An ontology-driven method and system for multimodal combat rescue data fusion and aggregation
By using an ontology-driven multimodal combat rescue data fusion method, semantic consistency of combat rescue data across devices, stages, and systems is achieved, reducing manual maintenance costs, improving data connectivity, supporting multidimensional indexing and visual statistical analysis, and solving the problem of data aggregation and retrieval that is difficult to achieve quickly in existing technologies.
Patent Information
- Application Number
- CN202611183108.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-05
- Publication Date
- 2026-11-17
Smart Images

Figure CN123120183A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of medical information technology, and in particular to an ontology-driven method and system for the fusion and aggregation of multimodal combat and rescue data. Background Technology
[0002] In combat rescue scenarios, casualties are complex and the treatment process is lengthy. Relevant information typically originates from multiple monitoring methods and various terminal devices. This includes multimodal vital sign data such as continuous blood pressure, ECG, blood oxygen saturation, body temperature, respiration, pulse rate, as well as EEG, EMG, non-invasive intracranial pressure, and cardiac output. It also includes operational data such as injury classification, treatment records, medication information, transport methods, destination, geographical location, and timestamps. This type of data exhibits significant multi-source heterogeneity, with substantial differences in output format, field naming, unit dimensions, and sampling frequency among different devices or systems. Furthermore, the same indicator often corresponds to multiple names, abbreviations, or encoding methods, resulting in inconsistent semantic expression and making standardization difficult. On the other hand, due to inconsistent sampling frequencies, timestamp drift, and changes in device status, the data commonly suffers from issues such as inconsistent timing, noise interference, missing data, and anomalies. More importantly, the lack of a unified data organization method across multiple stages—frontline response, evacuation, and rear treatment—makes it difficult to establish an object-event-observation link throughout the entire process. This results in the difficulty of rapidly aggregating, retrieving, and statistically analyzing combat rescue data according to dimensions such as region, time, injury severity, injury type, treatment method, and evacuation destination. Therefore, traditional data management methods based on fixed table structures or manual alignment are no longer sufficient to meet the actual needs of real-time performance, scalability, and unified management in combat rescue scenarios.
[0003] Currently, existing technical solutions mainly fall into two categories. The first is a heterogeneous data integration solution based on manual field alignment. This type of solution typically uses manually defined ETL (Extract-Transform-Load) rules to map fields and convert formats of data from different devices and systems, then integrates multi-source data into a unified relational database or data warehouse. This solution is applicable to scenarios with relatively fixed data sources and stable formats. However, in emergency rescue scenarios, due to the diverse types of equipment, inconsistent field naming, dynamic changes in business processes, and the high cost of manually maintaining mapping rules, it struggles to effectively handle cross-stage entity associations and semantic conflicts. The second category is a medical data management solution based on traditional knowledge graphs. This type of solution constructs a medical knowledge graph, structurally storing medical entities and their relationships in a graph database to achieve cross-system entity links and relationship retrieval. Representative works include clinical data semantic integration solutions based on standard terminology systems such as SNOMED CT or ICD. However, existing knowledge graph solutions are usually mainly geared towards structured electronic medical record data, lacking the ability to collaboratively manage continuous multimodal vital signs time-series data, and are not specifically designed for the multi-stage process links, real-time quality control, and multi-dimensional statistical retrieval needs in combat and rescue scenarios. Therefore, they are still difficult to meet the needs of data fusion, aggregation, and efficient application throughout the entire combat and rescue process. Summary of the Invention
[0004] To address the technical challenges of existing technologies in unifying the semantic representation of multi-source, heterogeneous, multi-modal, and cross-stage data in combat rescue scenarios, establishing object event chains, and simultaneously managing graph relationship retrieval and time-series data collaboratively, this invention provides an ontology-driven multi-modal combat rescue data fusion and aggregation method and system. The technical solution is as follows: On the one hand, an ontology-driven multimodal combat rescue data fusion and aggregation method is provided. This method is implemented by an ontology-driven multimodal combat rescue data fusion and aggregation device, and includes: S1. Acquire multi-source heterogeneous combat rescue data and attach metadata including device identifier, timestamp, geographic location, and stage identifier; wherein the multi-source heterogeneous combat rescue data refers to multi-source heterogeneous data from various combat rescue terminals, monitoring equipment, and business systems. S2. Based on the ontology model of the combat rescue domain and the pre-built terminology standard system, the field standardization, semantic mapping and unit normalization of multi-source heterogeneous combat rescue data are performed to obtain pre-processed standardized data. S3. Perform entity disambiguation, event association, and conflict resolution on the preprocessed standardized data to construct a semantic structure that includes objects, events, and observations; S4. Write the entities, relationships, and event chains in the semantic structure into a pre-built graph database, and write the continuous monitoring sequence and window features into a pre-built time series database, while generating a time series reference identifier; S5. Based on pre-configured search conditions for region, time, and injury severity, filter the sets of injured persons and events that meet the search conditions in the graph database, and perform time window aggregation and statistical calculations in the time series database according to the time series reference identifier, and output the joint query results including the sets of injured persons, the sets of events, and the time window statistics.
[0005] On the other hand, an ontology-driven multimodal combat rescue data fusion and aggregation system is provided. This system is applied to an ontology-driven multimodal combat rescue data fusion and aggregation method, and includes: The data acquisition and access module is used to acquire multi-source heterogeneous combat rescue data and add metadata including device identifier, timestamp, geographic location, and stage identifier; wherein the multi-source heterogeneous combat rescue data is multi-source heterogeneous data from various combat rescue terminals, monitoring equipment, and business systems. The ontology model and terminology standard module is used to perform field standardization, semantic mapping, and unit normalization on multi-source heterogeneous combat rescue data based on the combat rescue ontology model and the pre-built terminology standard system, so as to obtain pre-processed standardized data. The multimodal data preprocessing module is used to perform entity disambiguation, event association, and conflict resolution on the preprocessed standardized data, and to construct a semantic structure that includes objects, events, and observations. A unified storage module is used to write entities, relationships, and event chains in the semantic structure into a pre-built graph database, and to write continuous monitoring sequences and window features into a pre-built time-series database, while generating time-series reference identifiers. The multidimensional indexing and retrieval module is used to filter sets of wounded persons and events that meet the pre-configured search conditions of region, time and injury severity in the graph database, and perform time window aggregation and statistical calculations in the time series database according to the time series reference identifier, and output the joint query results including the set of wounded persons, the set of events and the time window statistics.
