Emergency treatment batch patient shunting data processing method, system and equipment and medium

Abstract

The invention relates to the technical field of emergency treatment shunting, and provides an emergency treatment batch patient shunting data processing method, system and device, and a medium, and the method comprises the steps: after an emergency plan is started, loading a hospital three-dimensional model, distinguishing a treatment space from a non-treatment space, taking the non-treatment space as a shunting candidate position, and building a multi-channel dynamic data matrix. Patient information and initial illness state grades are obtained through voice input, treatment priorities are formed, and initial shunting positions are distributed to patients according to the priorities and the shortest passing path. In addition, vital signs, medicine and treatment parameters of the patients are continuously monitored, emergency patients are immediately marked and alarmed when indexes are abnormal, and the nearest idle rescue space is immediately selected for rematching, so that orderly distribution of conventional patients and rapid rescue of critical patients are realized, and emergency rescue scheduling efficiency and safety are improved.

Description

Technical Field

[0001] This invention relates to the field of emergency triage technology, and more specifically, to a method, system, device, and medium for processing emergency batch patient triage data. Background Technology

[0002] The content in this section only provides background information related to this invention and may not constitute prior art.

[0003] In the field of emergency medicine, mass emergencies can easily lead to a surge of patients into hospitals. Efficient and accurate patient triage is a core element in ensuring orderly treatment and improving the success rate of rescue, directly determining the emergency response capacity of emergency medicine. Currently, the triage of large numbers of emergency patients in domestic hospitals still relies mainly on manual operation and experience-based judgment by medical staff, and a standardized, digital triage data processing system has not yet been formed, making it difficult to meet the needs of efficient handling when large numbers of patients flood in.

[0004] The existing triage model has revealed many problems in practical application: the lack of precise functional division and digital management of hospital physical space makes it impossible to quickly and reasonably delineate temporary triage areas, which can easily lead to situations where core emergency space is occupied by patients with mild symptoms and temporary placement locations are selected in an disorderly manner, resulting in low efficiency of medical space resource utilization; triage information collection mostly adopts handwriting and manual entry methods, which are not only extremely inefficient in chaotic on-site scenarios, but also easily lead to problems such as mismatch of patient information and subjective classification of the severity of illness, making it difficult to ensure priority treatment for critically ill patients; patient transfer route planning only considers geometric straight-line distance without taking into account the hospital building structure, the physical attributes of the passage, and real-time traffic conditions, resulting in a lack of practical operability in the calculated routes and a significant increase in transfer time and costs.

[0005] Meanwhile, traditional triage is a static, one-time placement model. After the initial triage is completed, there is a lack of continuous dynamic monitoring of the patient's vital signs and treatment-related data, making it impossible to capture sudden signals of deterioration in the patient's condition in a timely manner. Even if abnormalities are detected, it is difficult to quickly screen and match available core resuscitation spaces, resulting in a serious disconnect between triage and emergency resuscitation, delaying the treatment of critically ill patients.

[0006] In summary, existing technologies for triaging large numbers of emergency patients suffer from multiple shortcomings, including unscientific space utilization, inefficient and inaccurate information processing, unrealistic route planning, and a lack of dynamic monitoring and emergency rescue coordination mechanisms. There is an urgent need for a systematic data processing method for triaging large numbers of emergency patients to address the technical problems of unreasonable triage space planning, difficulty in standardizing triage information processing, lack of accurate calculation of transfer routes, insufficient dynamic monitoring of patient conditions, and difficulty in rapidly matching emergency rescue spaces in large patient scenarios. Summary of the Invention

[0007] The purpose of this invention is to provide a method, system, device, and medium for processing emergency department patient triage data in batches, in order to improve the aforementioned problems. To achieve the above objective, the technical solution adopted by this invention is as follows: Firstly, this application provides a method for processing emergency department batch patient triage data, including: In response to the emergency plan activation command, the system loads a 3D model of the hospital's interior and obtains the attribute labels for each spatial unit in the 3D model. The attribute labels include treatment attribute labels and non-treatment attribute labels. Spatial units carrying non-treatment attribute labels are marked as candidate triage locations. A dynamic matrix is ​​constructed, with each row of the dynamic matrix corresponding to an activated patient number and each column of the dynamic matrix corresponding to a data acquisition channel. The data acquisition channels include a voice input channel, a vital signs monitoring channel, a drug traceability channel, and a device treatment parameter channel. The system receives voice streams uploaded by medical staff through the voice input channel, extracts the patient number and initial severity level from the voice stream, writes the patient number into a new row of the dynamic matrix, and converts the initial severity level into a first priority weight and stores it in the corresponding column of the new row. Extract the spatial coordinates of each candidate triage location and calculate the path length; sort the activated patient numbers from high to low according to the first priority weight to obtain the allocation order; according to the allocation order, select the candidate triage location that is currently unoccupied and has the shortest path length for each patient in turn to generate the initial matching record; Continuously acquire real-time data streams from the vital signs monitoring channel, drug traceability channel, and equipment treatment parameter channel. The real-time data streams carry patient numbers and monitoring values. Associate fluctuation segments of the corresponding indicators in the monitoring values ​​that exceed the preset range with the corresponding patient numbers, mark the patient numbers as emergency numbers, and trigger alarms. Extract unoccupied rescue space units from the space units carrying treatment attribute tags; calculate the path distance between the real-time location corresponding to the emergency number and each unoccupied rescue space unit, select the rescue space unit with the smallest path distance as the new matching location, and output the diversion matching result.

[0008] Furthermore, the steps for loading the 3D model of the hospital's interior specifically include: The system calls upon the building structure data stored in the building information modeling system. The building structure data includes floor plans, wall partition lines, and room function labels. A three-dimensional mesh map is constructed based on the building structure data, and each mesh cell in the three-dimensional mesh map is bound to the room function label to generate attribute labels for each spatial cell in the three-dimensional model.

