduling system to track the on-the-job status and workload of specialists and nurses; (3) Equipment resource layer: monitor the operating status and usage of emergency equipment through the IoT sensing system; (4) Drug and consumables layer: real-time monitoring of inventory levels of important drugs and medical consumables; 4) Dynamic weight calculation mechanism: By designing an adaptive weight calculation model, the weight coefficients of various resources are dynamically adjusted according to different disease types: (1) For patients with myocardial infarction, the weight of catheterization laboratory equipment status was increased to 0.3, and the weight of specialist physicians was 0.25; (2) For stroke patients, the weight of CT / MRI equipment status was 0.28, and the weight of neurologist was 0.27; (3) For trauma patients, the weight of operating room status is 0.35, and the weight of emergency physician is 0.25; The weight coefficient is dynamically optimized based on historical treatment effects through a machine learning algorithm, and the weight parameter is updated once a week; 5) Multi-objective optimization scheduling algorithm, using an improved genetic algorithm to achieve multi-objective optimization scheduling. The algorithm designs three key objective functions: minimizing patient transfer time, maximizing medical resource matching, and balancing the load of each medical center; by introducing the Pareto optimal solution set, the NSGA-III algorithm is used to solve the multi-objective optimization problem; in the algorithm iteration process, the adaptive crossover operator and mutation operator are used to improve the diversity of solutions and the convergence speed; 6) Resource reservation strategy, based on the time series prediction model, to realize the dynamic reservation mechanism of medical resources: (1) Short-term forecast (0-2 hours): ARIMA model is used to predict emergency needs; (2) Medium-term forecast (2-24 hours): Analyze historical data trends in combination with LSTM networks; (3) Long-term forecast (1-7 days): seasonal factors are taken into account through the Prophet model; Based on the forecast results, appropriate resources are reserved for different levels of emergency needs in different time periods to ensure the system's ability to respond to emergencies.

5. The information-based process processing method for the chest pain stroke trauma treatment mode according to claim 4, characterized in that, The multi-objective optimization scheduling algorithm also includes: (1) constructing an objective function, wherein the objective function construction includes: an objective function of minimizing the transport time, an objective function of maximizing the medical resource matching degree, and an objective function of load balancing; (2) an improved NSGA-III algorithm design, wherein the improved NSGA-III algorithm design includes: a chromosome encoding scheme, an adaptive crossover operator, and a dynamic mutation operator; (3) Reference point generation and fitness evaluation, which includes reference point distribution optimization and fitness evaluation mechanism; (4) Elite retention strategy, which includes an archive set update mechanism and an environment selection strategy.

6. The information-based process handling method for the treatment model of chest pain stroke trauma according to claim 1, wherein, The rescue process also includes: 1) Medical resource pre-allocation mechanism, including emergency plan triggering process and resource allocation instruction issuance; the emergency plan triggering process is based on a hierarchical response mechanism, and the system automatically matches the plan level; the resource allocation instruction issuance adopts a parallel allocation mechanism to simultaneously activate multiple resource nodes; 2) Transfer route planning and navigation, including dynamic route planning algorithms and navigation information push mechanisms; the dynamic route planning algorithm constructs a real-time road network model by integrating multi-source data, and at the same time adopts an improved A* algorithm to calculate the optimal transfer route in real time; the navigation information push mechanism designs a hierarchical navigation information display system.

7. The information-based process processing method for the chest pain stroke trauma treatment mode according to claim 6, characterized in that, The pre-plan levels include: (1) P1 level: Unstable vital signs, immediate rescue required; (2) P2 level: Critically ill condition, urgent treatment required; (3) P3 level: Stable condition, timely treatment required; The above system automatically calculates the pre-plan level based on the patient classification and marking results, combined with physiological sign data, and pushes warning information to relevant departments through the medical information system; The resource nodes include: (1) Medical staff allocation: Push emergency assembly notices through mobile terminals; (2) Equipment preparation: Trigger the preheating program of relevant inspection equipment; (3) Bed reservation: Lock eligible bed resources; (4) Operating room preparation: Start the operating room disinfection program and equipment inspection; The above system ensures the orderly execution of various allocation instructions according to the predetermined priority through the workflow engine.

8. The information-based process processing method for the chest pain stroke trauma treatment mode according to claim 6, characterized in that, The multi-source data includes: (1) Basic road network data: Road grade, traffic conditions; (2) Real-time traffic data: Road section congestion index, accident information; (3) Weather environment data: Precipitation, visibility influencing factors; The evaluation indicators of the transfer route include: (1) Time cost: Estimated travel time; (2) Safety factor: Road condition score; (3) Treatment requirement: Adaptability to special road conditions; The hierarchical navigation information display system includes: (1) Macro layer: Overall route overview; (2) Middle layer: Key section prompts; (3) Micro layer: Real-time steering guidance.

9. An information-based process processing system for the treatment mode of chest pain, stroke and trauma, according to the information-based process processing method for the treatment mode of chest pain, stroke and trauma described in any one of claims 1-8, characterized in that, It includes three core modules, and the three core modules are respectively: 1) Pre-hospital emergency module: Equipped with multi-parameter monitoring equipment, adopts a modified chest pain scoring scale to realize real-time transmission of vital signs; 2) Resource scheduling module: Establish a medical institution database, record the status of beds and equipment, and maintain expert scheduling information; 3) Remote consultation module: Provide video conferencing functions, support sharing of inspection results and remote consultation.

10. The information-based process processing system for the treatment mode of chest pain and stroke trauma according to claim 9, wherein The information-based process processing system adopts a three-level architecture design, which includes: 1) Data collection layer, the data collection layer collects patient vital signs and 12-lead electrocardiogram data through pre-hospital emergency terminals; 2) Business processing layer, the business processing layer sets a resource scheduling engine, and allocates medical resources according to preset rules; 3) Application service layer, the application service layer provides a mobile first-aid platform, a hospital management platform and an expert consultation platform.

Images