[0006] On the other hand, a body-driven multimodal combat rescue data fusion and aggregation device is provided. The body-driven multimodal combat rescue data fusion and aggregation device includes: a processor; a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, any one of the methods in the body-driven multimodal combat rescue data fusion and aggregation method described above is implemented.
[0007] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, the at least one instruction being loaded and executed by a processor to implement any of the above-described ontology-driven multimodal combat rescue data fusion and aggregation methods.
[0008] The beneficial effects of the technical solutions provided by the embodiments of the present invention include at least the following: The embodiments of this invention achieve semantic consistency across devices, stages, and systems through ontology-driven approaches, significantly reducing the cost of manual field maintenance. By constructing object event observation links through entity disambiguation and event association, the ability to connect information across processes is improved, solving the problems of poor scalability and inability to handle semantic layer entity associations in dynamic combat and rescue scenarios of existing manual field alignment schemes.
[0009] This invention employs a dual-database storage approach, combining a knowledge graph library and a time-series library, and enables linked retrieval via TSRef. This achieves complementary advantages between semantic relationship retrieval and high-frequency time-series aggregation. Furthermore, it supports combat readiness assessment and resource scheduling through multi-dimensional indexing and visual statistical analysis, thus addressing the shortcomings of existing knowledge graph solutions in terms of multi-modal time-series data collaborative management capabilities and multi-dimensional retrieval capabilities for combat readiness operations. Attached Figure Description
[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0011] Figure 1 This is a flowchart of an ontology-driven multimodal combat rescue data fusion and aggregation method provided in an embodiment of the present invention; Figure 2 This is an overall architecture diagram of a body-driven multimodal combat rescue data fusion and aggregation system provided in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the linkage between a dual-database storage architecture and a Time-Sequenced Reference Token (TSRef) provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of a drill-back process that integrates statistical analysis and visualization modules, provided by an embodiment of the present invention. Figure 5 This is a schematic diagram illustrating the implementation process of an ontology-driven multimodal combat rescue data fusion and aggregation method provided in an embodiment of the present invention; Figure 6 This is a block diagram of a body-driven multimodal combat rescue data fusion and aggregation system provided in an embodiment of the present invention; Figure 7This is a schematic diagram of the structure of a body-driven multimodal combat rescue data fusion and aggregation device provided in an embodiment of the present invention. Detailed Implementation
[0012] The technical solution of the present invention will now be described with reference to the accompanying drawings.
[0013] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.
[0014] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.
[0015] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.
[0016] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.
[0017] This invention provides an ontology-driven method for fusing and aggregating multimodal combat rescue data. This method can be implemented using an ontology-driven multimodal combat rescue data fusion and aggregation device, which can be a terminal or a server. For example... Figure 1 The flowchart shown is for an ontology-driven multimodal combat rescue data fusion and aggregation method. The processing flow of this method may include the following steps: S1. Acquire multi-source heterogeneous combat rescue data and attach metadata including device identifier, timestamp, geographic location, and stage identifier; where multi-source heterogeneous combat rescue data refers to multi-source heterogeneous data from various combat rescue terminals, monitoring equipment, and business systems.
[0018] One feasible implementation method is, for example Figure 2The diagram shown is an overall architecture diagram of an ontology-driven multimodal combat rescue data fusion and aggregation system provided by an embodiment of the present invention. The overall system architecture includes: a data acquisition and access module, an ontology model and terminology standard module, a multimodal data preprocessing module, an ontology-driven semantic mapping and fusion module, a unified storage module, a multidimensional indexing and retrieval module, and a fusion statistical analysis and visualization module.
[0019] The data acquisition and access module is used to access various combat rescue information sources and detection equipment on mobile terminals, edge terminals or other combat rescue terminals, collect vital signs data, test and detection data, injury and treatment data, transfer and evacuation data, as well as time and location information, and upload each piece of data to the platform after adding metadata such as device identifier, timestamp, geographical location and stage identifier.
[0020] The ontology model and terminology standard module is used to construct an ontology model and terminology standard system for the combat rescue domain, define classes, attributes, relationships and constraint rules, and form machine-readable ontology files and standardized terminology dictionaries to support subsequent semantic mapping, rule verification and index dimension definition.
[0021] The multimodal data preprocessing module is used to perform time alignment, signal processing, missing data detection, anomaly detection, and windowing processing on the acquired data, and to generate a quality score by combining information such as signal quality index, device status, and missing data ratio.
[0022] The semantic mapping and fusion module is used to map heterogeneous fields, device outputs, encoded information or text descriptions to concepts, attributes and relationships in the ontology, and completes synonym unification, abbreviation unification, unit unification, entity disambiguation, event association and conflict resolution, forming a unified object-event-observation semantic structure.
[0023] In one feasible implementation, during the semantic mapping and fusion process, fields, codes, or text descriptions in the collected data are mapped to the corresponding concept identifiers in the ontology model to achieve semantic normalization and unit unification. For the same injured person across devices and stages, a unified entity identifier is created to form a PatientID. Related data is then organized into event-based groups based on time, location, stage, and business actions to generate EventIDs. Various observation data are then categorized into corresponding event chains. For situations where the same indicator has multiple source values at the same time or within the same time window, weighted fusion or optimal retention is performed based on time freshness, device reliability, and quality score, ultimately forming a unified object-event-observation semantic structure. The observation data element can be represented as Observation={ObservationID, PatientID, EventID, ConceptID, Value / ValueRange, Unit, Timestamp / Window, QualityScore, DeviceID,TSRef}, where ObservationID is the observation identifier, PatientID is the unique identifier of the injured person, EventID is the event identifier, ConceptID is the ontology concept identifier, Value / ValueRange is the observed value or value range, Unit is the unit of measurement, Timestamp / Window is the timestamp or time window, QualityScore is the quality score, and DeviceID is the device identifier; ConceptID and Unit are constrained by the ontology model, and TSRef is the time-series reference identifier, used to establish the association between the graph database / knowledge graph library and the time-series database.