[0009] Furthermore, the steps for extracting the patient ID and initial severity rating from the speech stream specifically include: The speech stream is input into a preset speech recognition model and converted into text data; named entity recognition is performed on the text data to extract the number field associated with the patient's identity as the patient number; medical entity recognition is performed on the text data to extract keywords describing the injury or symptoms. Based on a pre-defined severity grading mapping library, the extracted keywords are matched with the grading levels. The mapping library stores the association between different symptom keywords and their corresponding severity levels. The severity levels are divided according to the urgency of the illness indicated by the symptoms and are pre-assigned different level values. When the extracted keywords correspond to multiple severity levels, the highest level is selected as the initial severity classification.

[0010] Furthermore, the steps for calculating the length of the travel path specifically include: In the 3D model and its corresponding coordinates, walls are used as insurmountable constraints within the same plane to construct walkable areas on the same level. For each candidate triage location and the current real-time location of each activated patient, determine whether the candidate triage location to be calculated and the real-time location are on the same floor; If they are on the same floor, the shortest distance to bypass the wall obstacle is calculated using a path search algorithm within the walkable area of ​​that floor, and this distance is taken as the path length. If the locations are on different floors, a path search algorithm is used to calculate the first path distance from the current real-time location to each stair entrance on the current floor, and the second path distance from each stair exit on the floor where the candidate diversion location is located to the candidate diversion location. Then, the fixed vertical transition distance of the corresponding stair passage between the stair entrance and the stair exit is obtained. The first path distance, the second path distance, and the fixed vertical transition distance corresponding to the same stair are added together, and the shortest distance after the addition is selected as the passage path length.

[0011] Furthermore, the pathfinding algorithm employs a heuristic search strategy based on physical travel cost, specifically including: In the 3D model, the walkable area is discretized into a topological network consisting of nodes and edges. Nodes correspond to the center point of a spatial unit or the intersection of a passage, and edges correspond to the passage path segments connecting two nodes. Each edge is assigned a basic length value, and a dynamic passage cost is generated based on the physical properties of the spatial unit or passage unit in which the edge is located. The physical properties include at least the passage width, passage direction restrictions, and real-time patient density. Starting from the patient's current real-time location node and targeting candidate triage locations, a heuristic search algorithm is used to search for a path. During the search, the actual travel cost from the starting point to the current node is the sum of the dynamic travel costs of all edges traversed, and the estimated travel cost from the current node to the target point is the Euclidean distance between the two points divided by the preset maximum travel speed. The node with the smallest sum of actual and estimated travel costs is selected for expansion until the target point is found. The output path is formed by connecting nodes and travel edges in sequence.

[0012] Furthermore, the dynamic toll cost is calculated using the following preset formula: In the formula, The dynamic passage cost of an edge; This represents the basic length value of the edge; The actual channel width of the edge; This is the standard reference width; This is the width influence coefficient; The direction inverse coefficient; This is the directional influence coefficient; This represents the real-time patient density in the area where the edge is located. The patient density influence coefficient.

[0013] Furthermore, the step of extracting unoccupied rescue space units from space units carrying treatment attribute tags specifically includes: Traverse all spatial units with treatment attribute tags in the 3D model, obtain the current occupancy status of each spatial unit, filter out the spatial units that are currently vacant, and use them as unoccupied rescue spatial units.

[0014] Secondly, this application also provides an emergency bulk patient triage data processing system, including: The spatial candidate labeling module is used to load a 3D model of the hospital in response to the emergency plan activation command, and obtain the attribute labels of each spatial unit in the 3D model. The attribute labels include treatment attribute labels and non-treatment attribute labels; the spatial units carrying non-treatment attribute labels are marked as candidate diversion locations. The dynamic matrix construction module is used to construct a dynamic matrix. Each row of the dynamic matrix corresponds to an activated patient number, and each column of the dynamic matrix corresponds to a data acquisition channel. The data acquisition channels include a voice input channel, a vital signs monitoring channel, a drug traceability channel, and a device treatment parameter channel. The voice grading assignment module is used to receive the voice stream uploaded by medical staff through the voice input channel, extract the patient number and initial severity grade from the voice stream, write the patient number into a new row of the dynamic matrix, and convert the initial severity grade into the first priority weight and store it in the corresponding column of the new row. The path triage module is used to extract the spatial coordinates of each candidate triage location and calculate the path length; sort the activated patient numbers from high to low according to the first priority weight to obtain the allocation order; according to the allocation order, select the candidate triage location that is currently unoccupied and has the shortest path length for each patient in turn to generate the initial matching record; The real-time monitoring and alarm module is used to continuously acquire real-time data streams from the vital signs monitoring channel, drug traceability channel, and equipment treatment parameter channel. The real-time data stream carries the patient number and monitoring value. It associates the fluctuation segments of the corresponding indicators in the monitoring value that exceed the preset range with the corresponding patient number, marks the patient number as an emergency number, and triggers an alarm. The emergency matching module is used to extract unoccupied emergency space units from the space units carrying treatment attribute tags; calculate the path distance between the real-time location corresponding to the emergency number and each unoccupied emergency space unit; select the emergency space unit with the smallest path distance as the new matching location; and output the diversion matching result.

[0015] Thirdly, this application also provides an electronic device, including: Memory, used to store computer programs; A processor is used to implement the method steps as described in the first aspect when executing a computer program.

[0016] Fourthly, this application also provides a readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, it implements the method steps of the first aspect.

[0017] The beneficial effects of this invention are as follows: After the emergency plan is activated, this invention first loads a three-dimensional spatial model of the hospital and distinguishes between treatment and non-treatment spatial units. Non-treatment spaces are designated as candidate locations for patient triage. Simultaneously, a multi-channel dynamic data matrix is ​​constructed, encompassing voice input, vital sign monitoring, drug traceability, and equipment treatment parameter collection. Patient information and initial condition classification are obtained through the voice channel and converted into treatment priorities. Candidate triage locations are assigned to patients according to priority and the principle of shortest travel path to achieve initial orderly triage. During the process, real-time monitoring data of patients is continuously collected through other data channels. Once abnormal fluctuations in relevant indicators are detected, the corresponding patient is marked as an emergency and an alarm is triggered. Unoccupied professional resuscitation spaces are then retrieved, and the optimal resuscitation location is re-matched for emergency patients according to the shortest path. This not only achieves standardized initial triage of large numbers of patients but also enables rapid allocation of dedicated resuscitation spaces for patients whose conditions suddenly deteriorate, improving the scheduling efficiency and emergency response capabilities of hospital patient treatment in emergency scenarios. Attached Figure Description

[0018] Figure 1 A flowchart of the emergency department batch patient triage data processing method provided by the present invention; Figure 2 This is a data flow diagram of dynamic matrix construction and multi-channel data acquisition in this invention; Figure 3 This is a diagram illustrating the calculation of travel path data and the initial traffic matching data processing in this invention. Figure 4 A schematic diagram of the emergency batch patient triage data processing system provided by the present invention; Figure 5 This is a schematic diagram of an electronic device provided by the present invention.

[0019] In the diagram: 201, Spatial candidate labeling module; 202, Dynamic matrix construction module; 203, Voice hierarchical assignment module; 204, Path diversion module; 205, Real-time monitoring and alarm module; 206, Rescue matching module; 301, Processor; 302, Memory. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0021] like Figure 1 As shown in the embodiment of the present invention, a method for processing emergency batch patient triage data includes: S101, in response to the emergency plan activation command, loads the three-dimensional model inside the hospital and obtains the attribute label of each spatial unit in the three-dimensional model. The attribute label includes treatment attribute label and non-treatment attribute label; the spatial unit carrying the non-treatment attribute label is marked as a candidate diversion location.

[0022] Specifically, in emergency situations where a large number of patients flood in, the activation of the emergency plan signifies the formal commencement of the standardized triage process. The system is immediately activated to prevent disorderly triage operations. Subsequently, a 3D model of the hospital's interior is loaded. The core principle is to overcome the spatial limitations of traditional two-dimensional drawings and rely on a 3D visualization platform to fully recreate the hospital's physical spatial structure, achieving precise spatial location and providing an intuitive and accurate spatial basis for subsequent triage location selection. Next, the attribute labels of each spatial unit in the 3D model are obtained. These attribute labels are clearly divided into two categories: treatment attribute labels and non-treatment attribute labels. The hospital space is finely categorized by functional attributes. The treatment attribute tag corresponds to the core medical space with emergency treatment and rescue functions, while the non-treatment attribute tag corresponds to the transitional and resettlement space without core treatment capabilities. The strict distinction between the two can ensure that core treatment resources are not occupied by the diversion of mild cases. Subsequently, the space units with the non-treatment attribute tag will be marked as candidate diversion locations. The principle is that in the case of a large number of emergency patients, temporary diversion areas should be planned for patients with relatively stable conditions. By marking non-treatment spaces in advance, vacant areas suitable for large-scale resettlement can be quickly screened out, improving the efficiency of diversion site selection.

[0023] The detailed process of loading the hospital's internal 3D model first involves calling the building structure data stored in the Building Information Modeling (BIM) system. This data includes floor plans, wall partitions, and room function labels. The principle is to rely on the standardized structured data of the BIM system to ensure that the 3D model is completely consistent with the actual physical space of the hospital, eliminating spatial data deviations. Then, a 3D mesh map is constructed based on the building structure data, and each mesh unit in the 3D mesh map is bound to a room function label. This generates attribute labels for each spatial unit in the 3D model. The principle is to divide the overall space of the hospital into the smallest independent spatial units through mesh discretization, so that each spatial unit is bound to a unique functional attribute, achieving precise assignment of spatial attributes. For example, after a top-tier general hospital activates its emergency plan for a group of patients affected by a landslide, the system immediately accesses the floor plan, wall partitions, and clinic function identification data from the hospital's emergency building building information model system. It quickly constructs a three-dimensional grid map of the emergency building, marking spaces such as the waiting hall, public corridors, and temporary observation cubicles as non-treatment attribute labels and simultaneously designating them as candidate triage locations. It also marks the emergency resuscitation room and intensive care unit as treatment attribute labels, completing the initial spatial preprocessing and laying a solid foundation for subsequent triage matching.

[0024] S102, construct a dynamic matrix. Each row of the dynamic matrix corresponds to an activated patient number, and each column of the dynamic matrix corresponds to a data acquisition channel. The data acquisition channels include a voice input channel, a vital signs monitoring channel, a drug traceability channel, and a device treatment parameter channel.

[0025] Specifically, its implementation method is as follows: Figure 2 As shown, the data acquisition channels are initialized in the column dimension through channel type definition, and the batch of emergency patients are initialized in the row dimension through patient number activation. After the row and column dimension initializations are synchronized, a dynamic matrix is ​​constructed. Subsequently, multi-channel data is aggregated to form patient-specific data rows. The core data carrier framework is built using standardized initialization of the matrix's row and column dimensions. Each row of the dynamic matrix corresponds to an activated patient number. The principle is that the matrix row dimension serves as the exclusive data carrier for a single patient. For triage scenarios involving the unified placement and individual numbering of batch emergency patients, this ensures that each patient's entire triage process data has an independent collection unit, fundamentally avoiding cross-contamination of multiple patient data. This achieves a precise one-to-one binding between patients and data, ensuring the targeted nature of subsequent data retrieval and triage matching, and preventing triage errors caused by data mismatch.