[0024] The unified storage module is used to collaboratively store the fused data using a graph database or knowledge graph library and a time-series database. The graph database or knowledge graph library stores ontology, entities, relationships, and event chains, while the time-series database stores continuous monitoring sequences and their window features. The two are associated through unified identifiers such as PatientID (unique identifier for the injured person), EventID (event identifier), ObservationID (observation identifier), DeviceID (device identifier), and TSRef (Time Series Reference). Raw high-frequency waveforms such as ECGs are stored in object storage as file objects and pointed to by time-series references.
[0025] Among them, such as Figure 3The diagram illustrates a dual-database storage architecture and the linkage between time-series reference identifiers (TSRefs) provided in an embodiment of the present invention. In one feasible implementation, during unified storage, wounded individuals, events, observations, and their relationships are first created or updated in a pre-built graph database or knowledge graph database. Then, original time-series data or window features associated with ObservationIDs (observation identifiers) or EventIDs (event identifiers) are written into the pre-built time-series database. The corresponding time-series reference identifiers are then backfilled into the observation nodes in the pre-built graph database or knowledge graph database, forming TSRefs. Simultaneously, this is recorded in a log to support data traceability, recalculation, and playback. During linked retrieval, the target set of wounded individuals or events is first filtered based on concepts, attributes, and relationships in the pre-built graph database or knowledge graph database. Time window aggregation, curve playback, or statistical calculations are performed in the time-series database based on the TSRefs, thereby outputting the joint query results. Through the aforementioned dual-database storage and linked retrieval mechanism, collaborative support for graph relation semantic querying and high-frequency time-series data management is achieved.
[0026] The multidimensional indexing and retrieval module is used to build indexes around dimensions such as time, region, injury severity, injury type, treatment type, post-transfer destination, and risk events to support rapid filtering and linked queries.
[0027] Among them, the integrated statistical analysis and visualization module is used to perform multi-dimensional statistics, situation display, drill-down analysis and replay display on the fusion results, and combine resource consumption and inventory status to conduct bottleneck early warning and risk event monitoring, realizing the step-by-step backtracking from statistical results to the casualty event link and then to the physiological curve of a specific time window. Among them, such as Figure 4 The diagram illustrates a drill-back process integrating statistical analysis and visualization modules provided in an embodiment of the present invention. In one feasible implementation, the specific steps of the drill-back process include: The content displayed based on multidimensional statistical situation includes: the number of wounded and their expected arrival trend, the distribution of injury severity, resource inventory and bottleneck warning, risk event monitoring, and the status of transportation tools. The target set of wounded is constructed by drilling down by dimension. Based on the target set of wounded, the wounded are selected and a single wounded event chain is constructed by filtering according to time, region, injury level, injury type, treatment type and evacuation destination. Based on a single patient event chain, the selected treatment nodes include: physiological curve playback and key treatment node display. Physiological curve playback refers to retrieving continuous monitoring sequences such as blood pressure, electrocardiogram, and blood oxygen from the time series database according to a specified time window for a single patient's event chain, and reproducing the changes in the patient's physiological parameters during that period in the form of a time curve for medical staff to review and debrief. Key treatment node display refers to the marking of medication administration, intravenous infusion, and transportation times.
[0028] S2. Based on the ontology model of the combat rescue domain and the pre-built terminology standard system, the field standardization, semantic mapping and unit normalization of multi-source heterogeneous combat rescue data are performed to obtain pre-processed standardized data.
[0029] In one feasible implementation, in addition to field normalization and semantic mapping and unit normalization, time alignment, signal processing, missing detection, anomaly detection and quality scoring are also included.
[0030] Optionally, the ontology in the combat rescue domain includes core classes such as wounded, injuries, events, observations, equipment, locations, and resources; Among them, six semantic relationships are defined between the categories, including: the association between the wounded and the injuries, the association between the wounded and the events, the inclusion relationship between the events and the observations, the generation relationship between the observations and the equipment, the occurrence relationship between the events and the locations, and the consumption relationship between the events and the resources.
[0031] Optionally, the combat rescue domain ontology is used to define four types of data constraints and two types of normalization rules; Among them, the four types of data constraints include: the first type of constraints are used to specify the standard unit of measurement, the set of convertible units, and the unit and dimension constraints of the conversion coefficient for each observation concept; The second type of constraint is used to define the range of legal values for each observation concept. The third type of constraint is used to specify the nominal sampling frequency and the sampling frequency constraint of the allowable fluctuation range for various types of signals; The fourth type of constraint is used to constrain the time consistency of observation timestamps within the same event, the time sequence of events at different stages for the same injured person, and the relationship between observation time and upload time. Among them, four types of data constraints are used to verify the legality of the observation data. Among them, the unit and dimension constraints specify the standard unit of measurement, the set of convertible units, and the corresponding conversion coefficients for each observation concept. For example, blood pressure is uniformly defined as mmHg (millimeters of mercury), body temperature as ℃ (degrees Celsius), or heart rate as beats / min. The legal value range constraints define the physiologically legal value range for each observation concept. For example, heart rate is [0,300] beats / min, blood oxygen saturation is [0,100]%, or body temperature is [20,45]℃. Values outside the range are marked as abnormal and need to be verified. The sampling frequency constraints specify the nominal sampling frequency and allowable fluctuation range for various signals. For example, the sampling frequency of electrocardiogram waveforms is not less than 125Hz, and the sampling interval of routine vital signs is not greater than 60s. This is used to identify downsampling, interrupted sampling, and timestamp anomalies. The time consistency constraints require that the timestamps of each observation within the same event fall within the start and end time window of the event, that the events of each stage of the same patient increase monotonically according to the treatment process, and that the observation time is not later than the data upload time. Data that violates the constraints will trigger time correction or manual verification.
[0032] Among them, the two types of normalization rules include: the first type of normalization rule is a synonym normalization rule used to map the standard name of each concept to common aliases, abbreviations and custom names; Among them, the second type of normalization rule is used to maintain the encoding mapping rule between the ontology concept identifier and various types of codes.