[0026] Simultaneously, each column of the dynamic matrix corresponds to a data acquisition channel. The principle is to divide data acquisition dimensions from different sources by matrix column dimensions, achieving classification and organization of multiple data types, eliminating the drawbacks of scattered storage, and allowing various types of data to be orderly categorized according to the acquisition path. The data acquisition channels specifically include a voice input channel (for input), a vital signs monitoring channel, a drug traceability channel, and a device treatment parameter channel. The principle is to comprehensively cover the core data needs of the entire process of triage for batch emergency patients, including voice input data from the initial triage stage, as well as real-time data from subsequent dynamic monitoring of the condition and control of the treatment process, forming a full-process, multi-dimensional data acquisition system. This achieves comprehensive collection of triage data, taking into account the data support needs of initial triage site selection and subsequent emergency alarms and resuscitation space allocation, improving the data integrity of the overall triage process. For example, when a top-tier hospital receives a large number of emergency patients due to a group of workplace injuries, after activating the emergency plan and completing the initial spatial preprocessing, it immediately constructs this dynamic matrix. Each time a patient's number is activated, a new row corresponding to that patient's number is added to the matrix. Simultaneously, four fixed data channels are defined: voice input, vital sign monitoring, medication traceability, and equipment treatment parameters. Subsequently, during the patient's triage, the voice input parameters of medical staff, vital sign monitoring parameters such as ECG and blood pressure, on-site emergency medication traceability parameters, and emergency treatment equipment parameters (i.e., the treatment parameters of the equipment used when administering treatment) are all collected into that row according to the corresponding channel column. This achieves integrated storage and real-time updates of comprehensive triage data for a single patient, improving data processing efficiency and accuracy.

[0027] S103: Receive the voice stream uploaded by medical staff through the voice input channel, extract the patient number and the initial severity classification from the voice stream, write the patient number into a new row of the dynamic matrix, and convert the initial severity classification into the first-priority weight value and store it in the corresponding column of the new row.

[0028] Specifically, in the emergency scenario where a large number of emergency patients suddenly flood into the hospital, to avoid chaos at the scene, chaos in triage information transmission, and cross-confusion of multi-patient data, the hospital will first temporarily place all patients to be triaged in a dedicated triage area at the emergency entrance. The on-site medical staff will assign a unique activated patient number to each patient in the order of arrival, completing the preliminary independent identification of the patient's identity. Subsequently, the attending doctor will quickly screen and judge the injuries, symptoms, and urgency of each patient on-site. At this time, the doctor uses a communication device to achieve rapid collection, accurate identification, and standardized assignment of triage information through voice interaction, fully meeting the requirements of high efficiency and accuracy in emergency response. The specific implementation method is as follows: First, receive the voice stream uploaded by medical staff through the mobile phone. This operation addresses the actual pain point that in a large-scale emergency scenario, the doctor's hands need to be engaged in the preliminary treatment of patients and cannot perform handwritten registration or manual keyboard input. Relying on the dedicated voice input data channel supporting the dynamic matrix, it realizes the non-contact and real-time upload of triage information. Thus, it abandons the inefficient mode of traditional manual paper records and manual input, adapts to the characteristics of a noisy emergency scene and a compact triage rhythm, and does not require medical staff to be distracted by operating input devices. Only through voice expression can the information be uploaded, ensuring the smoothness of the emergency triage work; Next, the patient ID and initial severity level are extracted from the speech stream. This extraction is not simply speech-to-text processing, but rather a multi-level text recognition technology that accurately filters core triage information. Specifically, it involves three interconnected technical steps: First, the acquired speech stream is input into a pre-set speech recognition model to accurately convert speech data into text data. This model is a neural network specifically trained for emergency medicine scenarios, effectively filtering environmental noise and accurately recognizing medical terminology and colloquial triage expressions, fundamentally eliminating information errors caused by speech conversion distortion. Second, named entity recognition is performed on the converted text data, specifically extracting the unique ID field associated with the patient's identity and identifying it as the corresponding patient ID. The principle is to use named entity recognition technology to lock onto the unique identifier in the text, achieving an initial binding between the patient ID and the triage speech information, fundamentally preventing confusion and mismatch of triage information for multiple patients. The third step involves simultaneously performing medical entity recognition on the text data, accurately extracting the core keywords describing the patient's injuries and symptoms, and removing interjections and scene-related descriptive words that have no actual triage significance. The principle is to focus on the core basis for determining the emergency condition and retain only medical keywords that indicate the urgency of the condition, thus laying a solid foundation for subsequent triage and matching.

[0029] After keyword extraction, the extracted keywords are matched with severity levels according to a pre-defined severity grading mapping library. This library stores the associations between different symptom keywords and their corresponding severity levels. Severity levels are strictly divided from high to low according to the urgency of the condition indicated by the symptoms, and each level is pre-assigned a unique grade value. The principle behind this is to establish a unified standard for emergency condition grading, avoiding subjective judgment biases from manual triage by different doctors, and achieving standardization and normalization of batch patient condition grading. When extracted symptom keywords correspond to multiple severity levels, the highest grade value is selected as the initial severity level. This follows the core principle of emergency medicine: "prioritizing critically ill patients." Even if a patient has multiple symptoms, the most critical and highest-level condition is used as the grading basis, preventing mild cases from masking potentially critical conditions and ensuring the safety and rationality of emergency triage. After the initial severity grading is completed, the corresponding patient number is accurately written into the new row of the dynamic matrix. At the same time, the initial severity grading is converted into the first priority weight, and the weight is stored in the column corresponding to the voice input channel in the new row. This operation continues the construction logic of the dynamic matrix in step S102, that is, each row corresponds to an activated patient number and each column corresponds to a data acquisition channel. The core triage data of voice acquisition is accurately embedded into the corresponding data position of the matrix, realizing the integrated binding and storage of patient identification number, initial condition grading, and triage priority weight.