[0033] Two types of normalization rules are used to eliminate semantic differences in heterogeneous data. The synonym normalization rule maintains a mapping between the standard name of each concept and common aliases, English abbreviations, and device manufacturer-defined names using a terminology dictionary. For example, it maps "heart rate," "HR," and "heart rate" to the same concept identifier. The encoding mapping rule maintains a lookup table between ontology concept identifiers and standard injury codes, device output codes, and local codes of various business systems. When multiple codes coexist, the mapping is completed in the following order: standard codes take priority, device codes take the next priority, and local codes take the last priority.
[0034] Optionally, field normalization and semantic mapping employ a three-level mapping strategy: exact matching, synonym matching, and fuzzy matching, including: The original field name and code are matched precisely with the standard name and standard code in the terminology dictionary. When the exact match is not found, that is, when the original field name or code does not have an exact corresponding standard name or standard code in the terminology dictionary, the synonym and abbreviation mapping table pre-maintained in the terminology dictionary is queried to complete the match. If no match is found after matching the synonym and abbreviation mapping table, the string similarity between the original field name and the candidate standard term is calculated by normalized edit distance; the term with the highest similarity and not lower than the preset threshold is taken as the mapping result.
[0035] In one feasible implementation, the process of calculating the string similarity between the original field name and the candidate standard term is represented by the following formula (1): (1) in, The edit distance between string A and string B is the minimum number of single-character insertions, deletions, or replacements required to transform string A into string B. The length of string A; The length of string B; This represents the similarity between string A and string B.
[0036] In one feasible implementation, the term with the highest similarity and not lower than a preset threshold is taken as the mapping result, and the original field and its value are mapped to the corresponding ontology concept identifier ConceptID. Fields with similarity lower than the preset threshold are transferred to manual confirmation, and the confirmation result is written back to the term dictionary for subsequent reuse.
[0037] In one feasible implementation, unit normalization is performed by converting the original observation value x into a value in the standard unit according to the standard unit specified in the ontology, using the following formula (2). : (2) Where a and b are conversion factors from the original unit to the standard unit. For example, when converting body temperature from Fahrenheit to Celsius, use... , When converting blood pressure from kPa to mmHg, take... , Units that only require proportional conversion are taken as follows: .
[0038] In one feasible implementation, the time alignment process includes: First, unify the time of all devices to the same time base: record the time of message transmission when the devices interact with the platform. Platform receiving time Platform response time and equipment receiving time The clock difference between the device and the platform is estimated using the following formula (3). : (3) Among them, according to After shifting and correcting the timestamps of the data uploaded by the device, each signal is resampled to a unified time grid at the nominal sampling frequency. For slowly varying vital signs such as blood pressure and body temperature, the grid point values are obtained by linear interpolation of adjacent sampling points, as expressed by the following formula (4): (4) in, This represents the signal value obtained by linear interpolation at grid time t; Represents the original sampling time adjacent to the raster time t. The sampled value at that location; Represents the original sampling time adjacent to the raster time t. The sampled value at that location; and The original sampling time adjacent to the raster time t; For high-frequency waveforms such as electrocardiograms, cubic spline interpolation or band-limited interpolation is used to complete resampling in order to avoid amplitude distortion and high-frequency component loss introduced by linear interpolation.
[0039] The signal processing includes: performing bandpass filtering on the continuous waveform to suppress baseline drift and high-frequency noise (e.g., using a 0.5-40Hz bandpass for ECG signals); performing median filtering on the beat-by-beat parameters to remove pulse interference; and smoothing using the mean of a sliding window of length N, expressed by the following formula (5): (5) in, This represents the value of the i-th sampling point within the window; This represents the mean value of each sampling point within the sliding window, i.e., the smoothed output value.
[0040] In one feasible implementation, the missing rate is calculated using the following formula (6): (6) in, Indicates the window missing rate; Indicates the actual number of missing points; This indicates the number of sampling points that should be present in the window according to the sampling frequency. Short-term missing data with no more than 3 sampling intervals are filled using linear interpolation, while long-term missing data are simply marked without being filled.
[0041] In one feasible implementation, the anomaly detection combines value range verification and statistical verification: considering that anomalies within the window will simultaneously raise the mean and standard deviation, thereby masking the anomaly itself, the statistical verification adopts robust statistics, using the median and absolute deviation of the median observed within the window instead of the mean and standard deviation. When the value of the sampling point satisfies the following formula (7), or the change of adjacent sampling points exceeds the maximum physiological rate of change allowed by the indicator, the sampling point is determined to be an anomaly.
[0042] (7)
[0043] in, The coefficient 1.4826 makes Under a normal distribution, it is consistent with the standard deviation. Anomaly determination coefficient (preferred) ).
[0044] The quality score is derived by comprehensively considering signal quality, missing data, anomalies, and equipment status, with the signal quality index being a key factor. The signal-to-noise ratio of the window signal is normalized using the following formula (8): (8) in, The ratio of useful signal power to noise power within the window (in dB); estimated separately from the signal power after bandpass filtering and the filtering residual power, and then... The calculation takes values falling within the interval (0,1) and monotonically increases with the signal-to-noise ratio. The quality scoring process is represented by the following formula (9): (9) in, Indicates the proportion of outliers within the window; Indicates quality score; Indicates the first non-negative weight; Indicates the second non-negative weight; This represents the device status score obtained by normalizing factors such as device power consumption and connection stability. Indicates the third non-negative weight; Indicates the fourth non-negative weight; Indicates the window missing rate; The weights are non-negative and satisfy the following conditions: ; , , , Data with Q below a set threshold is marked as low quality and archived for future reference only, without being included in subsequent statistical analysis.
[0045] S3. Perform entity disambiguation, event association, and conflict resolution on the preprocessed standardized data to construct a semantic structure that includes objects, events, and observations.
[0046] Optionally, entity disambiguation is used to unify the identity of wounded persons across devices and stages of data; When the standardized data record to be disambiguated carries a unique identifier, it is directly merged according to the identifier's exact matching method; Otherwise, calculate the overall similarity between candidate record pairs, where the overall similarity is obtained by weighting string similarity, number similarity, feature consistency score, and spatiotemporal proximity determined by the record time difference and location distance; When the overall similarity is not lower than the set merging threshold, they are determined to be the same injured person and are assigned a unified unique identifier. When the overall similarity is between the set pending threshold and the set merging threshold, manual review is performed; when the overall similarity is lower than the set pending threshold, they are determined to be different injured persons.
[0047] Alternatively, the overall similarity can be expressed by the following formula (10): (10) in, Indicates overall similarity; Indicates the first weight; Indicates the similarity of name strings; Indicates the second weight; Indicate the similarity of the ticket or wristband number; Indicates the third weight; It represents the spatiotemporal proximity determined by both the recording time difference and the location distance; Indicates the fourth weight; The score represents the consistency between injury type and injury severity grading characteristics.
[0048] in, The similarity of the name strings is calculated according to formula (1); The similarity of the injury ticket or wristband number is calculated according to formula (1).
[0049] Optionally, the spatiotemporal proximity, determined by both the recording time difference and the location distance, is expressed by the following formula (12): (12) in, It represents the spatiotemporal proximity determined by both the recording time difference and the location distance; d represents the time difference between the two records; d represents the distance between the location points of the two records. Indicates the time decay scale; This indicates the distance attenuation scale.
[0050] Among them, the event association is based on the spatiotemporal window rule to classify observations into events: when the observation timestamp falls within the start and end time window of the event. Furthermore, when the distance between the observation location and the location where the event occurs does not exceed the set threshold, the inclusion relationship between the event and the observation is established, and each event is linked together according to the stage identifiers such as frontline rescue, en route evacuation, and rear rescue, forming an event chain that runs through the entire rescue process.
[0051] Among them, conflict resolution handles the situation where the same concept has multiple source values at the same time or within the same time window: the fusion value is calculated for numerical indicators and expressed by the following formula (13): (13) in, Indicates the fusion value; For the i-th source value; The fusion weight for this source value is expressed by the following formula (14): (14) in, This indicates the decay of freshness over time. This indicates the time difference between the source value and the fusion time. Indicates the decay time constant; Indicates equipment reliability; This indicates the quality score of the source value.
[0052] For enumerated or text-based fields such as injury severity rating, the source value with the highest weight is retained, while the remaining source values along with their weights are archived for traceability and recalculation.
[0053] S4. Write the entities, relationships, and event chains in the semantic structure into the pre-built graph database, and write the continuous monitoring sequence and window features into the pre-built time series database, while generating time series reference identifiers.
[0054] S5. Based on pre-configured search conditions for region, time, and injury severity, filter the sets of injured persons and events that meet the search conditions in the graph database, and perform time window aggregation and statistical calculations in the time series database according to the time series reference identifier, and output the joint query results including the sets of injured persons, the sets of events, and the time window statistics.
[0055] Following the step of outputting the joint query results, the process also includes: The system performs multidimensional statistics, situation display, drill-down analysis, and replay of query results. It also combines resource consumption and inventory status to provide bottleneck warnings and risk event monitoring, enabling a step-by-step backtracking from statistical results to the casualty event chain and then to the physiological curve of a specific time window.
[0056] In this embodiment of the invention, an ontology model and terminology standard system for the field of combat rescue are constructed to perform semantic alignment, entity association, conflict resolution, and quality labeling on casualty-related data. A dual-database storage architecture that works in collaboration with a graph database or knowledge graph database and a time-series database is adopted to achieve efficient fusion and aggregation, unified management, multi-dimensional index retrieval, and visual statistical analysis of multimodal information throughout the combat rescue process. This provides real-time and scalable data support for combat rescue situation assessment, casualty management, and resource scheduling.
[0057] In this embodiment of the invention, the system is validated using a combat rescue drill simulation dataset. The simulation scenario covers the entire process of multi-type detection equipment access, multi-stage response links, and multi-dimensional statistical retrieval, validating ontology-driven semantic regression. For example, Figure 5 The diagram illustrates the implementation process of an ontology-driven multimodal combat rescue data fusion and aggregation method according to an embodiment of the present invention. In one feasible implementation, the specific implementation process of the multimodal combat rescue data fusion and aggregation method includes: Step 1, Data Acquisition and Access: The frontline mobile terminal connects to the multi-parameter monitor, finger-clip pulse oximeter, and electronic injury ticket terminal via Bluetooth or serial port to collect continuous vital sign data such as blood pressure, ECG, blood oxygen, body temperature, and respiration, as well as business data such as injury classification, tourniquet use, medication information, and evacuation destination. Each data point is appended with a device ID, acquisition timestamp, location information, and stage identifiers such as frontline rescue, evacuation en route, and rear rescue at the acquisition terminal. After being cached by the edge gateway, the data is uploaded to the rear platform.
[0058] Step 2, Preprocessing and Semantic Mapping: Perform field standardization and semantic mapping on the uploaded data according to the three-level mapping strategy of exact matching, synonym matching and fuzzy matching. For example, map the field PR output by a certain type of monitor and the "pulse rate" recorded by another business system to the same concept identifier; convert the blood pressure value reported in kPa to mmHg according to formula (2); then perform time alignment, filtering, missing detection and anomaly detection, and calculate the quality score Q for each observation window according to formula (9).
[0059] Step 3, entity disambiguation and event association: For the two records of the same wounded person generated by the front-line registration and the en route retest, the electronic wound ticket number is accurately matched and merged into the same PatientID; the tourniquet treatment and en route infusion and other business actions are organized into events and EventID is generated; the vital signs observations in the corresponding time window are classified into the corresponding events according to the time consistency constraint, forming the wounded person-event-observation link; for the blood oxygen values reported by the monitor and pulse oximeter at the same time window, the fusion value is calculated according to formula (13) and the original source value is retained for future reference.
[0060] Step 4, Dual Database Writing and Association: Write the injured, injuries, events, observation nodes and their relationships into a pre-built graph database or knowledge graph database, write the continuous monitoring sequences such as ECG or blood pressure and window features into the time series database, generate TSRefs to be backfilled to the corresponding observation nodes, and record them in the log to support traceability, recalculation and playback.