[0030] For example, in a specific emergency scenario, a top-tier general hospital suddenly received a group of injured patients from a construction site collapse. A total of 22 patients were placed in the triage area at the emergency entrance. Medical staff assigned activated numbers 001 to 022 to the patients according to their arrival order. After examining each patient's injuries, the doctor clearly stated, through the voice input channel, "Patient number 012, chest compression injury with difficulty breathing, upper limb abrasions," to patient number 012. The system received this voice stream, first converting it into complete text data using a preset speech recognition model, then accurately extracting patient number 012 using named entity recognition, and finally processing it through medical... Body recognition extracts three symptom keywords: "chest crush injury," "dyspnea," and "upper limb abrasion." After matching with a preset severity grading mapping library, "dyspnea" corresponds to Level 1 critical illness, "chest crush injury" corresponds to Level 2 severe illness, and "upper limb abrasion" corresponds to Level 3 mild illness. The system then selects Level 1 critical illness, which has the highest grade value, as the patient's initial severity grade, converts it into the corresponding first priority weight, and simultaneously adds the patient number 012 to a new row in the dynamic matrix. The first priority weight is stored in the corresponding column of the voice input channel in that row, completing the standardized collection of the patient's core triage data.

[0031] S104, extract the spatial coordinates of each candidate diversion location and calculate the path length; sort the activated patient numbers from high to low according to the first priority weight to obtain the allocation order; according to the allocation order, select the candidate diversion location that is currently unoccupied and has the shortest path length for each patient in turn, and generate the initial matching record.

[0032] Specifically, the digitized spatial coordinates corresponding to each candidate triage location marked in the 3D model are retrieved, and then the path length is calculated. The calculation process is as follows: In the 3D model and its corresponding coordinates, walls are used as insurmountable constraints within the same floor plane to construct walkable areas on that floor. This aims to replicate the access rules of the actual physical building of the hospital, setting walls as hard constraints that cannot be crossed within the plane. The principle is to avoid virtual invalid paths that do not conform to the actual scenario, such as paths that pass through or cross walls, ensuring that the calculated path is feasible on-site. Subsequently, for each candidate triage location and the current real-time location of each activated patient, it is determined whether the candidate triage location to be calculated and the real-time location are on the same floor. The principle is to determine the floor affiliation by whether the floor coordinates of the 3D model are on the same horizontal plane, thereby adapting to the structural characteristics of multi-story emergency buildings in hospitals and avoiding path errors caused by a single calculation logic. If the determination result indicates the patient is on the same floor, the path search algorithm is used within the walkable area of ​​that floor to calculate the shortest distance around wall obstacles, which is then used as the travel path length. The principle is to simplify the 3D calculation logic for same-floor scenarios, focusing on in-plane obstacle avoidance and geometric shortest distance selection, balancing calculation efficiency and accuracy. The beneficial effect is the rapid completion of path calculation for patients on the same floor, adapting to the efficient handling needs of batch emergency triage. If the determination result indicates the patient is on a different floor, the path search algorithm is used to calculate the first segment path distance from the current real-time location to each staircase entrance on the current floor, and the second segment path distance from each staircase exit on the candidate triage location's floor to the candidate triage location. Then, the fixed vertical transition distance between the staircase entrance and exit is obtained. The first segment path distance, the second segment path distance, and the fixed vertical transition distance for the same staircase are added together, and the shortest distance obtained is selected as the travel path length. This segmented calculation logic decomposes the complex 3D cross-floor path into planar and vertical segments, where the fixed vertical transition distance is a pre-stored standardized staircase parameter and does not require real-time calculation. It transforms the problem of three-dimensional passage in multi-story buildings into a quantifiable problem of summing segmented distances, solves the complexity problem of three-dimensional spatial path calculation, accurately adapts to the actual structure of multi-story emergency buildings in hospitals, and achieves unbiased quantification of cross-floor diversion paths.

[0033] The path search algorithm used is based on a heuristic search strategy using physical travel costs, as detailed below: First, in the 3D model, the walkable area is discretized into a topological network composed of nodes and edges. Nodes correspond to the center point of a spatial unit or the intersection of channels, and edges correspond to the path segments connecting two nodes. The principle is to transform the continuous physical walkable area into a discretized topological network structure, adapting to the path search logic of heuristic algorithms, allowing spatial paths to be accurately identified and traversed by the algorithm. Then, a basic length value is assigned to each edge, and a dynamic passage cost is generated based on the physical attributes of the spatial unit or channel unit in which the edge is located. The physical attributes include at least channel width, passage direction restrictions, and real-time patient density. This breaks through the traditional single mode of calculating only geometric straight-line distance. The principle is to incorporate the physical constraints of actual passage in the emergency room and real-time human flow factors, so that the passage cost not only includes spatial length, but also covers actual influencing factors such as on-site congestion and channel width, avoiding the problem of the geometric shortest path being disconnected from the actual optimal passage path. Finally, starting from the patient's current real-time location node and targeting candidate triage nodes, a heuristic search algorithm is used to search for a path. During the search, the actual travel cost from the starting point to the current node is the sum of the dynamic travel costs of all edges traversed, and the estimated travel cost from the current node to the target point is the Euclidean distance between the two points divided by the preset maximum travel speed. The node with the smallest sum of actual and estimated costs is selected for expansion until the target point is found. The output is the travel path formed by connecting nodes and travel edges in sequence. The principle is to combine the actual cost of the already traversed path with the estimated cost of the remaining path to achieve rapid convergence of the optimal path. This avoids excessive time consumption from blind searches and ensures that the final path is the one with the highest actual travel efficiency on site. The beneficial effect is that in emergency scenarios, it balances the timeliness and optimality of path search, meeting the core need for rapid triage of batches of patients.

[0034] The dynamic toll cost is calculated using the following preset formula: In the formula, The dynamic passage cost of an edge; This represents the basic length value of the edge; The actual channel width of the edge; This is the standard reference width; This is the width influence coefficient; The direction inverse coefficient; This is the directional influence coefficient; This represents the real-time patient density in the area where the edge is located. The patient density influence coefficient.