[0061] Step 5: Retrieval Statistics and Backtracking Verification: Set retrieval conditions such as region, time, and injury severity level, filter the sets of injured persons and events that meet the conditions in the pre-built graph database or knowledge graph library, perform time window aggregation in the time series database based on TSRef, and return the joint query results; The statistical interface displays the distribution of the number of injured persons for each type of injury, each treatment method, and each destination. It also supports drilling down from the statistical charts to the event chain of an individual injured person, and then replaying the physiological curve within a specific time window, realizing a step-by-step backtracking from statistical results to event chain and then to physiological curve.
[0062] The embodiments of this invention achieve semantic consistency across devices, stages, and systems through ontology-driven approaches, significantly reducing the cost of manual field maintenance. By constructing object event observation links through entity disambiguation and event association, the ability to connect information across processes is improved, solving the problems of poor scalability and inability to handle semantic layer entity associations in dynamic combat and rescue scenarios of existing manual field alignment schemes.
[0063] This invention employs a dual-database storage approach, combining a knowledge graph library and a time-series library, and enables linked retrieval via TSRef. This achieves complementary advantages between semantic relationship retrieval and high-frequency time-series aggregation. Furthermore, it supports combat readiness assessment and resource scheduling through multi-dimensional indexing and visual statistical analysis, thus addressing the shortcomings of existing knowledge graph solutions in terms of multi-modal time-series data collaborative management capabilities and multi-dimensional retrieval capabilities for combat readiness operations.
[0064] Compared with existing technologies, the embodiments of this invention take the ontology model of the combat rescue domain as the core, realize the unified semantic expression across devices, stages and systems, significantly reduce the cost of manual field mapping and rule maintenance, and improve the scalability and reusability of the system in dynamic combat rescue scenarios; by establishing a unified object-event-observation link through entity disambiguation and event association, the ability to connect data across multiple stages such as front-line handling, en route transfer and rear treatment is enhanced; by storing data in a dual database of graph database / knowledge graph database and time series database, and by establishing a linkage mechanism with the help of TSRef, the semantic relationship retrieval capability and the aggregation, playback and analysis capability of high-frequency time series data are taken into account.
[0065] Figure 6 This is a block diagram of an ontology-driven multimodal combat rescue data fusion and aggregation system provided in an embodiment of the present invention. This system is used for an ontology-driven multimodal combat rescue data fusion and aggregation method. (Refer to...) Figure 6The system includes a data acquisition and access module 610, an ontology model and terminology standard module 620, a multimodal data preprocessing module 630, a unified storage module 640, and a multidimensional indexing and retrieval module 650. Among them: The data acquisition and access module 610 is used to acquire multi-source heterogeneous combat rescue data and add metadata including device identifier, timestamp, geographic location, and stage identifier; wherein the multi-source heterogeneous combat rescue data is multi-source heterogeneous data from various combat rescue terminals, monitoring equipment, and business systems. The ontology model and terminology standard module 620 is used to perform field standardization, semantic mapping, and unit normalization on multi-source heterogeneous combat rescue data based on the combat rescue ontology model and the pre-built terminology standard system, so as to obtain pre-processed standardized data. The multimodal data preprocessing module 630 is used to perform entity disambiguation, event association and conflict resolution on the preprocessed standardized data, and to construct a semantic structure containing objects, events and observations. The unified storage module 640 is used to write entities, relationships and event chains in the semantic structure into a pre-built graph database, and to write continuous monitoring sequences and window features into a pre-built time series database, while generating time series reference identifiers. The multidimensional indexing and retrieval module 650 is used to filter sets of wounded persons and sets of events that meet the pre-configured search conditions of region, time and injury severity in the graph database, and perform time window aggregation and statistical calculations in the time series database according to the time series reference identifier, and output the joint query results including the set of wounded persons, the set of events and the time window statistics.
[0066] Optionally, the field standardization and semantic mapping employ a three-level mapping strategy of exact matching, synonym matching, and fuzzy matching, including: The original field name and code are matched precisely with the standard name and standard code in the terminology dictionary. When the exact match is not found, that is, when the original field name or code does not have an exact corresponding standard name or standard code in the terminology dictionary, the synonym and abbreviation mapping table pre-maintained in the terminology dictionary is queried to complete the match. When no match is found after matching the synonym and abbreviation mapping table, the string similarity between the original field name and the candidate standard term is calculated by normalized edit distance; the term with the highest similarity and not lower than the preset threshold is taken as the mapping result.
[0067] Optionally, the combat rescue domain ontology includes core classes such as wounded, injuries, events, observations, equipment, locations, and resources; Among them, six semantic relationships are defined between the categories, including: the association between the wounded and the injuries, the association between the wounded and the events, the inclusion relationship between the events and the observations, the generation relationship between the observations and the equipment, the occurrence relationship between the events and the locations, and the consumption relationship between the events and the resources.
[0068] Optionally, the combat rescue domain ontology is used to define four types of data constraints and two types of normalization rules; Among them, the four types of data constraints include: the first type of constraints are used to specify the standard unit of measurement, the set of convertible units, and the unit and dimension constraints of the conversion coefficient for each observation concept; The second type of constraint is used to define the range of legal values for each observation concept. The third type of constraint is used to specify the nominal sampling frequency and the sampling frequency constraint of the allowable fluctuation range for various types of signals; The fourth type of constraint is used to constrain the time consistency of observation timestamps within the same event, the time sequence of events at different stages for the same injured person, and the relationship between observation time and upload time. Among them, the two types of normalization rules include: the first type of normalization rule is a synonym normalization rule used to map the standard name of each concept to common aliases, abbreviations and custom names; Among them, the second type of normalization rule is used to maintain the encoding mapping rule between the ontology concept identifier and various types of codes.
[0069] Optionally, the entity disambiguation is used to unify the identity of the wounded in data across devices and stages; When the standardized data record to be disambiguated carries a unique identifier, it is directly merged according to the identifier's exact matching method; Otherwise, calculate the overall similarity between candidate record pairs, wherein the overall similarity is obtained by weighting string similarity, number similarity, feature consistency score and spatiotemporal proximity determined by record time difference and location distance; When the overall similarity is not lower than the set merging threshold, the injured person is determined to be the same injured person and assigned a unified unique identifier. When the overall similarity is between the set pending threshold and the set merging threshold, manual review is performed; when the overall similarity is lower than the set pending threshold, they are determined to be different injured persons.