[0035] After calculating the path length, the activated patient IDs are sorted from highest to lowest priority weight to determine the allocation order. The sorting logic follows the core principle of emergency medicine: "prioritizing critically ill patients." The principle is to intuitively distinguish the urgency of patients' conditions using quantified priority weights, placing critically ill patients with high weights at the forefront of triage allocation, thus avoiding disordered triage and the problem of mild cases occupying priority resources when a large influx of patients occurs. Subsequently, according to the allocation order, candidate triage locations with the shortest path length and currently unoccupied status are selected for each patient, generating an initial matching record. This operation simultaneously verifies the real-time occupancy status of candidate triage locations during triage matching, filtering only unoccupied locations. The optimal location is locked using path length as the core screening indicator, combining patient condition priority and spatial accessibility for matching. This ensures the rationality of the triage order while reducing patient movement distance and medical staff transfer time through the shortest path.

[0036] To further clarify the practical application effect of this step, let's take a car accident emergency scenario from the previous embodiment as an example: A tertiary general hospital receives a large number of patients from a group car accident. The waiting hall and temporary observation cubicles in the emergency building have been marked as candidate triage locations through step S101. Step S103 completes the activation of the numbers and the assignment of the first priority weight for 20 patients. Among them, patients numbered 003 and 007 have severe injuries and their first priority weights are ranked first and second, while patients numbered 011 to 020 have minor injuries and their weights are relatively low. The system first extracts the three-dimensional spatial coordinates of the waiting hall and each temporary observation cubicle, determining that all patients are located in the triage area on the first floor of the emergency building, and that all candidate triage locations are also on the first floor. Then, a walkable area on the first floor is constructed with the walls as constraints. A heuristic search algorithm based on physical travel cost is used to calculate the shortest path length from each patient to each candidate triage location. Subsequently, patients 003 and 007 are sorted from high to low according to their first priority weight, and patients 003 and 007 are assigned priority. After verifying that each candidate triage location is unoccupied, the system selects temporary observation cubicle No. 1, which has the shortest path distance from the real-time location, for patient 003, and temporary observation cubicle No. 2, which is adjacent to patient 007. Then, the remaining unoccupied waiting hall area locations with the shortest paths are assigned to the other patients with mild symptoms in turn. Finally, an initial matching record containing the patient number, the corresponding candidate triage location, and the path length is generated.

[0037] S105 continuously acquires real-time data streams from the vital signs monitoring channel, drug traceability channel, and equipment treatment parameter channel. The real-time data streams carry patient numbers and monitoring values. It associates fluctuation segments of the corresponding indicators in the monitoring values ​​that exceed the preset range with the corresponding patient numbers, marks the patient numbers as emergency numbers, and triggers an alarm.

[0038] Specifically, this step is implemented based on the dynamic matrix multi-data acquisition channel system built by S102. The principle is that after a batch of emergency patients have completed the initial triage and placement, their condition is not in a constant state. Simply relying on the initial severity classification is a static management mode, which is very easy to miss the risk of sudden deterioration of the condition. Therefore, it is necessary to establish a dynamic monitoring mechanism that is in all-time and uninterrupted to comprehensively track the changes in the physical condition of patients after triage and treatment-related data. This continues the core logic of one-to-one precise binding between patients and data in the dynamic matrix, ensuring that each set of real-time monitoring data corresponds to a unique patient number, thereby eliminating the problem of cross-contamination, mismatch and omission of real-time data from multiple patients from the root.

[0039] Building upon this, the system further associates fluctuations in monitored indicators exceeding preset ranges with corresponding patient IDs, then marks these patient IDs as emergency numbers and triggers an alarm. The principle behind this operation is to pre-set normal ranges or thresholds for various indicators that conform to emergency medicine standards, establishing standardized and unified criteria for judging abnormal conditions. This eliminates the lag and subjective bias of manual comparison, and through automated real-time comparison of thresholds or preset ranges, quickly filters out abnormal data segments. Then, relying on the patient ID, it accurately binds the abnormal data to the corresponding patient. Using the unique identifier of the emergency number, it clearly distinguishes between stable, routinely triaged patients and those experiencing sudden critical illness. Simultaneously, it activates the system's built-in alarm module to push risk alerts to medical staff terminals immediately, preventing delays in emergency treatment due to untimely responses. The specific implementation method is as follows... Figure 3 As shown, vital sign monitoring parameters, drug traceability parameters, and equipment treatment parameters are received through multi-channel data reception. After comparing the indicators (i.e. whether they meet the pre-set normal range or threshold of each indicator), abnormal monitoring values ​​are determined, and normal or abnormal data is obtained from the determination. For normal data, monitoring is maintained. For abnormal data, an emergency number is generated and an alarm is triggered while monitoring is maintained.

[0040] Using the landslide emergency scenario example above, after patient number 012 with chest compression injury was initially triaged to a temporary observation room, the system continuously collected real-time data such as vital sign monitoring parameters and equipment treatment parameters through the corresponding channel. Subsequently, when the patient's blood oxygen saturation suddenly dropped below the preset normal threshold and the heart rate fluctuated abnormally, the system automatically locked the abnormal fluctuation segment and accurately associated it with number 012, immediately marking 012 as an emergency number and simultaneously triggering a dedicated alarm at the emergency medical station. This facilitated medical staff in quickly locating the suddenly critically ill patient and initiating subsequent emergency treatment and triage adjustment procedures.

[0041] S106, extract the unoccupied rescue space units from the space units carrying treatment attribute tags; calculate the path distance between the real-time location corresponding to the emergency number and each unoccupied rescue space unit, select the rescue space unit with the smallest path distance as the new matching location, and output the diversion matching result; For critically ill patients with marked emergency numbers, the initial non-treatment triage locations are discarded. Instead, the focus is on treatment attribute spaces with core treatment capabilities. First, a full traversal of the 3D model is used to locate all treatment attribute spatial units, simultaneously verifying real-time occupancy status and precisely selecting available resuscitation units. This avoids occupancy of already-used core treatment resources and ensures compliant allocation of treatment space. Then, using the 3D model path calculation logic, the effective path distance between the patient's real-time location and each available resuscitation unit is calculated. The unit corresponding to the shortest path is selected, aligning with the core principle of "rapid and nearby treatment" in emergency medicine, minimizing the time required for critically ill patient transfer. The beneficial effects include rapidly completing secondary triage matching for critically ill patients, achieving precise connection of core treatment resources, and mitigating the treatment risks caused by transfer delays. Taking the previous example of a mass collapse emergency scenario: After emergency patient number 012 is marked, the system traverses the treatment attribute spaces such as the emergency room and the intensive care unit debridement room in the 3D model, selects the vacant intensive care unit debridement room No. 3, calculates that the shortest path between the patient's real-time location and the debridement room is, and sets it as the new matching location. Then, it outputs the final diversion matching result containing the patient number and the new rescue space location.

[0042] like Figure 4 As shown, based on the same inventive concept, this embodiment provides an emergency batch patient triage data processing system, including: The spatial candidate labeling module 201 is used to load a three-dimensional model of the hospital in response to the emergency plan activation command, and obtain the attribute labels of each spatial unit in the three-dimensional model. The attribute labels include treatment attribute labels and non-treatment attribute labels; and mark the spatial units carrying non-treatment attribute labels as candidate diversion locations. The dynamic matrix construction module 202 is used to construct a dynamic matrix. Each row of the dynamic matrix corresponds to an activated patient number, and each column of the dynamic matrix corresponds to a data acquisition channel. The data acquisition channels include a voice input channel, a vital signs monitoring channel, a drug traceability channel, and a device treatment parameter channel. The voice grading assignment module 203 is used to receive the voice stream uploaded by medical staff through the voice input channel, extract the patient number and initial severity grade from the voice stream, write the patient number into a new row of the dynamic matrix, and convert the initial severity grade into the first priority weight value and store it in the corresponding column of the new row. The path diversion module 204 is used to extract the spatial coordinates of each candidate diversion location and calculate the length of the travel path; sort the activated patient numbers from high to low according to the first priority weight to obtain the allocation order; according to the allocation order, select the candidate diversion location that is currently unoccupied and has the shortest travel path length for each patient in turn to generate an initial matching record; The real-time monitoring and alarm module 205 is used to continuously acquire real-time data streams from the vital signs monitoring channel, the drug traceability channel, and the equipment treatment parameter channel. The real-time data stream carries the patient number and the monitoring value. It associates the fluctuation segment of the corresponding indicator in the monitoring value that exceeds the preset range with the corresponding patient number, marks the patient number as an emergency number, and triggers an alarm. The rescue matching module 206 is used to extract unoccupied rescue space units from the space units carrying treatment attribute tags; calculate the path distance between the real-time location corresponding to the emergency number and each unoccupied rescue space unit; select the rescue space unit with the smallest path distance as the new matching location; and output the diversion matching result.

[0043] like Figure 5 As shown, based on the same inventive concept, this embodiment provides an electronic device, including: Memory 302 is used to store computer programs; Processor 301 is used to implement the method steps as described in the first aspect when executing a computer program.

[0044] Based on the same inventive concept, this embodiment provides a readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, it implements the method steps as described in the first aspect.

[0045] It should be noted that the specific methods by which each module performs operations in the system described in the above embodiments have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0046] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for processing emergency department patient triage data, characterized in that, include: In response to the emergency plan activation command, a three-dimensional model of the hospital is loaded, and the attribute labels of each spatial unit in the three-dimensional model are obtained. The attribute labels include treatment attribute labels and non-treatment attribute labels. The spatial units carrying the non-treatment attribute labels are marked as candidate diversion locations. A dynamic matrix is ​​constructed, wherein each row of the dynamic matrix corresponds to an activated patient ID, and each column of the dynamic matrix corresponds to a data acquisition channel, including a voice input channel, a vital signs monitoring channel, a drug traceability channel, and a device treatment parameter channel. The system receives voice streams uploaded by medical staff through a voice input channel, extracts the patient number and initial severity rating from the voice stream, writes the patient number into a new row of a dynamic matrix, and converts the initial severity rating into a first priority weight and stores it in the corresponding column of the new row. Extract the spatial coordinates of each candidate diversion location and calculate the path length; The activated patient IDs are sorted from highest to lowest according to their first priority weight to obtain the allocation order; According to the allocation order, candidate diversion positions that are currently unoccupied and have the shortest travel path length are selected for each patient in turn to generate an initial matching record; Real-time data streams are continuously acquired from the vital signs monitoring channel, the drug traceability channel, and the equipment treatment parameter channel. The real-time data streams carry the patient number and the monitoring values. Fluctuations in the monitoring values ​​that exceed the preset range are associated with the corresponding patient number. The patient number is marked as an emergency number and an alarm is triggered. Extract unoccupied rescue space units from the space units carrying the aforementioned treatment attribute tags; Calculate the path distance between the real-time location corresponding to the emergency number and each of the unoccupied rescue space units, select the rescue space unit with the smallest path distance as the new matching location, and output the diversion matching result.

2. The method for processing emergency batch patient triage data according to claim 1, characterized in that, The step of loading the 3D model of the hospital's interior specifically includes: The system calls upon the building structure data stored in the building information modeling system. The building structure data includes floor plans, wall partition lines, and room function labels. A three-dimensional mesh map is constructed based on the building structure data, and each mesh cell in the three-dimensional mesh map is bound to the room function label to generate attribute labels for each spatial cell in the three-dimensional model.