[0070] Optionally, the overall similarity is represented by the following formula (15): (15) Where Sim represents the overall similarity; Indicates the first weight; Indicates the similarity of name strings; Indicates the second weight; Indicate the similarity of the ticket or wristband number; Indicates the third weight; It represents the spatiotemporal proximity determined by both the recording time difference and the location distance; Indicates the fourth weight; The score represents the consistency between injury type and injury severity grading characteristics.
[0071] Optionally, the spatiotemporal proximity determined by the recording time difference and location distance is expressed by the following formula (16): (16) in, It represents the spatiotemporal proximity determined by both the recording time difference and the location distance; d represents the time difference between the two records; d represents the distance between the location points of the two records. Indicates the time decay scale; This indicates the distance attenuation scale.
[0072] The embodiments of this invention achieve semantic consistency across devices, stages, and systems through ontology-driven approaches, significantly reducing the cost of manual field maintenance. By constructing object event observation links through entity disambiguation and event association, the ability to connect information across processes is improved, solving the problems of poor scalability and inability to handle semantic layer entity associations in dynamic combat and rescue scenarios of existing manual field alignment schemes.
[0073] This invention employs a dual-database storage approach, combining a knowledge graph library and a time-series library, and enables linked retrieval via TSRef. This achieves complementary advantages between semantic relationship retrieval and high-frequency time-series aggregation. Furthermore, it supports combat readiness assessment and resource scheduling through multi-dimensional indexing and visual statistical analysis, thus addressing the shortcomings of existing knowledge graph solutions in terms of multi-modal time-series data collaborative management capabilities and multi-dimensional retrieval capabilities for combat readiness operations.
[0074] Figure 7 This is a schematic diagram of the structure of a body-driven multimodal combat rescue data fusion and aggregation device provided in an embodiment of the present invention, as shown below. Figure 7 As shown, the body-driven multimodal combat rescue data fusion and aggregation device may include the above-mentioned Figure 6 The illustrated system is a body-driven multimodal combat rescue data fusion and aggregation system. Optionally, the body-driven multimodal combat rescue data fusion and aggregation device 710 may include a first processor 2001.
[0075] Optionally, the body-driven multimodal combat rescue data fusion and aggregation device 710 may also include a memory 2002 and a transceiver 2003.
[0076] The first processor 2001, memory 2002, and transceiver 2003 can be connected via a communication bus.
[0077] The following is combined Figure 7 A detailed introduction to each component of the 710 body-driven multimodal combat rescue data fusion and aggregation device is provided: The first processor 2001 is the control center of the multimodal combat rescue data fusion and aggregation device 710 driven by the body. It can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 can be one or more central processing units (CPUs), application-specific integrated circuits (ASICs), or one or more integrated circuits configured to implement the embodiments of the present invention, such as one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs).
[0078] Optionally, the first processor 2001 can execute various functions of the body-driven multimodal combat rescue data fusion and aggregation device 710 by running or executing software programs stored in the memory 2002 and calling data stored in the memory 2002.
[0079] In a specific implementation, as one example, the first processor 2001 may include one or more CPUs, for example... Figure 7 CPU0 and CPU1 are shown in the diagram.
[0080] In a specific implementation, as one example, the ontology-driven multimodal combat rescue data fusion and aggregation device 710 may also include multiple processors, for example... Figure 7 The first processor 2001 and the second processor 2004 are shown in the diagram. Each of these processors can be a single-core processor or a multi-core processor. Here, a processor can refer to one or more devices, circuits, and / or processing cores used to process data (such as computer program instructions).
[0081] The memory 2002 is used to store the software program that executes the present invention, and is controlled by the first processor 2001 to execute it. The specific implementation method can be referred to the above method embodiment, and will not be repeated here.
[0082] Optionally, the memory 2002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 2002 may be integrated with the first processor 2001 or may exist independently, and may be connected via the interface circuit of the multimodal combat data fusion and aggregation device 710 driven by the processor itself. Figure 7 (Not shown in the image) is coupled to the first processor 2001, and this embodiment of the invention does not specifically limit this.
[0083] The transceiver 2003 is used to communicate with network devices or with terminal devices.
[0084] Alternatively, transceiver 2003 may include a receiver and a transmitter. Figure 7 (Not shown separately). The receiver is used to implement the receiving function, and the transmitter is used to implement the transmitting function.
[0085] Optionally, the transceiver 2003 can be integrated with the first processor 2001, or it can exist independently, and its interface circuit can be driven by the multimodal combat rescue data fusion and aggregation device 710. Figure 7 (Not shown in the image) is coupled to the first processor 2001, and this embodiment of the invention does not specifically limit this.
[0086] It should be noted that, Figure 7 The structure of the body-driven multimodal combat rescue data fusion and aggregation device 710 shown in the figure does not constitute a limitation on the router. Actual body-driven multimodal combat rescue data fusion and aggregation devices may include more or fewer components than shown, or combine certain components, or have different component arrangements.
[0087] Furthermore, the technical effects of the body-driven multimodal combat rescue data fusion and aggregation device 710 can be referred to the technical effects of the body-driven multimodal combat rescue data fusion and aggregation method described in the above method embodiments, and will not be repeated here.
[0088] It should be understood that the first processor 2001 in the embodiments of the present invention may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor, or it may be any conventional processor.
[0089] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).
[0090] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable system. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.
[0091] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.
[0092] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.
[0093] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.
[0094] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.
[0095] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, systems, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0096] In the embodiments provided by this invention, it should be understood that the disclosed devices, systems, and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or units may be electrical, mechanical, or other forms.
[0097] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0098] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
[0099] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0100] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A ontology-driven method for fusing and converging multimodal combat rescue data, characterized in that, The method includes: S1. Acquire multi-source heterogeneous combat rescue data and attach metadata including device identifier, timestamp, geographic location, and stage identifier; wherein the multi-source heterogeneous combat rescue data refers to multi-source heterogeneous data from various combat rescue terminals, monitoring equipment, and business systems. S2. Based on the ontology model of the combat rescue domain and the pre-built terminology standard system, the field standardization, semantic mapping and unit normalization of multi-source heterogeneous combat rescue data are performed to obtain pre-processed standardized data. S3. Perform entity disambiguation, event association, and conflict resolution on the preprocessed standardized data to construct a semantic structure that includes objects, events, and observations; S4. Write the entities, relationships, and event chains in the semantic structure into a pre-built graph database, and write the continuous monitoring sequence and window features into a pre-built time series database, while generating a time series reference identifier; S5. Based on pre-configured search conditions for region, time, and injury severity, filter the sets of injured persons and events that meet the search conditions in the graph database, and perform time window aggregation and statistical calculations in the time series database according to the time series reference identifier, and output the joint query results including the sets of injured persons, the sets of events, and the time window statistics.
2. The ontology-driven multimodal combat rescue data fusion and aggregation method according to claim 1, characterized in that, The field standardization and semantic mapping employ a three-level mapping strategy: exact matching, synonym matching, and fuzzy matching, including: The original field name and code are matched precisely with the standard name and standard code in the terminology dictionary. When the exact match is not found, that is, when the original field name or code does not have an exact corresponding standard name or standard code in the terminology dictionary, the synonym and abbreviation mapping table pre-maintained in the terminology dictionary is queried to complete the match. If no match is found after matching the synonym and abbreviation mapping table, the string similarity between the original field name and the candidate standard term is calculated by normalized edit distance; the term with the highest similarity and not lower than the preset threshold is taken as the mapping result.
3. The ontology-driven multimodal combat rescue data fusion and aggregation method according to claim 1, characterized in that, The combat rescue domain ontology includes core classes for wounded, injuries, events, observations, equipment, locations, and resources; Among them, six semantic relationships are defined between the categories, including: the association between the wounded and the injuries, the association between the wounded and the events, the inclusion relationship between the events and the observations, the generation relationship between the observations and the equipment, the occurrence relationship between the events and the locations, and the consumption relationship between the events and the resources.
4. The ontology-driven multimodal combat rescue data fusion and aggregation method according to claim 1, characterized in that, The combat rescue domain ontology is used to define four types of data constraints and two types of normalization rules; Among them, the four types of data constraints include: the first type of constraints are used to specify the standard unit of measurement, the set of convertible units, and the unit and dimension constraints of the conversion coefficient for each observation concept; The second type of constraint is used to define the range of legal values for each observation concept. The third type of constraint is used to specify the nominal sampling frequency and the sampling frequency constraint of the allowable fluctuation range for various types of signals; The fourth type of constraint is used to constrain the time consistency of observation timestamps within the same event, the time sequence of events at different stages for the same injured person, and the relationship between observation time and upload time. Among them, the two types of normalization rules include: the first type of normalization rule is a synonym normalization rule used to map the standard name of each concept to common aliases, abbreviations and custom names; Among them, the second type of normalization rule is used to maintain the encoding mapping rule between the ontology concept identifier and various types of codes.
5. The ontology-driven multimodal combat rescue data fusion and aggregation method according to claim 1, characterized in that, The entity disambiguation is used to unify the identity of the injured in data across devices and stages; When the standardized data record to be disambiguated carries a unique identifier, it is directly merged according to the identifier's exact matching method; Otherwise, calculate the overall similarity between candidate record pairs, wherein the overall similarity is obtained by weighting string similarity, number similarity, feature consistency score and spatiotemporal proximity determined by record time difference and location distance; When the overall similarity is not lower than the set merging threshold, the injured person is determined to be the same injured person and assigned a unified unique identifier. When the overall similarity is between the set pending threshold and the set merging threshold, manual review is performed; when the overall similarity is lower than the set pending threshold, they are determined to be different injured persons.
6. The ontology-driven multimodal combat rescue data fusion and aggregation method according to claim 5, characterized in that, The overall similarity is expressed by the following formula (1): (1) in, Indicates overall similarity; Indicates the first weight; Indicates the similarity of name strings; Indicates the second weight; Indicate the similarity of the ticket or wristband number; Indicates the third weight; It represents the spatiotemporal proximity determined by both the recording time difference and the location distance; Indicates the fourth weight; The score represents the consistency between injury type and injury severity grading characteristics.
7. The ontology-driven multimodal combat rescue data fusion and aggregation method according to claim 6, characterized in that, The spatiotemporal proximity, determined by the recording time difference and location distance, is expressed by the following formula (2): (2) in, It represents the spatiotemporal proximity determined by both the recording time difference and the location distance; d represents the time difference between the two records; d represents the distance between the location points of the two records. Indicates the time decay scale; This indicates the distance attenuation scale.
8. An ontology-driven multimodal combat rescue data fusion and aggregation system, wherein the ontology-driven multimodal combat rescue data fusion and aggregation system is used to implement the ontology-driven multimodal combat rescue data fusion and aggregation method as described in any one of claims 1-7, characterized in that, The system includes: The data acquisition and access module is used to acquire multi-source heterogeneous combat rescue data and add metadata including device identifier, timestamp, geographic location, and stage identifier; wherein the multi-source heterogeneous combat rescue data is multi-source heterogeneous data from various combat rescue terminals, monitoring equipment, and business systems; The ontology model and terminology standard module is used to perform field standardization, semantic mapping, and unit normalization on multi-source heterogeneous combat rescue data based on the combat rescue ontology model and the pre-built terminology standard system, so as to obtain pre-processed standardized data. The multimodal data preprocessing module is used to perform entity disambiguation, event association, and conflict resolution on the preprocessed standardized data, and to construct a semantic structure that includes objects, events, and observations. A unified storage module is used to write entities, relationships, and event chains in the semantic structure into a pre-built graph database, and to write continuous monitoring sequences and window features into a pre-built time-series database, while generating time-series reference identifiers. The multidimensional indexing and retrieval module is used to filter sets of wounded persons and events that meet the pre-configured search conditions of region, time and injury severity in the graph database, and perform time window aggregation and statistical calculations in the time series database according to the time series reference identifier, and output the joint query results including the set of wounded persons, the set of events and the time window statistics.
9. A body-driven multimodal combat rescue data fusion and aggregation device, characterized in that, The body-driven multimodal combat rescue data fusion and aggregation device includes: processor; A memory storing computer-readable instructions that, when executed by the processor, implement the method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium contains program code that can be invoked by a processor to execute the method as described in any one of claims 1 to 7.