3. The method for processing emergency batch patient triage data according to claim 1, characterized in that, The steps of extracting the patient ID and initial severity rating from the speech stream specifically include: The speech stream is input into a preset speech recognition model and converted into text data; named entity recognition is performed on the text data to extract the number field associated with the patient's identity as the patient's number; medical entity recognition is performed on the text data to extract keywords describing the injury or symptoms. According to a preset severity grading mapping library, the extracted keywords are matched with the grading levels. The mapping library stores the association between different symptom keywords and corresponding severity levels. The severity levels are divided according to the urgency of the illness pointed to by the symptoms and different level values ​​are pre-assigned. When the extracted keywords correspond to multiple severity levels, the level with the highest level value is selected as the initial severity classification.

4. The method for processing emergency batch patient triage data according to claim 1, characterized in that, The steps for calculating the length of the travel path specifically include: In the three-dimensional model and its corresponding coordinates, walls are used as insurmountable constraints within the same-level plane to construct walkable areas on the same level. For each candidate triage location and the current real-time location of each activated patient, determine whether the candidate triage location to be calculated and the real-time location are on the same floor; If they are on the same floor, the shortest distance to bypass the wall obstacle is calculated using a path search algorithm within the walkable area of ​​that floor, and this distance is taken as the path length. If the locations are on different floors, a path search algorithm is used to calculate the first path distance from the current real-time location to each stair entrance on the current floor, and the second path distance from each stair exit on the floor where the candidate diversion location is located to the candidate diversion location. Then, the fixed vertical transition distance of the corresponding stair passage between the stair entrance and the stair exit is obtained. The first path distance, the second path distance, and the fixed vertical transition distance corresponding to the same stair are added together, and the shortest distance after the addition is selected as the passage path length.

5. The emergency department batch patient triage data processing method according to claim 4, characterized in that, The path search algorithm employs a heuristic search strategy based on physical travel cost, specifically including: In the three-dimensional model, the walkable area is discretized into a topological network consisting of nodes and edges, where nodes correspond to the center point of a spatial unit or the intersection of a passage, and edges correspond to the passage path segments connecting two nodes; each edge is assigned a basic length value, and a dynamic passage cost is generated based on the physical properties of the spatial unit or passage unit in which the edge is located, the physical properties including at least the passage width, passage direction restrictions, and real-time patient density; Starting from the patient's current real-time location node and targeting candidate triage locations, a heuristic search algorithm is used to search for a path. During the search, the actual travel cost from the starting point to the current node is the sum of the dynamic travel costs of all edges traversed, and the estimated travel cost from the current node to the target point is the Euclidean distance between the two points divided by the preset maximum travel speed. The node with the smallest sum of actual and estimated travel costs is selected for expansion until the target point is found. The output path is formed by connecting nodes and travel edges in sequence.

6. The method for processing emergency batch patient triage data according to claim 5, characterized in that, The dynamic passage cost is calculated using the following preset formula: In the formula, The dynamic passage cost of an edge; This represents the basic length value of the edge; The actual channel width of the edge; This is the standard reference width; This is the width influence coefficient; The direction inverse coefficient; This is the directional influence coefficient; This represents the real-time patient density in the area where the edge is located. The patient density influence coefficient.

7. The method for processing emergency batch patient triage data according to claim 1, characterized in that, The step of extracting unoccupied rescue space units from the space units carrying the treatment attribute tags specifically includes: Traverse all spatial units carrying treatment attribute tags in the three-dimensional model, obtain the current occupancy status of each spatial unit, filter out the spatial units whose current occupancy status is vacant, and use them as unoccupied rescue spatial units.

8. An emergency department batch patient triage data processing system, based on the emergency department batch patient triage data processing method described in claim 1, characterized in that, include: The spatial candidate labeling module is used to load a three-dimensional model of the hospital in response to the emergency plan activation command, and obtain the attribute label of each spatial unit in the three-dimensional model. The attribute label includes a treatment attribute label and a non-treatment attribute label; and mark the spatial unit carrying the non-treatment attribute label as a candidate diversion location. A dynamic matrix construction module is used to construct a dynamic matrix. Each row of the dynamic matrix corresponds to an activated patient ID, and each column of the dynamic matrix corresponds to a data acquisition channel. The data acquisition channels include a voice input channel, a vital signs monitoring channel, a drug traceability channel, and a device treatment parameter channel. The voice grading assignment module is used to receive the voice stream uploaded by medical staff through the voice input channel, extract the patient number and initial severity grade from the voice stream, write the patient number into a new row of the dynamic matrix, and convert the initial severity grade into a first priority weight value and store it in the corresponding column of the new row. The path splitting module is used to extract the spatial coordinates of each candidate splitting location and calculate the length of the travel path; The activated patient IDs are sorted from highest to lowest according to their first priority weight to obtain the allocation order; According to the allocation order, candidate diversion positions that are currently unoccupied and have the shortest travel path length are selected for each patient in turn to generate an initial matching record; The real-time monitoring and alarm module is used to continuously acquire real-time data streams from the vital signs monitoring channel, the drug traceability channel, and the equipment treatment parameter channel. The real-time data streams carry patient numbers and monitoring values. The module associates fluctuation segments of the corresponding indicators in the monitoring values ​​that exceed the preset range with the corresponding patient numbers, marks the patient numbers as emergency numbers, and triggers an alarm. The rescue matching module is used to extract unoccupied rescue space units from the space units carrying the treatment attribute tags; Calculate the path distance between the real-time location corresponding to the emergency number and each of the unoccupied rescue space units, select the rescue space unit with the smallest path distance as the new matching location, and output the diversion matching result.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the emergency batch patient triage data processing method as described in any one of claims 1 to 7.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program that, when executed by a processor, implements an emergency batch patient triage data processing method as described in any one of claims 1 to 7.

Images